On April 25, 2015, Nepal was hit by a massive earthquake of 7.8 on the Richter scale. It sent shockwaves through the country’s 31 out of 75 districts, killing 9,000 people and injuring many more. In the wake of the quake, most of rural Nepal lay in ruins – requiring a nationally coordinated reconstruction effort. As a first step to reconstruction, the country needed to carry a comprehensive damage assessment to determine the level of damage of each house and identify beneficiaries eligible for government assistance. A preliminary estimation of around one million houses was calculated for the assessment. Compounded with complicated topography and limited accessibility to rural parts of Nepal, reaching those many houses and collecting damage assessment data were massive and challenging tasks.
In the old days, this would have meant sending enumerators to field with stacks of paper questionnaire, recording data in papers, bringing back heaps of filled up forms, spending months and years in transcribing them, before analyzing the data to formulate intervention plans. In this case, obviously, there was no time to wait that long to start reconstruction work. Keeping that in mind, the Government of Nepal decided to use digital technology for data collection. This is a historic decision and a radically different route. Kathmandu Living Labs was selected to design and develop overall technological solution for this challenge. Our key responsibilities were to: conceptualize, design and develop tablet App; train engineers on the App; test, configure and prepare tablet devices; provide helpdesk support to the field engineers; design and maintain backend database to ensure it is up-and-running; develop an MIS system to support the government for data validation process; and design and develop visualization system to track survey progress.
The government deployed over 2,500 engineers for this task. Instead of using paper forms, they used tablets to record data. The tablet also allowed them to capture location of buildings along with pictures which acted as evidence to support their damage assessment. The data was collected offline, but the engineers uploaded the collected data to a central server in Kathmandu using Wi-Fi or telecom network depending on the availability of the services. Using an electronic device to capture damage data meant better data quality, zero transcription cost, shorter turnaround period, enhanced data security, and richer dataset with GPS coordinates and photographs. One of the biggest advantages of using an electronic device was that the data could be transferred to the data center and used almost instantaneously.
During the first phase, we captured data of over 7,60,000 houses from the 11 most earthquake-affected districts of Nepal (almost 76% of total houses that were eventually surveyed) and made it available to National Reconstruction Authority within the first 100 days. In those districts, engineers went to every house as in the national census. Three districts in Kathmandu Valley and additional 17 districts in other parts of the country were surveyed in the second and third phases respectively. In these later two phases, engineers went only to houses that were reported to the government as damaged.
By the end of the survey in all 31 districts, the statistics read as below:
|No. of surveyors deployed||2,513|
|No. of records created||9,97,522|
|No. of houses surveyed||10,53,874|
|No. of households surveyed||10,37,525|
|No. of household members covered||50,81,967|
|No. of photographs captured||93,36,576|
|Total size of the dataset||9.39 terabytes|
To the best of our knowledge, this is the largest mobile survey ever in terms of numbers of surveyors deployed and volume of data collected in such a short period of time. Perhaps the biggest outcome of this project is the story of grit and inspiration that Nepal has to offer to the world – an outstanding example of solidarity in the face of a crisis – how the government, development partners, tech innovators and victims worked together to create data infrastructure as a foundation to building back better. You may also read this cover story in one of the nation’s leading newspapers.
Of course, achieving this milestone was no mean feat. This project encountered technical as well as non-technical challenges at several steps. A select-few of those challenges that we could manage to fit in the space below are:
Technical (digital survey) challenges:
- Bringing digital survey data with photographs from rural fields to the data center in Kathmandu on a daily basis using the infamously unstable (and in many cases, effectively absent) internet connectivity.
- Setting up, configuring and keeping the server infrastructure up-and-running 24/7 for the entire survey period. Initially, the plan was to use cloud-based service to store data. However, this decision was abruptly changed by the government over concerns of national information security after the first batches of engineers were deployed to the field.
- Making tablet batteries (power) last on the field. How do we charge them? There was no electricity in many rural places.
- Convincing all stakeholders to make momentous shift from paper survey to digital survey.
- Hiring, deploying, and managing 2,500 surveyors and 2,500 social mobilizers. Each surveyor was paired with a social mobilizer with local knowledge.
- Keeping all stakeholders on the same page. This project had at least seven agencies taking different roles at any given time.
- Procuring and managing survey logistics for 5,000 surveyors and social mobilizers.
Explaining how these problems were solved, or even approached, undoubtedly deserve separate blog posts of their own. A single blog like this cannot do justice to the innovation, out-of-the-box thinking, and the sheer hard work and determination of everyone involved in making this project a success.
However, a few accolades that could (read: should) be a part of any sort of writing on this project are: guidance by the National Reconstruction Authority, overall coordination by the Central Bureau of Statistics, the impeccable human resource (i.e. recruitment of those many engineers) provided by HERD and Real Solutions, and the overall project management services by the United Nations’ Office for Project Services (UNOPS). Special mention goes to UNOPS consultants, Dipesh Raj Sharma and Vikas for their relentless work on the server-side infrastructures. And of course, to The World Bank and DFID for funding, trusting and supporting in etching out a new digital roadmap in Nepal.
A special mention goes out to our friends at ONA, a Nairobi and Washington DC based social enterprise who helped setup, manage and support the server-side data collection software (Ona Platform) on local servers in Kathmandu. Despite time differences, they demonstrated true professionalism in responding to Nepal’s needs for local hosting and providing continuous tech support throughout the entire project.
One of the larger outputs of this project is the highly rich 10 terabytes data of over one million houses. The dataset not only contains damage data of the buildings, but also the location of each of those buildings along with immensely useful socio-economic profiles of its residents. In addition to identifying housing beneficiaries, this data has a potential to serve as an information infrastructure for development planning and interventions. As Dr. Budhathoki, Executive Director of Kathmandu Living Labs, who led the digital team of this project, was quoted saying, “… this is a gold mine of data. The government can overlay other information, e.g. roads, schools, businesses and communities on top of the existing data for informed decisions in future.”
Recently, the government has opened this data to the public. It organized an event at Central Bureau of Statistics premises on June 11, 2017 to announce this decision (read the announcement here). This has inspired open data communities in Nepal and around the world. Virtually, anyone can now get this entire dataset in digital form with a minimal fee (of course, in anonymized form to maintain individual privacy).
Many people were reluctant to use mobile technology to collect housing damage data when we started the project. They suggested we switch to the traditional pen and paper even after the surveyors were already in the field. It was a radical approach which they had never witnessed before. Hence, their hesitation is understandable. Now, within this short time, many government departments including Central Bureau of Statistics have already started using digital data collection method in their surveys. Further, during the event organized to make this housing damage data public, the joint secretary from National Planning Commission, Mr. Tulasi Prasad Gautam, enthusiastically encouraged Central Bureau of Statistics to explore possibilities of using mobile data collection during the 2021 census of Nepal.
One big outcome of this digital survey is the paradigm shift in data collection, data validation and data usage triggered by this experience. This experience has potentially transformed forever the way Nepal would now collect and use data (let us hope). We feel that this project has helped Nepal develop its confidence in digital technology. For those working in the field of civic technology, there is possibly nothing more satisfying than this!