91832 OPEN DATA FOR RESILIENCE INITIATIVE: PLANNING AN OPEN CITIES MAPPING PROJECT OPEN DATA FOR RESILIENCE INITIATIVE: PLANNING AN OPEN CITIES MAPPING PROJECT The World Bank D Humanitarian OpenStreetMap Team © 2014 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contribu- tions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or accep- tance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: pubrights@worldbank.org. TABLE OF CONTENTS 1 INTRODUCTION 11 1.1 History of Open Cities and Open Data for Resilience Initiative 11 1.2 Rationale for the Open Cities Project 12 1.3 How to Use this Guide 15 2 PROJECT DESIGN AND PREPARATION 19 2.1 Key Partnerships 19 2.1.1 Why Would Groups Join an Open Cities Project? 19 2.1.2 Outreach 22 2.2 Defining the Scope of Work 23 2.2.1 Overview of the Process 23 2.2.2 Why Are You Collecting the Data? 25 2.2.3 Defining the Target Area 25 2.2.4 Building the Data Model 28 2.2.5 Estimating a Time Frame 28 2.3 Building the Team (Staffing) 29 2.3.1 Management Structure, Roles, and Responsibilities 29 2.3.2 Finding Mappers and Surveyors 31 2.3.3 Compensation and Incentives 32 2.4 Assessing Existing Map Resources 33 2.4.1 Existing Digital and Paper Data 33 2.4.2 Availability of Imagery for Digitization 34 2.4.3 GPS Devices, Field Papers, and Smartphones 36 3 GETTING STARTED 43 3.1 Finding a Workspace 43 3.2 Equipment and Other Costs 44 3.2.1 Hardware Needs 44 3.2.2 Printing Needs 44 3.2.3 Mapping Workshop Expenses 44 3.2.4 Surveying Expenses 44 3.3 Training 45 3.3.1 Training on Mapping Tools 45 3.3.2 Training on Surveying 46 4 IMPLEMENTATION AND SUPERVISION 49 4.1 Collecting the Data 49 4.1.1 Creating Reference Maps 49 4.1.2 Defining Daily Mapping Areas 50 4.1.3 Collecting Field Data 51 4.1.4 Developing Survey Forms 51 4.1.5 Entering the Data (Editing) 52 4.1.6 Scheduling Project Tasks 55 4.2 Common Challenges 56 4.2.1 Bottlenecks 56 4.2.2 Time Management 56 4.2.3 Survey Fatigue 56 4.3 Quality Control 56 4.3.1 Daily Data Checks 56 4.3.2 Resurveying 57 4.3.3 Data Analysis 57 4.4 Reporting 57 5 LESSONS LEARNED AND RECOMMENDATIONS 61 5.1 Government Ownership 61 5.2 Partnerships with Universities 61 5.3 Access to Imagery 62 5.4 Community Connections 62 5.5 Building Trust in the Data 62 5.6 Sustained Engagement 62 CASE STUDIES 65 Batticaloa, Sri Lanka 67 Kathmandu, Nepal 71 Dhaka, Bangladesh 75 6 APPENDIXES 79 6.1 Data Model Design 79 6.2 Sample Survey Form 83 6.3 Sample Data Authorization Letter 87 6.4 Sample Project Report 89 6.5 Sample Letter of Support for Mappers 93 6.6 Open Cities Project Planning Checklist 95 ACKNOWLEDGMENTS This publication was prepared by a team consisting of Marc Forni, Jeff Haack, Robert Soden, Dustin York, Ryan Sommerville, and Vivien Deparday. The publication benefited from inputs and contributions from the following: Alanna Simpson [GFDRR], Nama Budhathoki [Kathmandu Living Labs], Abby Baca, Eric Dickson, Ignacio Urrutia [World Bank]. The Open Cities project is the result of the hard work, dedication, and creativity of numerous local participants in government, civil society, private sector in Bangladesh, Nepal, and Sri Lanka. The Open Cities project team is grateful to contributions from Anil Pokhrel, Suranga Kahandawa, Swarna Kazi, Mirva Tulia, Marc Ellery, Steven Rubinyi and Sonam Velani [World Bank], Kate Chapman [Humanitarian OpenStreetMap Team], Josh Campbell, Benson Wilder, and Patrick Dufour [US State Department Humanitarian Information Unit], Shadrock Roberts and Chad Blevins [USAID], Robert Banick and Maya Kapsoka- vadis [American Red Cross], and Alyssa Wright [Mapzen]. THE OPEN CITIES PARTNERSHIP The Open Cities project is a partnership between GFDRR, the World Bank, the Humanitarian OpenStreetMap Team, American Red Cross, US State Department Humanitarian Information Unit, USAID, and Development Seed. ABBREVIATIONS BUET Bangladesh University of Engineering and Technology DRM Disaster Risk Management (World Bank unit) GIS Geographic Information System GPS Global Positioning System GFDRR Global Facility for Disaster Reduction and Recovery (World Bank) HOT Humanitarian OpenStreetMap Team JOSM Java OpenStreetMap Editor NGO Nongovernmental Organization ODbL Open Database License OpenDRI Open Data for Resilience Initiative (World Bank/GFDRR project) OSM OpenStreetMap USAID United States Agency for International Development planning an open cities mapping project 10 1 INTRODUCTION 1.1 HISTORY OF OPEN CITIES AND THE OPEN DATA FOR RESILIENCE INITIATIVE The World Bank, through its Global Facility for Disaster Reduction and Recovery (GFDRR), launched the Open Cities Project in November 2012 to create open data ecosystems that will facilitate innovative, data-driven urban planning and disaster risk management in South Asian cities. Open Cities is one component of a broader World Bank and GFDRR program, the Open Data for Resilience Initiative (OpenDRI), further described in box 1.1. Box 1.1 About OpenDRI Open Cities is part of the Open Data for Resilience Initiative, sponsored by the World Bank and GFDRR. OpenDRI brings the philosophies and practices of the open data movement to bear on the challenges of building resilience to natural hazards and the impacts of climate change. In partner- ship with governments, international organizations, and civil society groups, this initiative develops open systems for creating, sharing, and using disaster risk and climate change information to ensure that a wide range of actors can participate in meeting these challenges. Since its launch in 2011, OpenDRI has worked to implement these ideas in over 25 countries around the world. Since its inception, Open Cities has brought together Cities Project, the major components of its implementa- stakeholders from government, donor agencies, the tion to date, and some of the most salient lessons learned private sector, universities, and civil society groups to from the project so far. create usable information through community mapping techniques, to build applications and tools that inform The Open Cities Project launched its efforts in three decision making, and to develop the networks of trust cities: Batticaloa, Sri Lanka; Dhaka, Bangladesh; and and social capital necessary for these efforts to become Kathmandu, Nepal. These cities were chosen for sustainable. This process has been evolutionary, with DD Their high levels of disaster risk; opportunities for experimentation, learning, failure, and adaptation incorporated into the project planning. This guide discusses the rationale and design of the Open 11 planning an open cities mapping project Box 1.2 The Open Cities Website The Open Cities website provides case studies and visualizations of, and access to, much of the data collected during the first round of Open Cities projects. It was designed and built by our partners at Development Seed and can be accessed at http://opencitiesproject.org. It will be updated continually as the program expands and more data become available. DD The presence of World Bank-lending activities related manage urban sprawl; and to identify potential sites for to urban planning and disaster management that parks and public services. In addition, growing popu- would benefit from access to better data; and lations, unplanned settlements, and unsafe building practices all increase disaster risk. DD The willingness of government counterparts to partic- ipate in and help guide the interventions. As urban populations and vulnerability grow, manag- ing urban growth in a way that fosters cities’ resilience In each of these projects, Open Cities has supported the to natural hazards and the impacts of climate change creation of new data while also attending to the cities’ becomes an ever-greater challenge that requires detailed, broader ecosystems of open data production and use. To up-to-date geographic data of the built environment. view the city-specific project data collected thus far, visit To meet this challenge requires innovative, affordable, the Open Cities website (box 1.2). precise, open, and dynamic data collection and mapping processes that support management of urban growth Leveraging robust, accurate data to improve urban and disaster risk. In response to the 2010 earthquake in planning and disaster risk management decisions Haiti, OpenStreetMap pioneered such efforts, as box 1.3 requires not only high-quality information but also the further describes. requisite tools, skills, and willingness to commit to a Open Cities approaches risk assessment differently data-driven decision-making process. With this in mind, from catastrophic-risk-modeling firms, whose data are Open Cities also has developed partnerships across typically used either for broad awareness raising or government ministries, donor agencies, universities, for the insurance industry. These professional assess- private sector technology groups, and civil society ments often involve computationally intensive mod- organizations to ensure broad acceptance of the data eling analysis, but they also tend to rely on statistical produced, facilitate data use, and align investments representations, proxies, or estimations of the exposed across projects and sectors. assets, which are expressed in monetary terms. Such data are insufficient for driving specific investments to 1.2 RATIONALE FOR THE OPEN reduce disaster risk because they typically do not locate, describe, and value individual assets. CITIES PROJECT By contrast, Open Cities engages local expertise and South Asia is one of the most rapidly urbanizing regions stakeholders in identifying all building structures in in the world. A deep understanding of the built environ- an area and assigning vulnerability attributes to each ment is critical to providing relevant services, managing through community mapping and crowdsourcing. In urban growth, and visualizing disaster risk in this this way, a risk assessment that identifies particular context. For example, good characterization of the built at-risk structures can be completed. An assessment this environment allows urban planners, engineers, and precise can identify structures based on importance and policy makers to plan for and design appropriate trans- risk level and therefore can guide plans to reduce disas- portation systems and adequate water supply systems; to ter and climate risk through physical investment. estimate the population distribution of cities; attempt to 12 chapter 2 | project design and preparation Box 1.3 OpenStreetMap Open Cities drew from, and was inspired by, a number of projects involving community mapping, primarily the OpenStreetMap (OSM) response to the 2010 Haiti earthquake and the “Community Mapping for Exposure” effort by Australian and Indonesian governments with the Humanitarian OpenStreetMap Team (HOT). Open Cities and the other projects used the OSM platform to harness the power of crowd and community to create accurate, up-to-date spatial data about locations and characteristics of the built and natural environments. Dubbed “the Wikipedia of maps” by its founder, British technology entrepreneur Steve Coast, OSM is an online database and a global com- munity of over 1 million contributors who collaborate toward building a free and open map of the world to which anyone can contribute and which anyone can use in their own tools and analysis. OSM was first used in disaster response in a large-scale fashion in Haiti, following the January 12, 2010, earthquake. In the days following the disaster, the World Bank, Google, and several other entities made high-resolution imagery of the affected area available to the public. Over 600 indi- viduals from the global OSM community began digitizing the imagery and tracing roads, building footprints, and other infrastructure, creating what quickly became the most detailed map of Port au Prince in existence. OSM’s map became the de facto map for the area, and it was used on the global positioning system (GPS) devices of search and rescue teams, to help route supplies around the congested and devastated capital, and to coordinate many other aspects of the response and reconstruction effort. Subsequent work supported by the World Bank, the International Organi- zation for Migration (IOM), USAID, and others would go on to help develop an OSM community in Haiti, ensuring local contributors would continue to shepherd and maintain the resource that international volunteers had created. The experience of OSM Haiti demonstrated that volunteers who collaborated around open data could quickly create accurate and trusted information. Meanwhile, an OSM project in Indonesia called “Community Mapping for Exposure” has sought to create data—locally and in advance of a disaster—that could inform disaster preparedness and contingency planning activities. Working with local governments, university students, and civil society groups, the mapping work has focused especially on critical infrastructure in the capital region of Jakarta, including schools, hospitals, community centers, and places of worship. The resulting data have been combined with hazard information from a variety of sources to produce realistic impact scenarios using the InaSAFE tool. InaSAFE is an open-source project developed by the Australian Government, the Government of Indonesia, and the World Bank that was created specifically for this work but is now being deployed in other DRM projects. The OSM project has since expanded beyond Jakarta and has mapped over 1 million buildings across Indonesia since March 2011. 13 planning an open cities mapping project Moreover, Open Cities uses detailed information about and human capital, jobs, support to innovation and new roads, building infrastructures, and population to help business while ensuring that the overall efforts become develop land use plans, contingency plans, evacuation sustainable. An integral part of Open Cities as a com- routes, and plans for cash transfers to affected vulner- ponent of OpenDRI is the development of an ecosystem able households in an informed and effective manner. of data producers and data users through partnerships, Open Cities thus provides the platform to support capacity building, innovation, and software develop- activities through the entire disaster risk management ment—aspects to be further described throughout the cycle. The concept has proven effective and, in fact, can guide. be less expensive to complete than the exposure and vulnerability analyses by typical risk modelers, thanks to Open Cities has achieved several noteworthy outcomes the engagement of local experts and stakeholders. It also during its first year: helps to ensure ownership of, and trust in, the data by DD Comprehensive and accessible databases of the built the local stakeholders and community. environment. For instance, Batticaloa now has a As the data collected become openly accessible, they can detailed structural database of every building, and be reused, complemented, and enhanced for other appli- Kathmandu has a database of all schools and hospitals cations beyond the initial project, especially in sectors to use for risk assessment. such as energy and transportation where detailed, up-to- DD Improved in-country capacity to update, maintain, date geographic data are required to develop investment and use key datasets. For instance, Kathmandu has programs. As such, having dynamic and detailed data created innovation spaces such as the Kathmandu that are freely available to and enhanced by all sectors of Living Labs, internship opportunities, and university the government, development institutions, and private curricula that provide students with employable skills. business creates economic value far beyond the project DD Mainstreamed open-data use and strengthened data that initiates the data collection. (Box 1.4 describes the collection and management processes at different open license for availability and use of the data in con- levels of government. For instance, the Sri Lanka junction with the OpenStreetMap platform.) Survey Department asked for support to start incor- In addition to the value of the data’s sheer availability, porating crowdsourcing and community mapping the Open Cities Project’s approach of building the approaches into its regular workflow, and the Govern- capacity of local communities, local and national gov- ment of Sri Lanka sought support for the creation of ernment, academics, and the private sector creates social an Open and Spatial Data Infrastructure. Box 1.4 Open Data Licensing The information collected and added to the OpenStreetMap platform is distributed with the Open Database License (ODbL). This means that although individual contributors hold the copyright to the data they produce, the collective data of all contributors are available under this open license. The ODbL allows anyone to freely copy, distribute, and adapt OSM data. The only requirement is that OSM be given credit in any adapted works, and if the original data are altered, the result should be made available under the same license. 14 chapter 2 | project design and preparation DD Adoption of new applications by multiple levels of government and Bank-financed projects. For instance, forthcoming risk assessments will be driven by detailed data to design physical mitigation investment programs. DD Complementary new partnerships and increased collaboration. New partners to implement projects include the U.S. Department of State, the United States Agency for International Development (USAID), the Humanitarian OpenStreetMap Team (HOT), and the American Red Cross. 1.3 HOW TO USE THIS GUIDE This guide offers a comprehensive understanding of the design and implementation of an Open Cities mapping project—for both practitioners in the field and those interested in a higher-level understanding of the process. The guide’s content is based on experience in imple- menting the initial Open Cities projects in Bangladesh, Nepal, and Sri Lanka as well as on previous mapping project experience. Where relevant, it provides relevant examples from those projects in the text and full case studies at the end of guide. Figure 1.1 depicts the overall process that the chapters will discuss in detail. The next chapter, “Project Design and Preparation,” covers how a project design process begins: by identifying partners, clarifying a project’s objectives and scope, assembling a team of managers and mappers, and assessing the necessary resources for mapping. Chapter 3, “Getting Started,” then describes the steps after the initial planning stage: how to locate an appropriate workspace, assess equipment costs, and prepare staff training. Chapter 4, “Implementation and Supervision,” takes a practical look at data collection techniques from both the organizational and technical perspectives. It also addresses common challenges and mechanisms for quality control and reporting. Finally, chapter 5 examines the lessons learned from previous Open Cities projects and considers future improvements to the overall project design. 15 Open Cities Process Diagram PLANNING New Challenges Assessing Key Identified Actor’s Goals Data Usage D Structural data as input risk assessment that informs retrofitting investments D Location and characteristics of critical facilities for emergency preparedness planning D Data visualizations to communicate the city’s at-risk sections to decision-makers and the public. Quality Assessment EXECUTION Building the Ecosystem Each part of an Open Cities project offers opportunities for involving new participants, demonstrating the value of open data, and supporting the growth of the networks organizations and individuals who can continue to Figure 1.1 Designing and executing Open Cities projects is a complex task that involves a great deal of coordination with partners, technical and scientific work, team and volunteer coordination and management, and logistical work. While the format of this book necessarily presents these steps as linear, in practice this tasks are ongoing, iterative, and happening in parallel. Defining Assessing Scope Existing Resources Assess the geographic areas or features not covered, as this provides the opportunity to collect additional data Setting Up Logistics Quality Assessment is performed so that problems can be caught early, and mappers can be re-trained as necessary. Training Data Collection PREPARATION update the data or champion the work after the project itself is complete. Finding ways to build the ecosystem of data contributors and users involved in an Open Cities project is key to long-term sustainability and impact. 18 2 PROJECT DESIGN AND PREPARATION CHAPTER HIGHLIGHTS: DD Identifying Key Groups for Outreach and Partnerships DD Defining Project Scope and Objectives DD Building the Project Team DD Assessing and Preparing Existing Mapping Resources 2.1 KEY PARTNERSHIPS DD Will benefit from the innovation to improve its workflow; Involvement of a wide community of partners is a central component of an Open Cities mapping project. DD Provides contextual knowledge; and The OSM community is already built around the idea DD Shares resources such as workspace, relationships, of collaboration, and its mapping software serves this staff, and supplies. purpose. Open Cities project data are sourced from and collected by many users and partners and are open and freely available to anyone who wants to use the data. 2.1.1 Why Would Groups Join an Open Thus, engaging a community of users and contributors Cities Project? is an important first objective of planning a mapping All organizations involved in disaster risk management project. and urban planning (such as local and national govern- ments, international partners, and nongovernmental Consider that a good partner organizations [NGOs]) need up-to-date, detailed data for DD Is already undertaking work in a similar context; their operations. Some of them may already be strug- gling to maintain their own datasets. By joining with DD Needs the data being collected in its operation; many partners, organizations find it easier to keep data DD Has a passion for change and an openness to relevant and up-to-date and to avoid reinventing the innovation; wheel. The data become more useful and more widely used, their application leading to greater innovation, 19 Open Cities Partnership Diagram This diagram illustrates some of the common examples where partnership feeds into the Open Cities process. New Challenges Assessing Key Identified Actor’s Goals Provide engineering or Data Usage geospatial expertise to quality assessment processes Quality Assessment Building the Ecosystem Using the data in different or interesting ways, incorporating OSM into university curriculum, hosting mapping parties. Figure 2.1 Help design what gets collected, provide insights into key DRM Contribute datasets and urban planning challenges Defining Assessing Scope Existing Resources Provide working Provide volunteers, staff, or meeting space. Setting Up or domain experts Donate hardware, Logistics or funding Data Collection Training Talking to partners and colleagues about the value of open data and community engagement. planning an open cities mapping project which in turn encourages further collection of valuable resources to maintain. In the long-term, government can data, which feed back into a robust open data ecosystem. incorporate the Open Cities data collection methodology into national initiatives by sharing data, experience, and Organizations that participate in an Open Cities project resources. This sort of buy-in will strengthen the project (a process illustrated in figure 2.1) therefore enjoy these over time. In turn, government support can aid your benefits: project in terms of planning, legitimacy, access to exist- DD Ability to contribute to a free and open map that ing data, and provision of local resources. Government, makes detailed, relevant data available to everyone particularly the national and local agencies that already work in mapping and geographic data, can provide valu- DD Access to high-quality geographic data to drive their able insight concerning how to plan mapping activities operations and strengthen informed decision making efficiently. Their involvement also adds legitimacy. For DD Increased capacities to update, maintain, and use key example, mappers may face challenges from local people datasets by when surveying an area. With the engagement of local government or other authorities, mappers may carry DD Sharing skills and knowledge with other local ID cards or letters of approval to demonstrate official collaborators; and support for the project during data collection in the field. DD Interacting with a global community of mappers Finally, useful existing data may be found in govern- and technologists to gain exposure and education ment storage. Although convincing government bodies on new mapping tools, methods, and skills to release their data freely to the public can be a chal- DD Use of innovative approaches and cutting-edge lenge (or some data may be private or sensitive), much technology that strengthens both data collection and of the data will be more useful if opened for public management processes while reducing costs access. Furthermore, if the information collected during the project can enhance and build upon government Collaborative, open mapping has moved beyond the data, the government will benefit through the return realm of hobbyists. Many private companies, national of updated datasets. Open Cities can also be an entry mapping agencies, governments, and NGOs around the point for the government to increase its capacities, world are already using OSM in connection with serious update employees’ skills, and start integrating innovative applications, business or development operations, and tools and methods into its workflow to reduce costs and job opportunities. improve regular operations. Many different levels and parts of the government can 2.1.2 Outreach be approached, depending on the focus of the project. Government support and participation in an Open When starting a project, identify those agencies that Cities mapping project is a key strategy for ensuring its are most important to approach for support. Relevant sustainability. The OSM community is also a essential government departments may cover the following areas: resource, along with other partners such as universities, civil society organizations, and the private sector. This DD Urban development subsection examines these likely partners and their DD Disaster management potential roles. DD Mapping or surveying 2.1.2.1 Government DD Census and statistics In most contexts, governmental support will benefit a DD Information and communication technology mapping initiative because different government units DD Transportation can provide resources and ensure better sustainability and legitimacy of the work. Therefore, they should be DD Water and sanitation included as partners from the early stages of planning. Government entities will benefit from access to up-to-date, detailed data that they already devote great 22 chapter 2 | project design and preparation DD Local government (such as regional and municipal DD Wiki pages or sites. It is important to check the status governments, village heads) of mapping in certain areas by visiting the OSM Wiki site, starting with the main page at http://wiki.open- streetmap.org/wiki/Main_Page. Most countries have at 2.1.2.2 OpenStreetMap Community least one dedicated page to the status of the project. Because Open Cities mapping is built upon the OSM DD The “Neis One!” blog. Another useful starting point platform, another group to communicate with is the for connecting with the OSM community is Pascal existing OSM community. It is important to understand Neis’s blog at neis-one.org, which provides tools that what the mapping community already looks like and show which users are actively mapping in certain how the Open Cities Project might best support and areas. work with this existing group. What is the OSM community? Simply put, anyone who 2.1.2.3 Other Groups actively contributes to OpenStreetMap in a given country or region is part of that community. In active communi- Other groups may also be considered potential collab- ties, mappers communicate with each other often, plan orators. Table 2.1 describes several sources of potential mapping events, and collaborate on OSM development. partners. In quieter communities, mappers communicate less but Building partnerships is essential to reaching the high- contribute for their own personal reasons. Active, vibrant er-level goals of an Open Cities mapping project, such communities are the most productive and sustainable, as strengthening the open-source mapping community. but no matter what the local OSM user base looks like, Ideally your partners will be directly involved in the there are always possibilities to engage. planning and implementation of the project. However, Even if direct partnership is not developed with existing if key partners cannot commit time and resources, it OSM contributors, it is important to maintain commu- is still important to keep them informed about project nication with them and inform them of the project goals activities. In this way, you can keep them connected and and target areas. Since Open Cities mapping will likely approach them again later as your project expands. take place in an area where others have already mapped, the project must not overwrite the information already entered by volunteer mappers who used local knowledge 2.2 DEFINING THE SCOPE to edit the map. It is critical that the project participants, OF WORK as a volunteer community, coordinate any large data imports with the existing community and respects work The first step of Open Cities project planning is to that has already been accurately completed. clearly define the objectives. This means answering questions such as these: There are several ways to get in touch with OSM mappers: DD What is the scale of the project? Has this been done here before? DD E-mail lists. In many countries, the mappers commu- DD What will the data be used for? nicate through dedicated e-mail lists. You can find such lists at lists.openstreetmap.org. DD What data will we collect? DD Facebook. In some places, the OSM communities DD What geographic area will we cover? have chosen to communicate using Facebook groups. DD What is the time frame for project completion? Therefore, you should search Facebook to find out whether there is a local group for OSM contributors in the areas you plan to map. 2.2.1 Overview of the Process Defining your project’s scope means considering how much area will be mapped, how much data will be col- lected, and how much time the project will require—a process illustrated in figure 2.2. These decisions will 23 planning an open cities mapping project Table 2.1: Potential Partners for Open Cities Mapping Projects Universities Universities and colleges are great potential collaborators. Participation in a mapping project offers students and faculty an opportunity to learn about cutting-edge open-source methodologies, ideas, and software. Students are often technically minded, have flexible time commitments, and enthusiasm to learn new things. Mapping projects that involve academia may also evolve into a permanent part of the university’s curriculum or even a course by itself. Additionally, Open Cities partners that build university capacity in this way help to develop the qualifications of students who can later be hired as interns or staff. Scientific Local scientific communities, whether involved in university research or in civil society communities organizations, can be called upon to support an Open Cities project in several important ways. First, these are important groups to involve in the data modeling process. Civil engineers, planners, and others have experience with relevant data and analysis, so they may be able to provide suggestions as well as important local context. Second, data quality assessment throughout the project, particularly at the end of the data collection period, is critical, and this group may be well placed to support it. Technical Open-source software communities (user groups and private companies) are often closely communities linked with OSM communities. Freelance software developers, geographic information system (GIS) specialists, and private software companies may be interested in providing software that assists the project as well as services for the community. The technical community also can share its collective professional expertise when hosting a skills workshop, and service providers can rent office space or equipment. For example, the widespread Open Source Geospatial Foundation (OSGeo, at osgeo.org) is closely related to OpenStreetMap. Civil society Civil society projects may use datasets similar to those of the Open Cities project, and organizations potential partnerships should be explored. The synergies that a collaborating NGO can bring to Open Cities include data sharing and collection, youth outreach, and additional connections to the community. International Many large development projects are funded by international organizations that have organizations significant datasets from previous projects. Their interests focus on a variety of areas and potential synergies may exist. Offices such as the United Nations Office for the Coordina- tion of Humanitarian Affairs (OCHA) often have existing mapping capacity and already have knowledge of the local context and existing data. Youth and adult Community groups, by their nature, attract members of society who are interested in community volunteering their time for the benefit of the community. A great example comes from Indonesia, where HOT worked with the Scouts to map thousands of buildings in the groups Bengawan Solo River Valley. 24 chapter 2 | project design and preparation enable you to determine your personnel and resource To determine which types of features and attributes the requirements as well. project should map, you will want to consult with the partners involved in the project. For example, you may Although your long-term objective might be to map a be working closely with the government’s water and san- large area such as a city, district, or even a country, it itation department—and they might not need buildings’ is advisable to begin with a smaller pilot project (box structural characteristics but will be very interested in 2.1) and later scale up. This provides the necessary time using data about drainage lines and solid waste. and experience to build capacity in your team, nurture partnerships, and gain experience in the process. One potential challenge is that, with numerous partners wanting different information, you may end up with The following sections will explain how to define the requests to map many different features along with target area and make rough estimates of the number a wide range of attributes for each. Such a large data of features to be mapped. We will discuss what a data model may not be practical to map with the resources model is, how it should be defined, and how to engage at your disposal. Increasing the number of attributes partners in this process. Last, we will consider how to collected has, in previous projects, decreased both the manage a project’s time frame and ensure that tasks are quantity of features mapped as well as the quality of the properly done in succession, so that your project effec- data. It is therefore important to limit the amount of col- tively meets its deadlines and its goals. lected information to only the most important features and attributes. Keeping things as simple as possible will 2.2.2 Why Are You Collecting the help ensure project success, particularly in pilot projects. Data? One of the first questions in designing a project is 2.2.3 Defining the Target Area simply, “What will the data be used for?” In other words, You should precisely determine your intended survey why are you collecting data? You should have a good area from the start. Are you mapping an entire city? Part answer for this at the outset because it will define the of a city? An arbitrary area? sort of data that you want to collect. Let’s say that you are mapping a city. You may simply For example, Open Cities projects are often oriented think, “We are mapping City X. That is the area, and it toward disaster risk reduction. In the case of disaster has been defined.” But what does that mean? Does that preparedness, this might mean collecting data about mean the official government boundaries of City X? buildings and other critical infrastructure. For starters, Are we interested in the entire urban area or only the the data would include the location of each building and city center? If the target area coincides with an existing its geometry (the building’s footprint on the ground), but administrative division, the area will be easy to define, it also includes information about other building attri- though it might not follow natural roads and features as butes such as the number of levels in the building, the well as you might like. shape of the roof, and the materials used for the walls. Box 2.1 Pilot Projects Pilot projects typically map a small part of a much larger area, over a short time frame. The goals of pilot projects are not only to collect data but also to experiment with the surveying methodology and determine the most effective and efficient ways of mapping. Such projects are also useful for demonstrating the effectiveness of innovative mapping tools and strengthening local support. 25 1 COMMUNITY MAPPING What Area Should Be Mapped? What You Province Need to City Answer First Neighborhood The project’s scope is informed by When marking out the total area (or extent) the goals enumerated at the outset. to map, constraints around time, budget, and partnerships should be realistically assessed by core decision makers in order to determine the Every project requires a clear-eyed physical size of the project. assessment of each individual locality and situation. Most community mapping efforts are SRI LANKA EXTENT MAPPING mobilized around a prior identified In Sri Lanka, Open Cities has focused on the need, so it is up to project leaders city of Batticaloa, which is affected by flood to best utilize resources to meet as well as potential for hurricane and storm their specified goals. surge. At the start of the project, there weren’t available resources to map the entire at-risk area so in consultation with government partners, Open Cities started a pilot project to map all the buildings and roads in the Manmunai North Divisional Secretariat (DS) that covers an area of 68 km2 and about 90,000 people. Figure 2.2 2 3 What Attributes to Collect What Objects within that Area? About these Objects? Hospital building name roofing type # of stories School Clinic building population Further specificity on a project can be enumerated The attributes to map is important when if certain priorities have identified the need considering a specific population’s vulnerability to to focus on mapping a certain type of object certain hazards, or it could perhaps be a matter of or structure. These decisions might center on improving outdated or incomplete risk data. mapping the structures of greatest vulnerability, or some other situation-specific reason. DHAKA PROJECT SCOPING KATHMANDU DATA MODELING Open Cities Bangladesh undertook detailed In Kathmandu, Open Cities project participants mapping of a section of the Old City in Dhaka. Key mapped schools and health facilities in order to partners in the project included government water support a risk modeling effort designed to help and sanitation experts and a local NGO focused prioritize seismic retrofitting. To fit the needs of the on heritage preservation. Therefore in addition risk model, specific attributes for each structure, to collecting building and road data for use in such as the number of stories, age of the building, emergency preparedness, the team also identified and construction type were necessary. These historic buildings and collected locations of wells, attributes were developed in partnership between drains, and sewers. risk experts and engineers from the World Bank and a Nepalese scientific organization. planning an open cities mapping project Finally, it will benefit you a great deal to know as much is a good idea to write down the list of features you want as you can about your target area from the beginning. to collect and, beneath each feature, to list the attributes Think about questions such as these: you want to collect. For example, one feature you will likely be mapping is roads. Defining the data model for DD How large is the target area in square kilometers? roads means determining which attributes you want to DD How many people live there? collect, such as the following: DD How many buildings and roads are in the area? DD Type of road DD What will be the main challenges of mapping this DD Name of road (in one or multiple languages) area? DD Surface types of road (asphalt, concrete, brick, stones, gravel, or unpaved) 2.2.4 Building the Data Model DD Direction of traffic (one-way, two-way, or variable by A data model is a detailed description of all the data you time of day) want to collect in your field survey: the features you want to identify and the attributes of those features you want to collect. To a large extent, the data model will relate to 2.2.5 Estimating a Time Frame the available time and resources for the project—either Project duration may not be something you can define determining how much you will need or what the from the start, but you should be able to estimate it. Will constraints might be due to predetermined limits. To it take a month? Four to eight months? A year? This will complete this data model, determine how much data you become clearer as you plan your project and consider the will collect, which methods you will use to collect it, and logistics, but you should begin developing an idea of the whether privacy or sensitivity concerns will limit what time required to complete the project. you collect. Be sure to think through the factors that will affect your Quantity of data. Defining a data model may be easy or it survey work, and include them when working on your may take some time. If many partners are involved who time frame. Will major holidays occur during the pro- want to collect many different types of data, deciding posed time period? How many days each week do you on which data to collect could prove to be a challenge. expect your staff to work? How many hours per day will Remember that you may not want your surveyors to they realistically be able to work? Keep in mind consid- collect 100 or even a couple of dozen attributes about erations such as regular work breaks, time for prayers, every feature on the ground. The more attributes you time for classes (if the staff will include students), travel must collect, the greater the time and resources you will time, and so on. Figure 2.4 illustrates part of a sample need for data collection. project time frame. Collection methods. You will also need to decide whether During the Open Cities Dhaka pilot project, for example, surveyors will collect only those attributes that are a single survey team of two could collect data for about directly observable, or whether they will be required to 50 buildings per day by mapping for three or four hours. speak with building owners. Speaking with building In some locations, it is common to cover 100 or more owners will take considerably longer than surveying buildings per day. Use metrics like these to estimate simply through observation. This decision will greatly how long it will take to map a certain region. If you can affect the surveyors’ mapping speed and therefore the estimate the number of features in an area and the time time frame and cost of the project. it will take to map them, you can better understand the likely time frame of your project. Sensitivity of data. Remember also that the data you are collecting will be shared on an open-source database. You will not want to collect data that may be considered private or sensitive. Figure 2.3 shows part of a sample data model (for a com- plete sample form, see Appendix 6.1, “Data Design”). It 28 chapter 2 | project design and preparation Figure 2.3: Sample Data Model of Features and Attributes Proposed Data Model Streets Name & Address Width Surface Type (Bituminous, Pitch, Brick, etc.) One way / Two way / not defined Buildings Footprints Usage Number of Stories Construction Type (categories consistent with CDMP 2007 survey) RCC/Brick/Block/ 2.3 BUILDING THE TEAM that these two tasks be split, or that different people are editing in the office at different times. (STAFFING) While preparing for your project, consider the logistics 2.3.1 Management Structure, Roles, that will be involved in terms of staff and office work. Remember that half of the job involves surveying the and Responsibilities target area, and the other half is editing work on com- You can organize your staff in various ways. A single puters. You may have a small project team, or you may manager can supervise a large group of mappers, or you have dozens of people or more. Box 2.3 discusses some may choose to divide the team into a number of smaller of these considerations. teams that can operate somewhat independently. (Box 2.4 describes some of these roles.) Either way, you will Consider whether all of your staff will be editing, need people to perform the following tasks: or whether you will divide the tasks of surveyor and editor. It is generally beneficial for mappers to do both DD Field surveying surveying and editing, rather than splitting the work, DD Editing because mappers remember what they have surveyed; even with detailed notes, nobody will be able to edit the DD Printing Field Papers (see 2.4.3) map as well as the person who surveyed the features DD Managing equipment being mapped. Editing tasks also help to overcome DD Organizing workflow “survey fatigue”—when surveyors tire from spending most of their days walking around to perform a single DD Procuring office supplies task (and vice versa for editors staying at the computer DD Quality assurance all day). That said, logistical considerations may require 29 planning an open cities mapping project Figure 2.4 Sample Project Time Frame Activity 7-Jul 14-Jul 21-Jul 28-Jul 4-Aug 11-Aug 18-Aug 25-Aug 1-Sep 8-Sep 15-Sep 22-Sep A1 Design and preparation A1.1 Define the data model A1.2 Develop data capture survey, tagging guidelines and strategy A1.3 Hire Staff A1.4 Find workspace and acquire equipment A2 Project launch A2.1 OSM training for managers A2.2 OSM training for mappers A3 Data Collection A3.1 Mapping of target area A3.2 Map beyond target area if ahead of schedule A4 Data Monitoring, Evaluation, and Validation 30 chapter 2 | project design and preparation DD Technical support 2.3.2 Finding Mappers and Surveyors DD Outreach, training, and volunteer mobilization Survey staffs are often sourced from universities and from recent university graduates. The demands of Mappers are responsible for the day-to-day activities of a lengthy field activity can be demanding and is suitable project. They are the ones going out to different locations for young people who have an interest in fields related each day within the target area, identifying features, to GIS, geography, urban planning, and engineering. and writing down attributes. They also spend time in However, the surveying and editing process is not the project office, editing the data they collect each day difficult; with the right training, anyone can fill the role, and adding those data to OpenStreetMap. Because their even those with limited computer experience. Those skills will grow, the quality and pace of their work will with less computer and geography experience may take improve over time. Mappers may also perform office longer to train but can still fulfill the job requirements. tasks, procure supplies, work with volunteers, or train Mappers are also often found through other partners, partner organizations. such as government agencies, youth groups, and high schools. A good manager should keep the mappers on sched- ule and help them as needed. Managers will be more As you reach out to potential partners, key influencers experienced in GIS and OpenStreetMap, so that they can and casual mappers will be identified who are interested effectively advise mappers during the editing process. in the project focus and can be added to the team as the It is important that managers check for mistakes and project expands. A unique aspect of the Open Cities data errors in the data, especially early in a project, so that collection methodology is the use of crowdsourcing from they can inform mappers of their mistakes before they both the international and local OSM mapping commu- are repeated hundreds of times. nities. When engaging the local community, it is often students who have the time and interest to contribute to Good managers also maintain strong communication the project. An Open Cities project provides the opportu- with their mappers and should occasionally accompany nity for knowledge exchange between academia and the them during surveys to observe their work and offer implementing organization. As a result, the surveyors advice. Managers are responsible for performing quality can be sourced from local university students and recent assurance surveys, which will be discussed in section graduates in geography, computer science, or other 4.3, “Quality Control.” project-related technical disciplines. Box 2.2 Different Methodologies Different structural approaches to mapping offer varying advantages. This guide focuses on the most common approach: systematically mapping an area with an Open Cities project team. But the same task can also be completed in other ways, such as by engaging volunteers, hosting mapping parties, or working through local governments. These alternative methods tend to take longer and be less precise, but their advantages should not be dismissed. Remember that the more individuals and organizations that are involved in your project, the stronger it becomes. The more participants you have, the more users, contributors, and awareness of the data you will have—which, in turn, strengthens the open data ecosystem for the future. We recommend engaging a project team because of the practical goals of a mapping project, but you are encouraged to combine these methods to engage others in the mapping process to realize these additional benefits. 31 planning an open cities mapping project Box 2.3 Common Roles Open Cities mapping project teams generally include people in the several roles described below. Surveyor and Editor (Mapper) Surveyors typically work in pairs. Each pair is responsible for field surveying—usually during half of the day, as it can be quite physically tiring. They may also engage in small tasks such as printing field papers. In most projects, each survey team also edits its own work, but these roles can be split. Each day, the pairs will be tasked with part of the target area to map, and they will spend three to five hours surveying, with the remainder of their time spent editing the data. Some of the experi- ence mappers should dedicate some or all of their time to the quality assurance process. Manager A small pilot project may have only a single manager, or there may be multiple team managers, each of whom manages a subset of mappers, and who report to a top-level project manager. Man- agers are responsible for organizing the project workflow, supervising and aiding mappers, and managing equipment and logistics. Project Lead At the top of the management structure is typically a project lead (PL). The PL may be the same person as the project manager, or the two roles may be separated. Ideally, the PL will have in-depth local knowledge and relationships with government and academia. The PL, aside from supervising the mapping project, is usually responsible for high-level discussions with partners and community outreach. The PL should be well versed in field data collection and skilled in project management, mapping and GIS, internal and external communications, social networking, and community building and integration. 2.3.3 Compensation and Incentives concerns whether mappers should be paid during a tar- geted mapping project. On the one hand, it is preferable There are various methods for compensating project to find ways to encourage mapping other than payment members depending on their background and interests. in order to foster a sustainable initiative and mapping If you assemble a management structure with a team community. On the other hand, a targeted mapping of managers (known as a core project team), they will project may require mappers to dedicate a great deal most likely be paid as full-time staff. Others, such as of their time to complete the work within a given time volunteer students or young professionals, may either be frame. In the end, the choices about how to compensate strictly volunteers, incentivized through an internship or incentivize mappers will depend on the particular or academic credit, or offered per diem or even full-time demands of the project. payment. Volunteer mappers can be incentivized by designing the Because OpenStreetMap is an open data project built position as an internship opportunity, by working with around a community of volunteers, a frequent debate 32 chapter 2 | project design and preparation universities so that they receive academic credit, or by can be used to accurately trace polygonal features into simply offering the reward of community service and OpenStreetMap. skills development. The broader benefits of participation were discussed earlier, in section 2.1.1. 2.4.1 Existing Digital and Paper Data Another approach toward providing incentives is to Because Open Cities uses the OSM platform, there will develop competitions between universities, departments, almost certainly be preexisting data on the map. The or parts of the mapping community. Although monetary OSM project has been around for nearly a decade, and rewards have been found to lead to a spike in data col- there are few places in the world without at least a few lection, collection may also dramatically decrease after roads already drawn. This is helpful because it means a competition is over. Therefore, although competitions that Open Cities projects do not start from scratch. It provide a potentially useful way to gather a lot of data, is important to remember, however, that your project they are not necessarily a good way to support commu- mappers should not delete features that others have nity growth in the project focus area. In the case of a created. The purpose here is to collaborate and work with mapping competition, creative prizes or rewards (such as other users as well. Once you have defined your project mapping hardware) can be used as incentives that may target area, it is a good idea to examine the existing OSM not result in temporary, opportunistic mapping. map and assess how detailed it already is. You should also contact any active OSM users in that area to let them know what the project is about and to see whether there 2.4 ASSESSING EXISTING MAP are any opportunities for collaboration. RESOURCES Often, other sources of data will also be available to you. Once you have established the foundation of your Public government data might exist in either digital or project, you must assess the resources that are available paper format. In Open Cities Sri Lanka, regional paper for mapping. For example, maps already existed in each district. Surveyors used them as references to make improved, digital maps with DD Are there any preexisting data that will be useful and OpenStreetMap. available? Digital data often exist as well. In the Open Cities Dhaka DD Is the target area covered by Bing or Mapbox imagery? pilot project, for example, the team received permission Are other sources available? to use old government road and building data. They DD Do you have GPS units available? Mobile phones? Will decided it would be easier to use these data as a base you need them? layer that could be updated rather than starting from a clean slate, and so the data were first imported into The purpose of such resources is twofold. First, it is OpenStreetMap. always better to start with some objects already on the map, rather than with an empty canvas. If you already 2.4.1.1 Data Imports have two streets drawn on the map, you can easily add a school at their intersection. But if you don’t have those Importing data can save time and work if there are exist- streets drawn on the map, you have no landmarks by ing data that correspond to your data model and that you which you can correctly locate the school, thus requiring are free to use. For example, if you plan to map 5,000 much more work. Therefore, it is always best to have an buildings and the city government has footprints for existing base map before going into the field. all those buildings, you might consider importing the data into the OpenStreetMap database. Box 2.5 further Second, these resources aid you while you are mapping. discusses how to work successfully with the OSM com- For example, if you are carrying a satellite picture munity on data imports. with you while out surveying, you can usually identify objects by comparing what you see to what is shown on If it takes 30 seconds to draw the geometry of a building, the image. This will also help you to understand the then it would take approximately 40 hours to draw 5,000 correct sizes and shapes of features. Later, the imagery buildings—a week of someone’s time. 33 planning an open cities mapping project Before you think about importing data, you must con- toward urban development. Such agencies may possess a sider the following: wealth of urban-oriented geodata. DD Are the data of good quality? How and when were OCHA. The UN’s Office for the Coordination of they created? Are they better than, as good as, or Humanitarian Affairs has had a local presence in many worse than data you would collect yourselves? countries for many years. It typically has plenty of local knowledge and often is more willing than government DD Are the data in the public domain, or do you need agencies to make its data available. explicit permission from the agency that created the data? NGOs. Typically hundreds if not thousands of NGOs DD Is the net benefit of importing the data greater than exist in a given locale, and those working in certain the cost of creating it new? fields may have collected data from previous projects. Organizations working in disaster relief, water and san- DD Who will import the data? itation, or urban development are good places to begin asking around. Written permission is generally required from agencies that are willing to donate data to OpenStreetMap, and they must understand that once they offer the data, the 2.4.2 Availability of Imagery for data will go into an open-source database that anyone Digitization can use. Getting this permission can be easy in some Perhaps the most effective aid in mapping is aerial places, but often agencies are rather protective of their imagery, as shown in figure 2.5. Aerial images are photo- data. This is particularly true of governments that are graphs taken from satellites or airplanes, which are then not familiar with the ideas of open data, and it can be oriented accurately by their location on Earth. a challenge to get their permission and understand the meaning of an open license. For a sample permission Aerial images can be viewed as a background in the letter, see Appendix 6.3. OpenStreetMap editing software. Mappers can use what they see to trace lines. This can be useful before field Importing data is not difficult, but it requires an surveys, to create a skeleton base map that shows roads, advanced understanding of OpenStreetMap and can buildings, and anything else that is identifiable solely take a significant amount of time, because imported from the imagery. data must be manually merged with existing OSM data. If you do decide that importing data is the right Aerial imagery can also be used during field surveys course to take, you must have a clear plan of how to and editing. It can be printed out and carried in the import the data and discuss it beforehand with the OSM field, helping mappers to orient themselves and identify community. features on the ground. Later, when they are editing, they can reference the sketches and notes they made in 2.4.1.2 Other Local Data Sources the field with the imagery in the editing software and accurately draw objects on the map. Consult the sources described below to find additional existing local data. While preparing for a project, the identification of suitable imagery should be undertaken early. If good Mapping agencies. Most national governments have quality, high-resolution imagery is available, this will a department or agency that deals exclusively with benefit your project tremendously and mean less time mapping. Whether it is called the survey department or and less work. If you cannot find good imagery, however, something else, it is responsible for managing a coun- other methods will be required for digitizing objects on try’s geographic data. If the department or agency is the ground. willing to provide some of its data as public information, it can be a prime source for acquiring existing data. 2.4.2.1 Assessing Bing and Mapbox Imagery Urban development agencies. Some cities or countries Several imagery sources are available to OSM users, but have government departments specifically oriented those with the widest coverage and the highest-quality 34 chapter 2 | project design and preparation Box 2.4 Data Imports and the OSM Community What is the OSM Community? OpenStreetMap is a collaborative project involving hundreds of thousands of users and contributors, and the “OSM community” comprises those active members who continually participate by adding to or updating the map. Each country has what might be called its own community, a group of users who communicate, plan activities, and work together to improve the map. And each country’s community is also con- nected to the larger global OSM user base. Because an Open Cities project uses the OSM platform, it is automatically engaged in this community. Section 2.2.2.2 earlier provided more information about engaging the local OSM community. Most OpenStreetMap users go out and contribute to the map bit by bit, so they should be consulted before importing their data. The import might overwrite the work of other mappers who have actually conducted field surveys. You must also send an e-mail to imports@openstreetmap.org and let them know what you are planning. Ensuring that they are aware and supportive of the process before any importing takes place can save a lot of time explaining or reverting the changes after the fact. For a complete description of the steps to be taken for an OSM import, view this OSM Wiki page: http://wiki.openstreetmap.org/wiki/Import/Guidelines. images are Microsoft’s Bing search engine (figure 2.6) area will require another imagery source or a different and Mapbox, which provides open-source base maps, methodology. design tools, and OSM-powered street data. Box 2.6 discusses the importance of using only those sources 2.4.2.2 Alternative Imagery Sources that have given permission to use the imagery. Finding alternative imagery sources is a lot like finding After defining your target area, you should assess to possible data you can import. There is usually some what extent you will be able to use this imagery, assum- imagery floating around, but it is often of questionable ing that you plan to use it for your survey. If you are age and usefulness, and of even more questionable legal lucky, high-resolution imagery will cover your entire right to use. target area. Usually, however, parts of your target area will be obscured by clouds (figure 2.7) or not have Typically satellite images that have been purchased from high-resolution imagery at all. a major provider like DigitalGlobe are not licensed for use in mapping in OpenStreetMap unless specifically During the planning stage, consider using GIS software designated as such, and any company or agency that to draw lines around the parts of your target area where has these data does not have the right to reuse the data Bing imagery will be suitable for mapping as well as for this purpose. However, imagery created directly by those areas for which usable imagery is unavailable. This governments and companies may be usable. If someone will help you understand what proportion of the target is willing to donate imagery, the license should be inves- tigated at the start. 35 planning an open cities mapping project Figure 2.5 Use of an Aerial Image to Trace Lines on a Map Source: © 2014 Microsoft Corporation. Image provided courtesy of USGS Earthstar Geographics. Imagery can be purchased from companies such as 2.4.3 GPS Devices, Field Papers, and DigitalGlobe and GeoEye, but this can quickly become expensive. The cost per square kilometer is around Smartphones $15. It is usually best to find freely available sources of If imagery is unavailable for part or all of your target imagery, which are also more sustainable in the long area, having a good base map becomes even more run. important. If detailed, existing data such as road net- works and landmarks are already on the map, you should In some cases, imagery is made available through be able to use this as a reference to add new features in the U.S. Department of State’s Imagery to the Crowd their correct locations. If a good base map is not already initiative. This is typically provided in the aftermath available, you will need to use GPS to build up a base of disasters to aid remote mappers. Learn more about map of your area. Imagery to the Crowd at mapgive.state.gov. GPS devices allow mappers to record paths along which Remember, too, that using aerial imagery as a back- they walk as well as individual locations, which are ground layer is generally considered to be the easiest way stored with accurate latitudes and longitudes. These to map, particularly when many buildings need to be data can be imported into JOSM and then traced to add digitized. features to the database. When no imagery is available, it is a good idea to use GPS to build up a base map first. Survey teams can first traverse an area with GPS devices, locating roads and major landmarks. In a second stage, 36 chapter 2 | project design and preparation Box 2.5 Imagery and Licensing A common question during any mapping project is, “Why can we use imagery from Bing but not from Google?” The short answer, and one not always easily understood, is licensing. Any company or organization that sells or distributes imagery has a license describing what users are allowed to do with it. In most cases, people are forbidden from making derivative products. This means we are not allowed to use imagery to help make maps unless we receive specific per- mission to do so. Because Google has not given permission for the OSM community to use its imagery—or any other information from Google—as a resource, such use is not allowed, and edits derived from Google are illegal. On the other hand, Microsoft has given explicit permission for OSM contributors to use its imagery, which is of good quality in many parts of the world. DigitalGlobe, a satellite imagery company, has also released a great deal of imagery through Mapbox. It is always best to check these resources first and to assess the quality of the images in your project’s target area. Available imagery sources can be viewed in the Java OpenStreetMap Editor (JOSM). Figure 2.6 Example of Good Aerial Imagery from Bing 37 planning an open cities mapping project Figure 2.7 Cloudy Aerial Imagery they can use this base map as a reference to go out and Each day, mappers will print out Field Papers of small collect more detailed data. areas, which surveyors can use to sketch buildings and areas using the correctly located roads as a reference.1 A service for OSM known as Field Papers enables Figures 2.8 and 2.9 show examples using Field Papers mappers to print out a piece of paper that displays either “atlas maps.” an existing OSM map, satellite imagery, or a combina- tion of both. The mappers take the Field Papers into the Last, you can use mobile applications to collect Open- field, draw on them, write notes, and add features to the StreetMap data. Although applications have been map—an example of which is shown in figure 2.8. Later, developed to map using mobile devices, none of them the papers can be scanned back into the computer and provides all of the editing features of desktop software. automatically geo-referenced to their correct locations Mobile applications are best at mapping points of on the globe. It is then a simple process of digitizing the interest.2 hand-drawn notes. This process can be used for building up the base map, as well as for the daily mapping activities of a project. Creating rough sketches of building shapes and their locations is not very precise, and often mappers wonder whether this is a problem. Ideally, building 1 shapes will be more accurate, but such accuracy is difficult without imagery to use as a reference. If this mapping methodology is used, it is not a major problem. More important than precise building size and location is the topology—which helps to ensure that the buildings are placed correctly in relation to 38 other and in relation to the (accurately positioned) roads. each chapter 2 | project design and preparation Figure 2.8 A Field Papers Atlas Map, without Imagery, Used in Sumba Barat, Indonesia Source: Field Papers (http://www.fieldpapers.org). For a list of Android editing applications, see http://wiki.openstreetmap.org/wiki/Android#OpenStreetMap_editing_features. 2 39 planning an open cities mapping project Figure 2.9 A Field Paper Atlas Map Example Source: Field Papers (http://www.fieldpapers.org). 40 chapter 2 | project design and preparation Box 2.6 Design and Preparation Checklist During the project’s design and preparation phase, the following tasks should be completed: Partner Outreach rrIdentify relevant partners in government, academia, and civil society rrMeet with potential partners, discussing support and areas of collaboration rrCommunicate project goals with OpenStreetMap community and meet with local mappers Defining the Scope of Work rrObtain a map containing the boundary of the target area rrEstimate the number of features you will map in the area, such as roads, buildings, and so on rrWrite out your data model rrMap the data model to OpenStreetMap tags rrEstablish a rough time frame for survey completion, noting holiday periods that will halt the work Building the Team rrEstablish management structure rrDetermine sources of your staff and how staff will be compensated rrHire staff Assessing Resources rrList the existing digital and paper data rrDetermine whether you will import any data; if so, create an action plan and time frame rrDraw a map showing parts of your target area where no imagery is available rrDetermine whether you will use GPS devices, smartphones, or both; if so, acquire rrConfirm that the necessary base-map data are available to use in Field Papers 41 42 3 GETTING STARTED CHAPTER HIGHLIGHTS DD Finding an appropriate workspace DD Obtaining the right equipment DD Training staff 3.1 FINDING A WORKSPACE If Internet access is often lost, consider purchasing backup connections such as 3G modems. The additional A major consideration is the location of the project cost often will outweigh the cost of lost time while office. If you already have an office, you should consider waiting for service to return. whether it is in, or close to, the target area. Remember that travel between the editing location and the area Other considerations for your workspace include oppor- being mapped is time-consuming; hence, planning tunities for community engagement and collaborative transportation logistics is important. If you can save work. While not strictly necessary, it may be beneficial your mappers some time, the project should take less to have a meeting room that can accommodate a dozen time and cost less money. It may be worthwhile to orga- people—ideally one that is easily reconfigurable to nize an office space close to the target area so that the serve a wide variety of functions. Training could also be mapping process runs more smoothly. conducted here, although if you have many mappers to train, you may need to identify a larger facility for the In any case, care should be taken so that the office used initial training. for editing has a minimum of electrical, computer, and Internet problems. Because OpenStreetMap editing When possible, the office should provide a creative com- works best with a good Internet connection, frequent munal space furnished with items such as whiteboards disconnection is enough to effectively halt the process, and couches. This will encourage open discussion and wasting valuable time and slowing progress. If your staff facilitate brainstorming to identify current and future comes in for editing and has no Internet access, they challenges, examine alternatives, and develop solutions have nothing to do and you lose half a day of time. A or alternative actions. prolonged outage can be even more costly. 43 planning an open cities mapping project The initial project phases involve engaging the local 3.2.2 Printing Needs OSM community and hosting mapping workshops (or Printer: A printer will be needed to print the Field mapping parties) to raise awareness of the Open Cities Papers. If you are using paper survey forms to collect project and identify interested individuals, communities, attributes for many features such as buildings, you may schools, government departments, and private organi- be making thousands of photocopies. This process takes zations. Mapping parties can be organized either at the time and money, and it should be considered while plan- project office or at other venues in the focus area such ning logistics. For example, a pilot project in Sri Lanka as university labs, private or public offices, community faced the challenge of photocopying over 30,000 data centers, and libraries. Of course, the budget should take entry forms during the course of the project. into consideration the potential costs of renting these temporary venues. Business cards and brochures: These materials are useful when establishing credibility in the field. Com- munity members can be suspicious of mappers collect- 3.2 EQUIPMENT AND OTHER ing data, and it is necessary to ease their concerns. A COSTS printed brochure will outline the project methodology and goals. Cards and brochures can easily be distributed When you are preparing a mapping project budget, while collecting data in the field—raising awareness of be sure to consider a number of other potential costs, the project in the community and boosting credibility including the equipment, supply, and service expenses when fielding questions. Business cards and brochures described below. See box 3.1 for an estimate of the hard- should be distributed at all presentations and meetings. ware needed to serve a 10-person project team. Event signage and promotional materials: Other printed materials include banners and posters for larger events 3.2.1 Hardware Needs such as mapping workshops. Computers: Data entry and analysis for OSM generally requires desktop or laptop computers. Laptops, being portable, can also be used to facilitate training sessions 3.2.3 Mapping Workshop Expenses or mapping parties at venues away from the team’s Rent: Community mapping workshops should be held in office. Typically, one computer for every two members of a variety of locations. Where possible, a host should be your mapping team will suffice for most needs. identified to provide the venue. If this is not feasible, a venue should be rented. Projector: A projector is required for sensitization and awareness-raising presentations and in mapping work- Food: Although workshops will vary in length, they shops where the venue does not provide a projector. It should typically be half- or full-day events. Therefore, will also be used when these presentations are offered at provide a meal such as lunch for the mappers along with the office. coffee and snacks. For community mapping parties, it is customary to supply a cake that depicts a map of the Handheld GPS devices: GPS units will record the “track” local area on the top. of the mappers (where they went while collecting data) and can record points of interest (POIs). Internet connectivity and electricity: Internet connectivity should always be considered and, where possible, redun- Mobile Android mapping devices: Instead of GPS units, dant systems established as an option in case connec- you may opt to use Mobile Android devices. A variety tivity fails. Likewise, rechargeable batteries should be of smartphone applications provide specific OSM available for the projector and any GPS devices. mapping functionality. Currently, Android applications such as OSMTracker are the best. Mobile devices such as these usually have GPS chips inside them, allowing 3.2.4 Surveying Expenses them to behave like a GPS device, but with additional Surveying and office supplies: The mappers will need functionality. clipboards, pens, and pencils to complete surveying tasks. 44 chapter 3 | getting started Box 3.1 Hardware List The following hardware will serve a project team of at least 10 people, depending on mapping methodology: DD 5 laptops (assumes one mapping team per laptop) DD 2 GPS devices (assumes you will use GPS, but not all the time) DD 1 printer (for printing Field Papers and survey forms) along with necessary ink and paper DD 1 projector (for presenting the project to partners and for trainings) Transportation expenses: When specific sites have been each day to practice surveying and then editing in the identified to map, transportation may be required to take office. Although they were not experts after a week, they mappers into the field and return them to the venue, were able to lead a two-day training for a larger group of where data can be entered to OSM. mappers and answer most of their questions. 3.3.1 Training on Mapping Tools 3.3 TRAINING Basic OpenStreetMap training can take as few as two Once you have established your management struc- days. This is a typical amount of time to work through ture and hired staff, you will need to train them. On a the beginner training material available on learnosm. practical level, they will need to know how to map with org. However, learning OSM in depth requires time and OpenStreetMap, how to use the JOSM or iD editing plenty of practice. Training should include both class- software, and how to create and use Field Papers as well room time to work through the software step by step and as any specialized information about the data they will multiple field excursions. be collecting. For example, if they are collecting data on buildings’ structural attributes, they will need to learn While training core staff, it is a good idea to dedicate at some engineering basics. We can separate the training least one week and sometimes two simply to mapping components into two broad areas: those concerning and training. That way, the staff will have plenty of OpenStreetMap tools and software and those concern- supervised practice before they supervise others. This ing surveying techniques and understanding the data training period can also overlap with project preparation. model. Even after training is complete and mappers are doing their job, take care to watch for mistakes and help the Take the time and care to train the staff thoroughly. mappers continue to learn. Remember that you are essentially training your future trainers, because good managers and mappers will be A typical OSM training outline includes the following: able to teach others about mapping and offer trainings 1. Introduction to OpenStreetMap and with partner organizations and with volunteers. If it fits signing up for an account within your management structure, it is a good idea to a have a core team of managers who are more extensively 2. Using OSM’s online editor, iD trained than the mappers whom they oversee. The 3. Using the offline editing software, JOSM core team should become experts in OpenStreetMap and surveying techniques, so they will be equipped 4. Understanding OSM concepts such to manage and provide answers to other mappers. as nodes, ways, and tags In Open Cities Dhaka, for example, a core team was 5. Mapping in the field trained over the course of a week and went to the field 45 planning an open cities mapping project 6. Using aerial imagery to edit likely need to learn some engineering basics in order to identify various attributes. They will also need precise 7. Using a GPS device to map definitions of those attributes. For example, if one of the 8. Creating and using Field Papers to map attributes you are collecting is the building’s condition, and the possible values are “poor,” “average,” and “fair,” These are the main areas to cover when training you will need a precise description of each of these beginners, but the list can easily be either expanded or subjective terms. shortened, depending on the training needs. For specific It is a good idea to develop a survey training manual for training materials, visit learnosm.org and MapGive’s mappers that includes clear descriptions of the various short tutorials (mapgive.state.gov). attributes you are collecting along with photographs. In the Open Cities Dhaka project, for example, the survey 3.3.2 Training on Surveying required detailed engineering knowledge. The team’s The second training component concerns your project’s training manual clearly showed the differences between specific data model and how to collect accurate attributes types of masonry, for example. It was also particularly in the field. For example, in the pilot project in Dhaka, valuable to differentiate subjective attributes, such structural attributes about individual buildings were whether a building’s physical condition was “poor,” collected, such as the type of structure and whether the “average,” or “good.” Mappers had clear definitions for building had vertical irregularity. Because not all of the each so they could accurately and uniformly identify surveyors were not engineers, this training required those values in the field. several days, including practice identifying buildings in All said, a week of training is probably more than ade- the field. quate for most projects, and this can be combined with The amount of training time that should be dedicated actual mapping activities that managers lead or super- to this component depends on the complexity of your vise. The more experience mappers have, the better they data model and the background of the mappers. For will do, but they will need ongoing guidance as they example, not much instruction is required to collect only learn how to map properly. street names and surface types. But if you are collecting detailed structural attributes of buildings, mappers will 46 chapter 3 | getting started Box 3.2 Project Preparation Checklist During the project set-up phase, the following tasks should be completed: Logistics rrIdentify where your staff will work rrEnsure reliable Internet connectivity rrList the equipment you need and acquire it Training rrIdentify who will conduct the training and what material will be covered rrCollect necessary training materials, develop training manual rrIdentify a location to host the training (maybe the same as the project office) rrCreate a training schedule 47 48 4 IMPLEMENTATION AND SUPERVISION CHAPTER HIGHLIGHTS DD Managing the data collection process DD Implementing measures for quality control DD Project reporting Now that we’ve discussed the planning stage and project We will assume here that the surveyors are using Field logistics, we will consider the activities that go on during Papers and any additional survey forms required for a mapping project. We will discuss the methodology mapping. We assume this because, as noted in the for data collection (known as the data capture strategy), chapter on project design, Field Papers can be created project scheduling and daily workflow, quality control, even when imagery is not available. For now, we will and reporting. ignore the non-paper-based methodologies, such as the use of GPS and mobile devices. Of course, you are free to include them in your process as you design it. 4.1 COLLECTING THE DATA A principal consideration in a mapping project is the 4.1.1 Creating Reference Maps data capture strategy. How will your mappers go out Each day, mapping teams will venture into the field each day, collect information, come back to the office, carrying clipboards, pens, and papers. Before fieldwork and enter it into the OpenStreetMap database? And how begins, some preparation must be done. Mappers need will they do this efficiently and accurately? to know where they are going, how to orient themselves once they get there, and how to use existing features on Unfortunately, there is no single recipe for a perfect data the map to correctly position new ones. capture strategy. There are too many variables: the area being mapped, the data model, the available resources, For orientation purposes, you could supply mapping and so on. Still, we can suggest some methods that have teams with aerial imagery printed in color. This can be been successful in the past. a little costly to produce, but especially in dense areas, the imagery can be a great help in locating places on the ground. Alternatively, printed maps will also help teams 49 planning an open cities mapping project to orient themselves, particularly if the existing data maps) to cover an area that you estimate will take about include street names and significant landmarks. four hours to survey. You also might want to give them papers for an adjacent area of similar size, which they Once the mappers are oriented in the field—understand- can start working on if the first area takes less than four ing where they are and able to match their locations to hours. positions on their printed imagery and base map—they are ready to start collecting new features. Typically this As figure 4.1 shows, you can subdivide your target area is done with a combination of Field Papers and survey in two ways: by creating a grid or by defining areas with forms. (As section 2.4 discussed earlier, Field Papers are natural boundaries. printed OSM maps that the surveyors can sketch on to draw new features on the map.) In terms of preparation time, is easier to use a grid system because you can simply print out a large grid of Field Papers, and assign one paper to each mapping 4.1.1.1 Tracing Imagery team each day. On the other hand, from a surveying Tracing lines and polygons on top of imagery is a great perspective, it makes much more sense to follow existing way to produce base data before going into the field. boundaries such as roads and map everything within Because it is usually easy to identify roads on top of a given block. Organizing areas this way may require imagery, the tracing will make the roads available on more preparation but will be more efficient in the long your printed base map. run. Imagine you are collecting attributes about buildings. By It is a good idea to divide the target area early in the having your mappers draw the buildings before they go project. If possible, do this before fieldwork begins, into the field, they can print out Field Papers that already though you will need to have existing road data. Other- have all of the building polygons on it. They can then wise, you might set aside some time in the first few days easily collect attributes by enumerating each building for survey teams to simply map all of the roads, which on the Field Paper and filling out a survey form marked will then make it easier to divide the total area into daily with the same number. mapping areas. Once you do this, you will know how many of these 4.1.1.2 Producing Reference Maps and Field “blocks” are in your target area, and you can estimate Papers exactly the time and staff you will need to survey it all. In any project, someone should be responsible for Let’s say, for example, that you have a target area that producing reference maps and Field Papers. Managers you then subdivide into 20 blocks, a partial example of might either perform this task or delegate it to each which is shown in figure 4.2. You then estimate that mapping team individually. If possible, it is advisable for each block will take about four hours to survey and a managers to do it, because they will be keeping track couple of hours to edit, which means it will take one of the areas that each team maps and can ensure that pair of mappers to map one block every day. If you have teams are always equipped with the appropriate papers. one mapping team, it will take 20 workdays to map the Creating and printing Field Papers can take some time entire target area; if you have two mapping teams, it will each day, so that time should be considered while sched- take 10 workdays, and so on. uling activities. How would you manage this workflow? Each morning, you would give each team a block to map. You print 4.1.2 Defining Daily Mapping Areas out Field Papers that cover the areas they are supposed Each day, you will send mapping teams to different to map and draw a boundary around the specific area parts of the target area. Thus, one of the first tasks will or explain it to them directly. Give them additional be to determine what those smaller units of the area forms (survey forms, if you are using them) and the look like. For example, let’s assume that each pair of same materials for an adjacent square where they can mappers will go out for field surveys for four hours each map if they have extra time. (For this reason, as box 4.1 day. This means that you will want to provide them with explains, teams should receive block assignments that the resources (Field Papers, survey forms, and reference are not adjacent to each other on any given day.) 50 chapter 4 | implementation and supervision Figure 4.1 Options for Subdividing a Daily Mapping Area a. Gridding an area b. Subdividing an area by natural boundaries, such as blocks bounded by roads Source: Imagery in panel “a” from Field Papers (http://www.fieldpapers.org). In the afternoon, the teams will return to the office and DD (Optional) Printed map showing satellite imagery in begin editing. The manager will be there to answer color for orientation questions about surveying and editing and to ensure that everything gets uploaded to OSM without problems. With this survey methodology, the idea is to draw The next day, the process will repeat, thus ensuring that features on the Field Paper and mark them with identi- the target area is mapped in the expected time. fying numbers. Then, on a separate sheet of paper, the mappers can write down the attributes of each feature. In the case of detailed surveys, such as a building 4.1.3 Collecting Field Data survey, you may include a form that the mappers can Let’s take a closer look at the first half of the mapping fill out for each building to ensure that they collect all job: field data collection, or field surveys. We will assume attributes in an orderly fashion. that the workday is split in two, the first half for field surveys and the second half for editing. This schedule, of course, may differ based on the needs of the project. 4.1.4 Developing Survey Forms Mapping teams should receive sufficient survey forms Each mapping team will need to have the necessary each day. A good survey form should be organized, mapping materials each day as well as direction about simple, and clear. As shown in figure 4.3, the form where to map. Exactly which materials each mapping should be filled in with the surveyor’s name, the date, team has will vary based on the resources available an identification (ID) number for the part of the target and methodology of the project. At a minimum, each area they are mapping, an ID number for the Field Paper mapping team will need the following supplies: corresponding to the form, and an ID number for the specific feature being mapped. DD Field Paper of the day’s area DD Survey forms (if necessary) for collecting attributes The remainder of the form should reflect all of the attributes being collected as part of your data model DD Clipboard to keep papers and write on (discussed earlier in section 2.2.3). In other words, for DD Pen every attribute in the data model, the survey form should 51 planning an open cities mapping project Figure 4.2 Example of a “Blocked” Target Area in Dhaka, Bangladesh Source: OpenStreetMap. have a field to fill in. For attributes that can only have 4.1.5 Entering the Data (Editing) values within a certain range (for example, from “poor” The second component of mapping work is data entry. to “good”), it is a good idea to have checkboxes or mul- In a typical scenario, teams will go out mapping in tiple-choice options on the form.Survey forms should the morning, return to an office space, and edit in the be covered extensively in the second part of your project afternoon. This is usually a smooth process, though if training, which covers the identification of features. you don’t have the space or equipment for every mapping Mappers will need practice to identify all the different team to sit and edit their work at once, you may need to attributes covered in the data model and to understand come up with creative solutions, such as having half of how to properly fill out the survey forms. the teams map in the afternoon and edit in the morning. 52 chapter 4 | implementation and supervision Box 4.1 Mapping Nonadjacent Blocks It is recommended that adjacent blocks are not given to different teams to map on the same day. To do so would increase the likelihood of editing conflicts, which occur when two teams try to edit the same features at the same time. The assignment of nonadjacent blocks to each team also allows your mappers to continue mapping if their block takes less time than they expect. For these reasons, it is better to organize daily mapping activities following a checkerboard pattern, as shown in figure B4.1.1. Figure B4.1.1 Sample Mapping of Nonadjacent Squares in Guagua, Pampanga, Philippines Source: Data from OpenStreetMap It is best to do data entry on the same day as the data though this division has been done with some success in collection, or the following day at the latest. Although the past. mappers will be sketching out the features they collect and filling in data forms during data collection, they Conducting data entry directly after collection and often remember a lot of information simply from having ensuring that the mappers take part in the process will observed an area, and that information is best preserved help maintain an efficient and accurate flow of infor- if they edit the map soon after surveying. (See box 4.2 mation from the fieldwork into the OSM database. Be regarding custom presets, which can assist the mappers sure to keep the mapping forms and Field Papers well in adding features to their maps.) For the same reason, it organized and filed, however, in case you need to return is best for mappers to be involved in the editing process to them later when you review the data. rather than splitting up the jobs of editor and mapper, 53 planning an open cities mapping project Figure 4.3 Sample Building Survey Form Source: Open Cities Dhaka Project. Note: To view the full survey form, see Appendix 6.2. 54 chapter 4 | implementation and supervision 4.1.6 Scheduling Project Tasks Scheduling mapping activities is important for com- pleting the daily work, but care should also be taken The implementation phase will benefit from a clear to schedule appropriate time for preparing papers and project schedule. During project design, you should logistics for mappers. For example, if a manager is create a sample project timeline (as previously shown in overseeing four mapping teams, a typical day might look section 2.2.5). As the components of the project become like this: more well-defined, schedules should include more detail. DD 8–9 a.m. Assign each mapping team to an unclaimed block in the target area. Create and print Field Papers Of all of the tasks that go into a mapping project, for each team. Get survey forms and put all paper- several of the seemingly simple ones can be the most work into the mapping teams’ binders or clipboards. troublesome. For example, if you plan to import data, this becomes a critically important task because survey DD 9 a.m. Meet the mapping teams. Explain to each team work cannot commence until this is completed. If you where their area is and how to get there. Hand out schedule two weeks for data importing and it then takes paperwork and send teams to the field. four, it will delay other tasks and thus the whole project. DD 9 a.m.–noon Accompany one of the mapping teams to Similarly critical tasks include finalizing the data model the field to assist them and assess their work. and setting up alternative imagery sources, if applicable. Address these tasks right from the start and allow ample DD Noon–1 p.m. Lunch time to complete them. DD 1–4 p.m. Meet mapping teams back at the office and set them up at computers for editing. Help the Managers must also make clear schedules for them- teams with any questions they have about the editing selves and other mappers during the implementation process. Especially early in a project, review each phase. This is when the project will greatly benefit from team’s edits before they upload them to OpenStreet- the care taken in the project design phase to clearly Map. Make a list of common mistakes and explain define the target area and the data model. Managers them to everybody using a projector. should divide the project area into daily mapping blocks, enumerate them, and assign them each day to mapping DD 4–5 p.m. Organize the paperwork from the day and teams. The size of the blocks may change: during the begin preparing for the next day. File the completed first weeks of a pilot project, it will become clearer how survey forms and Field Papers for later reference. much area a single team can reasonably cover in a single Mark the completed areas on the project map. Meet day. Once this is understood, it will be easy to calculate with other managers to discuss any outstanding how many teams and how many days will be required to issues. complete the mapping activities. Box 4.2 Custom Presets During editing, your mappers will most likely be using either the offline JOSM software or the online iD editor. Because you will probably use a custom data model, you will want to add a cus- tomized menu to the editing software for adding the specific features that your mappers are collect- ing. This is done by creating a custom presets file, which implements a menu that makes it easy to attach the correct OSM tags to features. You can read more about creating custom presets here at http://learnosm.org/en/editing/creat- ing-presets/. HOT has created an easy-to-use visual editor for creating presets files, available at http://visualtags.hotosm.org/. 55 planning an open cities mapping project 4.2 COMMON CHALLENGES 4.2.3 Survey Fatigue There are a few common threats that many mapping Mappers who survey every day may tire of it over time. projects endure. It’s best to identify these potential It is physically demanding work and can sometimes feel threats early and develop backup plans in case some- monotonous. Generally mappers do better work if they thing goes wrong. maintain their enthusiasm. Finding ways to keep them motivated can aid a great deal—for example, by varying the work on occasion, providing frequent trainings to 4.2.1 Bottlenecks learn new things, and offering incentives for good work. Among the great threats to a mapping project are unexpected bottlenecks that result in the loss of much time and work. For example, if you are importing data 4.3 QUALITY CONTROL from a government agency, but you can’t begin mapping One of the manager’s primary jobs is to ensure quality until the data have been imported, make sure that you control of the data being collected and entered into allot enough time in advance to acquire the data, obtain OpenStreetMap. Quality control involves three principal permission to use it, and import it. tasks: If there is a weeklong religious holiday a month into DD Daily data checks your project and nobody will be working, you should identify that from the beginning. If the Internet con- DD Resurveying nection tends to go off every second day, you may need DD Data analysis to find another way of connecting your staff. Other common bottlenecks are the availability of hardware (such as having enough computers for your mapping 4.3.1 Daily Data Checks teams) or the ability to print enough Field Papers and The most immediate way to ensure quality is by staying survey forms to keep up with the mappers’ needs. in close contact with the mapping teams, keeping track of their edits, and discussing questions and concerns 4.2.2 Time Management with mappers when they arise. A great deal of time is lost in places where you would not New edits and additions should be reviewed and val- expect. This adds up and can significantly slow down idated daily or at most weekly. This is an important a project. One common loss of time is during morning supervisory task because catching mistakes and bad meetings. Out of 10 people, 1 or 2 are always late. editing practices early means that they can be corrected Don’t let their tardiness stop the rest of the group from and the editors can learn to do things properly. Manag- working, and try to enforce a policy of punctuality. Time ers should be highly proficient in the JOSM software, is also lost in the logistics—for example, printing Field using the validation and search tools to examine project Papers, preparing computers, and using the Internet. data. (Box 4.3 discusses some of these tools.) Key aspects Keep these processes running smoothly to save time and to review are energy. DD Topology errors (such as overlapping buildings or In addition, transportation can drain staff time unex- incorrect relations); pectedly. Be sure to build transportation time into the staff schedules. You can’t expect a team to map for a DD Tagging errors (misspelled tags or misused key-value full four hours if they need an hour to get to and from combinations); and their mapping area. Consider faster ways for them to get DD Completeness of data (whether all attributes in the there. For example, mapping teams may be allowed to data model have been collected). take survey equipment home with them and then head directly to the survey sites in the morning. This option, Managers should always be available during editing of course, requires trust that they will begin working on hours to answer questions and to check for mistakes. time. Each day they should review their mappers’ work and discuss any errors with them. They should also review 56 chapter 4 | implementation and supervision common mistakes using the projector so that everybody Resurveying has three main functions: may learn and benefit. DD It gives your mapping teams the opportunity to check and correct mistakes in their work. 4.3.2 Resurveying DD It allows managers to find recurring errors that Another method of quality control is to conduct second- mapping teams make and discuss them with the ary surveys at regular intervals. One day every week or teams to improve their work in the future. two, 5–10 percent of the completed work should be resur- veyed. Survey teams can resurvey each other’s work, DD It provides an important metric for project report- or, if possible, their more-experienced managers can ing: the percentage of error. A margin of error of 5 undertake the reviews. percent or less will probably be acceptable, although, of course, you should strive for the greatest accuracy The areas selected for resurveying should be randomly possible. selected from different areas, although at least one sample area should be taken from every mapping team throughout the course of the project. The resurveying 4.3.3 Data Analysis process involves collecting the same data twice, com- Just as resurveying helps to improve the data collec- paring those sets of data, and looking for mistakes. tion process, analyzing the data in JOSM improves For example, if you are mapping buildings and 1,000 the editing process. Another way to analyze the data buildings have been mapped, a resurvey would cover 50 is through GIS analysis, which may benefit both data to 100 of those buildings again. Based on the percentage collection and editing. of error you find, you can extrapolate the likely percent- age of error in the entire survey. It’s a good idea to regularly (once or twice a month) review the project data in GIS software such as Box 4.3 OSM Quality Assurance Tools Many tools can help OSM users check for errors in the data. The JOSM validation tool allows you to run an analysis and check for errors before uploading changes. An online tool called KeepRight (keepright.at) provides similar analysis and can help you find and correct mistakes. For a complete, up-to-date list of OSM quality assurance tools, visit wiki.openstreetmap.org/wiki/Quality_assurance. Quantum GIS. Such software allows for advanced query- 4.4 REPORTING ing and analysis of the data in order to answer question such as these: Most mapping projects have a financial backer and part- ners who will be interested in project reports. Depend- DD What features are missing attribute tags? ing on the length of the project, the reports may be DD Where do the values look out of the ordinary? issued at various intervals, and they will most certainly include a final project report. DD Where are names misspelled? How often should reports be submitted? Of course, this More detailed information on quality control and data will depend on the donors’ requirements, but for projects analysis can be found in HOT’s “Reviewing OSM Data3” of six months or less, a good rule of thumb is to deliver document. For more about quality reviews conducted a midterm report and a final report, as well as brief after project completion, refer to the “Scientific commu- snapshots of mapping progress every two to four weeks. nities” entry in table 2.1 on partnership sources. 57 planning an open cities mapping project For longer projects, you may want to deliver midterm GIS queries can help you to provide good metrics. These reports quarterly. queries can answer questions such as these: The midterm and final reports should include extensive DD What percentage of the target area has been mapped? evaluation of the project to date. They should summarize DD How many buildings have been mapped? the project activities and outcomes, including partner- ships, mapping activities, and the data collection prog- DD What is the average number of buildings a mapping ress. Lessons learned and changes to the original project team completes each day? design should also be detailed. In addition, these reports DD What is the estimated time to completion? are the appropriate place to document the results of your quality control processes, particularly the estimated level For a sample project report, see Appendix 6.1. of data accuracy and your error-checking methodology. The more–frequent interim “snapshots” need not be as detailed. The purpose of these regular updates is to report on mapping progress and maintain metrics on the pace of data collection. For example, if the project primarily involves building data collection, you will want to report the number of buildings mapped each week, the anticipated time to completion, the rate of mapping per day, and so forth. This will help to assess how your data collection process evolves over time. 58 chapter 4 | implementation and supervision Box 4.4 Project Implementation Checklist During the project implementation phase, the following tasks should be completed: Collecting the Data rrCreate and print reference maps rrDivide target area into daily mapping blocks; enumerate and make a statistics spreadsheet, providing each block its own row rrCreate and print survey forms rrCreate and print survey training manual rrMake daily work schedule rrCreate detailed project implementation schedule Quality Control rrDetermine methods to monitor data quality, and include these in both your project schedule and daily schedule Reporting rrDetermine the metrics to include in project reports rrAdd tasks to the schedule to ensure that metrics are collected and documented regularly rrMaintain statistics each day in a spreadsheet 59 60 5 LESSONS LEARNED AND RECOMMENDATIONS The preceding sections took a practical look at how to plan and implement a mapping project. Open Cities is ongoing, and improved methodologies and knowledge come out of each successive project. In this chapter, we conclude by highlighting several key lessons learned from the first year of the design and implementation of previous Open Cities initiatives. 5.1 GOVERNMENT OWNERSHIP 5.2 PARTNERSHIPS WITH Although many partners and participants in an Open UNIVERSITIES Cities project will be from civil society and the private Universities have been valuable allies during the first sector, it is critical that government counterparts be year of Open Cities work. Outreach to university depart- involved in a project’s development and execution. Gov- ments in engineering, geography, computer science, and ernments are primary stakeholders for many disaster planning provided critical connections and support to risk management and urban planning projects, and they Open Cities projects. Students from universities’ tech- provide necessary legitimacy to Open Cities work. nical departments frequently have the proper technical In Kathmandu, the involvement of the Department of backgrounds to quickly learn OpenStreetMap. In both Education helped to build their confidence in using Dhaka and Kathmandu, university students have played the data to prioritize seismic retrofitting activities. The important roles in mapping activities and software mapping team carried an official letter in support of the development. Some universities also require students to project, which was also critical to their ability to gain complete internships or volunteer projects before gradua- access to schools and health facilities to conduct surveys. tion, which led a number of students in Kathmandu to In Sri Lanka, local government authorities were directly participate in Open Cities. involved in the mapping activities to ensure government University faculty members have also provided useful ownership and use of the data. Key national agencies support. In Dhaka, professors from the civil engineering were regularly briefed and consulted. and planning departments at Bangladesh University of Involving government in this way also makes it easier to Engineering and Technology (BUET) provided necessary adopt a systematic approach to mapping, and it makes input into the design of the mapping project. Profes- it easier to obtain existing data sources for mappers in sors in Kathmandu University’s geomatics department areas that may not have many other partners to engage. have provided guidance to the project on quality control Engaging governments early in the planning process techniques for surveying and have incorporated Open- and ensuring close involvement throughout is an essen- StreetMap into their courses. Training future classes of tial component of a successful Open Cities project. university students will help the OSM community in 61 planning an open cities mapping project Kathmandu continue to grow after the formal project experienced OSM contributors who have provided signif- period has ended. icant support to the growing community in Kathmandu. In addition, the Kathmandu community helped digitize information into OSM after the Typhoon Yolanda hit the 5.3 ACCESS TO IMAGERY Philippines in November 2013. These relationships are an important factor in OSM’s global success, and project As shown in the OpenStreetMap community’s designers should seek opportunities to support them post-earthquake work in Haiti, access to high-resolu- through Open Cities work. tion satellite imagery is extremely useful for efficient mapping of infrastructure. However, such imagery is often prohibitively expensive or only available under licenses that would prevent digitization by the public. 5.5 BUILDING TRUST IN THE DATA Data quality is a frequently raised issue in community With this in mind, the U.S. Department of State’s and volunteer mapping projects. The Open Cities Project Humanitarian Information Unit (HIU) launched an has taken numerous measures to ensure that partners initiative in 2012 called “Imagery to the Crowd.” This and intended users of the data would trust its accuracy program makes high-resolution imagery— purchased and completeness. In Kathmandu, partner organiza- by the U.S. government from providers such as Digital tions including the National Society for Earthquake Globe—accessible to humanitarian organizations and Technology (a respected NGO working on seismic the volunteer communities that support them. Open resilience) and the Kathmandu University Geomatics Cities Kathmandu partnered with USAID and Imagery Department provided technical guidance to the project to the Crowd to release 2012 satellite photography for as well as independent quality assessments throughout the Kathmandu Valley and to organize volunteers in the process. In Dhaka, key stakeholders including BUET Germany, Nepal, the United Kingdom, and the United and representatives of government and civil society were States to digitize building footprints. The data created consulted throughout the project, and many of them by these volunteers have been incorporated into USAID received basic trainings in OpenStreetMap to familiarize disaster response planning, providing a solid foundation them with the platform. of data upon which the Nepali OSM community can continue to expand and improve. 5.6 SUSTAINED ENGAGEMENT 5.4 COMMUNITY CONNECTIONS Finally, it is important to note that, for these projects to be successful, sustained engagement with local partners OpenStreetMap and many of the other tools discussed in is necessary. Too often, technology and data projects of this toolkit have global networks of users, software devel- this sort are discrete, short-term endeavors. A workshop opers, and knowledgeable experts who interact and share or a weeklong training course is simply not enough time best practices through e-mail lists, in-person conferences to effect the kinds of shifts that Open Cities hopes to and meetups, and other forums. Connecting partici- support. pants in Open Cities projects to these communities can be a rewarding experience for them and provide valuable Although OpenStreetMap makes mapping more access to a global knowledge base. accessible to nonspecialists, it doesn’t change the fact that collecting and using geographic information is As part of the Open Cities Kathmandu project, the local a complex technical undertaking that requires more team helped to organize a number of events commem- training and involves a longer learning process than orating International Open Data Day in February 2013. many people assume. Building technical communities The day’s activities included a “mapathon” in which of OSM mappers and software developers who are teams worked together to digitize building footprints in familiar enough with the platform to comfortably deploy the Kathmandu Valley. Local groups were connected to it in their own tools and applications also takes time, but international volunteers, and, together, all participants such community building is integral to the sustainabil- mapped over 8,000 buildings in a single day. The local ity of Open Cities projects. team also established important relationships with 62 chapter 5 | lessons learned and recommendations Finally, Open Cities seeks to contribute to cultural and policy shifts within technical groups and government bodies that prioritize open data and broad participation in development challenges. When planning these kinds of projects, it is important that the parties involved understand and commit to sustained investment in their success. 63 CASE STUDIES case studies CASE STUDY BATTICALOA, SRI LANKA Overview Batticaloa, a major city in Sri Lanka’s Eastern Province severely affected by the Sri Lankan civil war and the 2004 Indian Ocean tsunami, is located in a hazard-prone area that has suffered near-annual droughts, floods and cyclones. Some limited hazard maps were available for the area, but no detailed digital geographic data of the built environment were available for use in risk studies or for informing contingency and response planning, potential infrastructure and risk mitigation projects. Scope of Work The objective of the Open Cities project was to map the complete building stock including critical assets and road infrastructures of the Manmunai North Divisional Secretariat, which covers an area of 68 km2 and includes about 90,000 people around the town of Batticaloa. The characteristics collected focused on basic information nec- essary for vulnerability assessment including number of floors, usage, and construc- tion materials of walls and roof. 67 planning an open cities mapping project Partners National Government — Several national government agencies including the Disas- ter Management Center, the Survey Department and the Census Department were involved from the beginning in the design of the project. Along the course of the project, the key national agency were updated on the progress and results of work. These partnerships provided the OpenCities with a strong country ownership, an important credibility, and a legitimacy to interact with the local government. The national agencies also provided staff from the local offices to support the project and build their capacity. Local Government — Locally, the work began with a series of meetings with the Batticaloa local authorities. In part, these were designed to establish the close collab- oration needed to carry out the actual mapping. They were also meant to ensure local understanding of and trust in the mapping process and in the data produced, so as to encourage local authorities to use the tools and data for their own DRM and urban planning projects. Humanitarian OpenstreetMap Team (HOT) — HOT provided some remote support to the team and facilitated a national workshop on OpenStreetMap at the end of the project to spread the word about OSM and build the capacity of key national agencies in order to start scaling up the project in Sri Lanka. Universities — The project was presented to most of the professors of the key depart- ments in the main Sri Lankan universities (University of Colombo, University of Peradeniya, Sri Lanka Institute of Technology, University of Moratuwa, University of Sri Jayewardenepura, Eastern University). Some of the professors participated in the national OSM workshop held by HOT. A small amount of mapping parties were held however as the location of Batticaloa is very far from most universities, the students were not involved directly in the data collection for this phase but will be involved in the next phase of the project. Building the Team A team of four technical experts (three recent GIS and IT graduates and one experi- enced GIS analyst) was hired and trained in OpenStreetMap techniques in order to supervise and support the overall mapping process. The team was completed by the staff from local partners, including the Batticaloa Municipal Council, the Batticaloa District, the Manmunai North Divisional Secretariat to constitute a team of about a dozen person fluent in OSM. For the field survey, the core team worked with the 48 recent university graduates that had been recently hired to work on the local planning and development of the 48 Grama Niladhari that make up the Manmunai North Divi- sional Secretariat. The graduates were appointed by the District Commissioner to work on the Open Cities project and received a stipend for the extra work. 68 case studies Workspace It was initially planned to host the team in the Manmunai North Divisional Secretariat premises but due to space, time and connectivity constraints, the team rented a house in Batticaloa that was used as a headquarter. It was set up with computers, internet connection, printers, tables, bedrooms in order to provide a space for trainings and to process field data. Existing Map Resources The team found an existing dataset for building footprints on the eastern coast. However, after examination, the quality was poor as it had been generated by auto- matic extraction so it was decided not to use it. Some datasets existed for infrastruc- tures like schools and roads but they were not cleared for use in OSM and outdated. Therefore, no pre-existing digital data was used in the project. However, the team worked with Divisional Secretariat and each GN divisions as they maintained paper maps for each of the 48 GN divisions. A booklet with all the paper maps were shared with the team and used to add landmarks and street names to the initial base map. Data Collection A small group began by tracing all building outlines into OSM using satellite imagery and then added landmarks, roads and road names, and points of interests using local paper maps provided by the divisional secretariat. This effort created a solid reference map for the surveying work. The work was then split into two components: buildings were surveyed by the 48 recent graduates, and surveyed data were entered by the core team who were also responsible for fixing the maps and refining the point of interests. Both groups were trained in OSM and surveying techniques by the Open Cities team. Quality Control Several measures were taken to ensure the data quality and ownership. First, the team worked with the local Building department, the Batticaloa Municipal Council civil engineer, and a civil engineering professor from the University of Moratuwa to create the building survey and ensure that it is usable by non-expert and taking into account local building practices. Second, the local team did ongoing quality assessment by doing a re-surveying of a sample of buildings for each GN divisions, making sure that the surveyors were not making mistakes. If some errors were found, corrective actions would be taken right away to ensure the rest of the data would be correct. Third, the team ran queries on the database to identify anomalies in the data collected and accordingly made fixes and took corrective actions through extra training. The trust and ownership in the data was ensured by the multiple partnerships with the national and local government and engaging through regular updates and feedback loops through the different phases of the project. 69 planning an open cities mapping project Results Footprints and basic characteristics including number of floors, usage, and construc- tion materials of walls and roof was collected for all 30,000 buildings in the area. These data are now freely available in OpenStreetMap and in the government geo- spatial data sharing platform RiskInfo (www.riskinfo.lk) for easy use by many stake- holders. At the local level, a wrap-up ceremony was held with all the stakeholders and participants where A3 maps were given to each GN division and T-shirt were handed out to all the participants. Future plans were also discussed including the request to extend the project to other DS Divisions in the district and to link the household survey recently conducted (150 questions per household) with the geographic data. To publicize the benefits of these techniques at the national level and promote their adoption, high-level managers of the relevant national agencies were briefed regularly and given final results when available. Two week-long training courses, one dealing with OSM techniques and the other with use of data for decision making (specifically the combination of data with existing hazard maps through GIS tools and the InaSAFE tool) were conducted at the national level with all relevant national agencies. Discus- sions are ongoing with various ministries concerning the next phase of the project. A second phase of the project will take place to cover the flood prone Attanagalu Oya river basin near Colombo. There is also strong interest in scaling up the project to cover a greater geographic area, in using OSM to update the Survey Department data, and in streamlining the use of the data in more DRM applications and sectors. 70 case studies CASE STUDY KATHMANDU, NEPAL Overview Kathmandu, the capital city of Nepal, is the most seismically at risk city in the world in terms of potential for loss of human life during a major event. In November 2012, in partnership with the government of Nepal, the World Bank and GFDRR launched a project to build seismic resilience in the Kathmandu Valley’s education and health infrastructure, in part by creating a disaster risk model to determine the relative vul- nerability of the relevant buildings. Once complete, the model will be used to prioritize plans for retrofits of schools and health facilities to improve structural integrity in the face of earthquake. Scope of Work The mapping goal for Open Cities Kathmandu was to collect data on location, occu- pancy, and structural characteristics for all of the schools and health facilities in the Kathmandu Valley. Partners Government — The Nepal Department of Education is a key partner in Open Cities Kathmandu. They are leading the risk assessment project to which the mapping work 71 planning an open cities mapping project will contribute. In addition to institutional support, the DOE also provided a letter of support that mappers carried with them as part of the work, instrumental to gaining access to some of the facilities. Technical Agencies — NSET, the National Society for Earthquake Technology, is a Kathmandu-based NGO with significant seismic risk and engineering expertise. They supported the project through assistance with data modeling, providing training on structural assessments to the mapping team, and conducting quality assessment of the data throughout the process. Universities — Universities were critical partners in the project. Connections to the Geomatics and Engineering departments in Kathmandu University and Tribhuvan University assisted with finding the core team and volunteers for the project and helped provide technical guidance and quality control during key project phases. Building the Team The core team comprised six graduates of Kathmandu University who were recruited based on their prior contribution to Nepal’s then-nascent OSM community. They were paid full-time salaries at rates commensurate with the local salary structure for recent graduates in technical disciplines. The project also recruited six part-time interns from Kathmandu University and 11 volunteers from Tribhuvan University. Workspace The project identified a local co-working space and technology startup incubator that provided access to meeting rooms, reliable Internet service, and opportunities to inter- act with other technologists and entrepreneurs, some of whom later became active in the local OpenStreetMap community as a result of the team’s outreach efforts. Existing Map Resources The mapping team obtained a spreadsheet containing an incomplete list of 1701 schools in the Kathmandu Valley from the Department of Education (DOE), which was then used as a starting point for the survey. The list did not provide the exact location of schools, just the wards in which they were located. For health facilities, Open Cities Kathmandu used three sources to generate a reference list of hospitals, health posts, and polyclinics: a list of 100 private and community hospitals received unofficially from the Ministry of Health and Population (MOHP), a list of 205 medical centers approved by MOHP that is accessible through the official website of the MOHP, and the names of 70 hospitals and 27 polyclinics from a private-sector health organization in Kathmandu. Together, these provided a list of over 400 unique health service pro- viders that the team worked to locate and survey using a similar approach. Data Collection The mapping teams used a snowball approach to find locations of the schools. They began with what location data were available through other sources, or by using their 72 case studies personal knowledge and social networks. They then talked with locals, school officials and health facilities personnel to locate remaining schools as well as ascertain the existence of schools not on their list. As a result of requests from school officials, the team also began carrying and distributing pamphlets and posters that provided basic information on earthquake preparedness and safety. Quality Control The team took a number of steps to ensure that data collected during the project were accurate and that they would be trusted by the wider group of stakeholders involved in disaster risk reduction in Nepal. First, the project was carried out with explicit support of Nepal’s Department of Education and the World Bank, two institutions with credi- ble reputations in the country. Second, the team consulted frequently on engineering matters with The National Society for Earthquake Technologies (NSET), a well-rec- ognized entity in disaster community in Nepal. NSET Engineers helped design the survey and provided training. The project also provided OSM training to a wide range of individuals from government, donor, and NGO community to build awareness of and comfort with the platform and approach. Finally, NSET conducted quality assess- ment of OSM data regularly during the surveying process. Through the involvement of NSET in Open Cities Kathmandu, the team regularly made quality assessments of the survey to both identify problems as they occurred and also to be able to assess the data collected in full at the project’s completion. Of particu- lar concern was the building structure surveying, which required training from NSET for the mapping team and volunteers to survey correctly. The first field verification was performed immediately after the first training and some initial in-field practice. The accuracy of the data for building structural system was 50% during this period. Given this result, NSET provided a “refreshers” training to the surveyors. Following this second training, the accuracy of the structural data increased collected by the mappers rose to 100%. Results Open Cities Kathmandu surveyed 2,256 schools and 350 health facilities in the Kath- mandu Valley. In addition to collecting a comprehensive list of structural data for health and school facilities, the team worked to create a comprehensive base map of the valley by digitizing building footprints, mapping the road network, and collecting information on other major points of interest. The Open Cities team also conducted significant outreach to universities, technical communities, and government in order to expand the OSM community. Over 2,300 individuals participated in OSM trainings or presentations during the first year of the project. The data have been used in plans to retrofit school and health facilities and in applications for transportation planning; moreover, USAID has incorporated the data into disaster preparedness planning exercises. That the American Red Cross has begun to contribute to OpenStreetMap in Kathmandu suggests that development partners remain interested in the project. A local NGO called the Kathmandu Living Labs (KLL), staffed by participants in the first 73 planning an open cities mapping project phase of the Open Cities project, has been created in order to continue the work. KLL has recently received funding from ICIMOD, a local technical organization, and the US Embassy in order to conduct OSM trainings and mapping activities. 74 case studies CASE STUDY DHAKA, BANGLADESH Overview Dhaka’s Old City is a crowded and complex area of immense historical value and an important locus of social and economic activity. In consultation with Dhaka Water and Sanitation, seismic risk experts from Bangladesh University of Engineering and Technology (BUET), and the Urban Study Group, a local nongovernmental organiza- tion (NGO) working on heritage preservation and restoration in Old Dhaka, the Dhaka Open Cities pilot sought to create detailed maps of three of the Old City’s 15 wards. These maps would provide data useful for planning of evacuation routes, managing water and sanitation infrastructure, and understanding the location and characteristics of heritage buildings. 75 planning an open cities mapping project Partners Dhaka Water and Sanitation Department Dhaka WASA was the primary government partner for this project and provided initial data describing water and sanitation infrastructure in Old Dhaka as well as input into the data modeling process. BUET University Professors from the Departments of Planning and Civil Engineering provided techni- cal guidance, assistance with recruitment of project participants, and workspace to the project. Urban Study Group The Urban Study Group provided data on location of critical heritage sites in Old Dhaka as well as input into the data modeling process. The Center for Environmental and Geographic Information Services (CEGIS) CEGIS is a local NGO with extensive experience working on GIS and surveying in Bangladesh. They provided overall management of the mapping work and quality assessment of the data collected. Humanitarian OpenStreetMap Team (HOT) HOT provided initial OSM training to project participants and assisted with the devel- opment of reporting and project tracking processes. Building the Team In partnership with BUET, which provided technical support and a working space, 20 engineering and planning undergraduates were hired as mappers and were trained for a three-month period. A local nonprofit GIS consulting organization, CEGIS, was contracted with to provide management and quality control for the work. The Human- itarian OpenStreetMap Team, a nonprofit specializing in the use of OpenStreetMap in development and humanitarian relief situations, also provided training and technical oversight to the project. Workspace BUET University provided a computer lab to serve as the primary workspace for train- ing of project participants and data entry work. Existing Map Resources The effort began by importing building footprint data for the three wards—created by CEGIS as part of a different project but until that point unavailable to the 76 case studies public—directly into OpenStreetMap. This allowed the mapping team to focus on field surveying, in which basic characteristics, such as building height, usage, construction materials, and age were collected through visual survey of each building. Data Collection Mapping teams divided the three wards into so-called blocks - sections which were bounded by roads on all sides. Each day a team would be assigned one or more of these blocks to survey, more easily mapping the densely set buildings due to the small compartmentalized blocks which they were sent to map. Imagery was less useful than normal due to the building density, but this methodology helped ensure accurate data collection. They also mapped road characteristics (width and surface type) along with important water and sanitation infrastructure. The data were added to OpenStreetMap during times when conditions prohibited field surveys (e.g., poor weather conditions, strikes). Quality Control The mappers were divided into small groups of surveying teams, each led by a team leader, who was responsible for organizing and planning their team’s work and ensur- ing smooth and accurate data entry. The team leaders were also responsible for imple- menting quality control measures. Twice each month they would randomly select ten percent of their mappers’ buildings and resurvey them. By doing so they could calculate the likely percentage of error of the entire survey, correct a limited number of mistakes, and discover repeated mistakes in the surveyors’ workflow. Results During this period, the team was able to finish complete maps of the three wards. In total, 8,500 buildings, 540 of which were deemed to have historical significance, were surveyed. Sections of roads measuring 43km and drainage works measuring over 50km were also assessed. These data are now available to the public through the OpenStreetMap platform. Several training courses and presentations on OpenStreet- Map were also given to university students, government partners, and private sector technology companies during the project period in order to help the OSM community in Dhaka grow. The results of the pilot were presented to the government and other key stakeholders in December 2013. Consultations are ongoing concerning the next phase of the project. 77 78 6 APPENDIXES APPENDIX 6.1 DATA MODEL DESIGN Your data model is a chart of the features and attributes agree on the same tagging conventions to represent of those features that you plan to collect in the field. As features in order to maintain data uniformity. For described in chapter 3, designing the data model is a example, this is why nearly all tags are in the English process of determining the priorities of your project in language. For most attributes that you want to map, you coordination with your partners. can probably find an existing tagging convention on the OSM Wiki site. Also useful is taginfo.openstreetmap. After a data model has been defined, you must deter- org, which provides statistics about tags that are being mine how the features will be mapped using the Open- used in the OSM database. In some cases when you are StreetMap tagging system. In OSM, features are drawn mapping new features or attributes, it may be necessary as points, lines, or polygons, and tags are applied to these to invent new OSM tags. It is a good idea to discuss this features to describe their attributes. Common tags can with your local e-mail list and by sending a message to be seen on the OSM Wiki site, at wiki.openstreetmap. tagging@openstreetmap.org. org/wiki/Map_Features. The data model below (figure A6.1.1), designed for the OpenStreetMap uses a free-tagging system. That is, you Open Cities Dhaka pilot project, shows how attributes can use any tags that you want to describe attributes. were mapped as OpenStreetMap tags. However, it is important that users around the world Figure A6.1.1 Sample Data Model, Open Cities Dhaka Project Open Cities Dhaka OSM TAGGING SYSTEM 1. Streets key possible values highway {primary | secondary | tertiary | residential | living_street} name name of street surface {asphalt | concrete | brick | unpaved} width number in meters of street width oneway {yes | no} 79 planning an open cities mapping project 2. Buildings key possible values addr:housenumber address number of the building, ie 25/5 or 19A addr:street street name name name of building building {yes | construction} building:levels number of levels in the building building:use {residential | commercial | industrial | utility | multipurpose | hospital | clinic | place_of_worship | government | school | college | community_centre} building:vertical_irregularity {yes | no} building:soft_storey {yes | no} building:material {plaster | brick | tin | cement_block | glass | bamboo_sheet | wood} building:structure {RCC_with_beam | RCC_without_beam | brick | steel | timber | bamboo} start_date year of construction, or range, ie. 2003..2013 building:condition {poor | average | good} 80 chapter 6 | appendixes 3. Storm Water Drainage 4.2. Hand Pump man_made = water_well 3.1. Drainage lines (ways): pump = manual waterway = drain pump:active = {yes | no} drain:covered = {yes | no} 4.3. Overhead Tank 3.2. Manhole covers (nodes): man_made = water_tower manhole = {drain | sewer} 4.4. Reserve Tank 4. Public Water Source man_made = water_tank 4.1. Stand Point amenity=drinking_water 5. Public Toilets pump:active = {yes | no} amenity = toilets toilets:num_chambers = # of toilets Source: Open Cities Dhaka. Note: OSM = OpenStreetMap. 81 chapter 6 | appendixes APPENDIX 6.2 SAMPLE SURVEY FORM Figure 6.2.1 Building Survey Form, Open Cities Dhaka Project Open Cities Dhaka – Building Survey Form Surveyor Name: Date: Field Paper ID: Ward / Block number: Map Building ID: 1. General Information: 1.1 Holding #: Building name: Street name: 1.2 Building usage: Residential Hospital Commercial Religious Industrial Government building Utility School Mixed College Community Center Other specify: 83 planning an open cities mapping project 2. Building Characteristics: 2.1 Number of Stories: 2.2 Vertical Irregularities? Yes / No 2.3 Is there a Soft Storey? Yes / No ( Long / Short direction ) 2.4 Walls: (Cladding & Partitions) 2.5 Main Load Bearing System Plastered RCC column with Beam Exposed brick RCC column without Beam (flat plate) Tin Sheet Brick masonry (load bearing wall) Exposed Cement Block Steel Glass Timber Bamboo sheet Bamboo Plywood/wooden Other specify: Other specify: 2.6a If available, Construction Year: (otherwise complete 2.6b), 2.6b Age: a) <10 yrs b) 10-20 yrs c) 20-50 yrs d) 50-100 yrs e) >100 yrs f) Unknown 2.7 Visible Physical Condition (Crake, Damp, Spoiling): a) Poor b) Average c) Good Notes: 84 chapter 6 | appendixes APPENDIX 6.3 SAMPLE DATA AUTHORIZATION LETTER The following letter (figure 6.3.1) is adapted from the EUROSHA, a European Union pilot project in Burundi, the Central African Republic, Chad, and Kenya. For more examples, visit http://wiki.openstreetmap.org/wiki/Import/ GettingPermission. Figure 6.3.1 Sample Data License Request Letter We, the undersigned, ………………………………………………………………………… grant to the OpenStreetMap Foundation, in the conditions as stated below, a worldwide, royalty-free, non-exclusive, perpetual, irrevocable licence to do any act that is restricted by copyright, database right or any related right over anything within the Contents, whether in the original medium or any other. These rights explicitly include commercial use, and do not exclude any field of endeavour. These rights include, without limitation, the right to sub-license the work through multiple tiers of sub-licensees and to sue for any copyright violation directly connected with OSMF’s rights under these terms. To the extent allowable under applicable local laws and copyright conventions, we also waive and/or agree not to assert against OSMF or its licensees any moral rights that we may have in the Contents. We grant to OSMF that it may only use or sub-license our Contents as part of a database and only under the terms of one or more of the following licenses: ODbL 1.0 for the database and DbCL 1.0 for the individual contents of the database; CC-BY-SA 2.0; or such other free and open licence (for example, http://www.opendefinition.org/okd/) as may from time to time be chosen by a vote of the OSMF membership and approved by at least a 2/3 majority vote of active contributors. We consent OSMF to quote us or cite the owner of the copyright, according to our wishes or these of the copyright holder. The citation will be showed on the following web page http://wiki.openstreetmap.org/ wiki/Attribution. We offer the OSMF the following data: ………………………………………………………………………………………………… …………………………………………………………………………………………………………………………………………………………… …………………………………………………………………………………………………………………………………………………………… …………………………………………………………………………………………………………………………………………………………… ……………………………………………………… Drawn up at ………………………………, the …………………………… in duplicate. Signature of the donor: Signature of receiver: (Preceded by the handwritten words «read and approved») Source: EUROSHA Project. 87 chapter 6 | appendixes APPENDIX 6.4 SAMPLE PROJECT REPORT It is a good idea to provide snapshot progress reports every two to four weeks during a mapping project. These can be brief, 1- to 2-page documents that highlight any news or changes in project design, mapping progress, and metrics on data collection and quality control. The sample below suggests content to include. Open Cities Project Update - July 1, 2014 I. Project Updates A brief paragraph summarizing new project developments. For example, any unexpected changes in the project, either positive or negative, may be recorded here. In other words, any project news, such as meetings with key partners, mapping activities, and adjustments to the project design. DD bullet points DD are good DD for documenting progress DD simply II. Mapping Progress A simple map produced in GIS software can give a quick visual explanation of your progress. Buildings completed are overlaid on the target area. 89 planning an open cities mapping project III. Data Collection and Quality Control Metrics Metrics Week 1 Week 2 Week 3 Week 4 Week 5 Week 6… Total blocks in the focus area 100 100 100 100 100 100 Number of buildings mapped 500 600 800 … Total buildings mapped 500 1,100 1,900 … Number of blocks complete 5 10 16 … Number of survey teams working 6 8 8 … Average buildings mapped 83 75 100 … per team Number of blocks remaining 95 90 84 … Estimated weeks to completion 19 18 16 … Number of buildings resurveyed 0 50 75 … Number of resurveyed buildings 0 5 6 … with error Percentage of buildings 0 10 8 … with error 90 chapter 6 | appendixes APPENDIX 6.5 SAMPLE LETTER OF SUPPORT FOR MAPPERS Government of Nepal Ministry of Education Department of Education (Physical Service Section) Letter No: Last Name, First Name Ref No: Date: Subject: To provide necessary support Dear all educational institutions in Kathmandu Valley, The World Bank is undertaking an exposure survey of all educational institutions within Kathmandu Valley for earthquake risk analysis and reduction. With this letter, we ask that you provide necessary support to the surveyors Name Title CC District Education Office, Kathmandu/Lalitpur/Bhaktapur World Bank Nepal, Kathmandu Source: Open Cities Kathmandu. 93 chapter 6 | appendixes APPENDIX 6.6: OPEN CITIES PROJECT PLANNING CHECKLIST Here is a complete checklist you may wish to follow when planning a mapping project. It will help you to consider the many factors that go into developing an efficient, effective process. Partner Outreach rrIdentify relevant partners in government, academia and civil society rrMeet with potential partners, discussing support and areas of collaboration rrCommunicate project goals with OpenStreetMap community, and meet with local mappers Defining the Scope of Work rrObtain a map containing the boundary of the target area rrEstimate the number of features you will map in the area, such as roads, buildings, and so on. rrWrite out your data model rrMap the data model to OpenStreetMap tags rrEstablish a rough time frame for survey completion, noting any holiday periods that will halt the work. Building the Team rrEstablish management structure rrDetermine sources of your staff from and how staff will be compensated rrHire staff Assessing Resources rrList the existing digital and paper data rrDetermine whether you will import any data; if so, create an action plan and time frame rrDraw a map showing parts of your target area where no imagery is available rrDetermine whether you will use GPS devices, smartphones, or both; if so, acquire rrConfirm that the necessary base map data are available in order to use Field Papers 95 planning an open cities mapping project Logistics rrIdentify where your staff will work rrEnsure reliable Internet connectivity rrList the equipment you need and acquire it Training rrIdentify who will conduct the training and what material will be covered rrCollect necessary training materials, develop training manual rrIdentify a location to host the training (maybe the same as the project office) rrCreate a training schedule Collecting the Data rrCreate and print reference maps rrDivide target area into daily mapping blocks; enumerate and make a statistics spreadsheet, providing each block its own row rrCreate and print survey forms rrCreate and print survey training manual rrMake daily work schedule rrCreate detailed project implementation schedule Quality Control rrDetermine methods to monitor data quality, and include these in your project schedule and daily schedule Reporting rrDetermine the metrics to include in your project reports rrAdd tasks into your schedule to ensure that metrics are collected and documented regularly rrMaintain statistics each day in a spreadsheet 96