Post-earthquake digital revolution in Nepal
In April 2015, a 7.8 Mg earthquake hit Nepal that affected over 1 million households and 5.08 million people. Disbursement of reconstruction funds called for a massive door-to-door survey in order to identify true beneficiaries. For the first time, Nepal and the Kathmandu Living Labs deployed a team of 3,000 engineers armed with electronic tablets to collect 10 TB data and 10 million photographs from remote parts of Nepal in just over 120 days.
Most of the places worst hit by the 2015 Nepal Earthquake were remote mountainous villages. Reaching the people living in these places and quickly assessing the damage was of paramount importance to the Government of Nepal (GoN). Considering the fragile economic conditions of most people in this region and the short time before the onset of monsoon rains followed quickly by winter meant that affected people needed to receive their earthquake assistance as soon as possible. However, assessing the damage and identifying the beneficiaries would be too time consuming if done through traditional method of pen and paper survey. Use of electronic media to capture, transfer and process the data seemed to be the only plausible way of fulfilling this massive task in a reasonable timeframe.
The biggest challenge of collecting and transferring 10 TB of data from the remote parts of Nepal was the unavailability of internet. Nepal is not known for its reliable internet connection. In addition, this was a time when the internet network might have been poorer than usual due to the damages to telecommunication towers by the earthquake and its aftershocks.
Even in places where the internet was available, the other challenge was that of limited bandwidth. In such cases, there needed to be a mechanism to send only textual data (of small data size) to the central server; while retaining the photographs (of large data size) on the surveying tablets. Once a reliable internet connection would become available to the surveyors they would then have to upload these photographs to the central server. In order to make sure that the correct photographs were attached to the already uploaded text data a system that would automatically match these data had to be developed. Failure to automate this would mean manually mapping 10 million photographs to 1 million households which would be an impossible task.
Third, since many families were spending their nights out in tents and Internally Displaced Person (IDP) camps, this survey was highly time-sensitive. It meant that the collected data had to be processed and the list of identified beneficiaries had to be published at the soonest. This required near real-time validation of the collected data in order to inform and expedite the subsequent repair, retrofitting and reconstruction efforts. On the other hand, it had to be ascertained that the surveyors actually travelled to the sites of houses for ‘real’ inspection. Hence, there had to be a way to ascertain in the survey data that they actually travelled to specific locations
In addition, since this project would involve a couple of 1,000 engineers on field at any given time, it required a robust progress tracking and visualization system in order to steer this project correctly and in time.
Therefore, the goal of this innovation was to beat these odds and devise an end-to-end ICT solution that would expedite the process of capturing survey data on field, transferring them over notoriously poor internet connections (or holding them safely on tablets for later uploads), processing the transferred data at the Central Data Center and identifying and publishing the list of beneficiaries entitled for earthquake assistance.
In hindsight, now that the project is successfully concluded (and dubbed as the world’s largest mobile survey executed given the timeframe), this innovation benefitted all that were involved. On the back of this innovation’s success, the Central Bureau of Statistics (CBS), the nation’s authority on statistics, is now planning to do the next census using tablets. The National Reconstruction Authority (NRA), Nepal’s top authority on post-earthquake reconstruction, was able to publish the list of beneficiaries in record time. Also, this intervention has been most beneficial to the earthquake displaced people who were able to receive reconstruction funds quickly and judiciously.
One of the best things about this survey dataset is that it contains a goldmine of socio-economic data in addition to data on earthquake damages. Hence, the CBS has released an anonymized dataset. In addition, the National Planning Commission (NPC), Nepal’s apex planning body, has commissioned and launched eq2015.npc.gov.np - an open data portal developed to increase the usability of this dataset to the wider audience.
What Makes Your Project Innovative?
Historically, all surveys led by the Government of Nepal prior to this one were pen-and-paper surveys. This was the first time that electronic tablets were being used to capture and transfer data at such a scale. Whereas traditional methods would have required additional time to capture, cleanse, transcribe and process data, use of electronic media curtailed this cycle by effectively removing transcribing process (and associated errors). It also made data cleansing easier by enforcing validation rules on the tablets, i.e. right at the source of data collection.
In addition, this system could capture data while remaining totally offline. It could not only capture photos but also compress them to save file storage space without compromising the visual quality of the images. Moreover, the system allows surveyors to split text data from photographs to be able to upload data using slower internet connections. None of these features existed in any single electronic survey tool at the time.
What is the current status of your innovation?
The survey is now complete in all 31 earthquake affected districts. It was done in three phases, each spanning an average of 40 days. A total of 10 TB data and 10 million photographs of 1 million households and 5.08 million people were captured in just over 120 days. Based on this data, the NRA published its first list of beneficiaries within days of starting the survey.
With the confidence gained from this survey, CBS is thinking about conducting its 2021 census using electronic tablets (at least in places where it is feasible). This is a wonderful example of the nation’s authority on statistics backing up new, digital technology. The NRA has also used the same system in order to collect and handle grievances related to the previous survey.
Collaborations & Partnerships
CBS managed the overall survey. Kathmandu Living Labs provided technical support to CBS on mobile app development and end-to-end ICT infrastructure management. UNOPS managed the overall logistics, HERD and Real Solutions hired and managed 3,000 engineers. The World Bank and DFID were the principal donors. Department of Urban Planning set out the damage assessment criteria and methods.
One of the biggest reasons for this intervention’s success was the smooth coordination between all stakeholders involved.
Users, Stakeholders & Beneficiaries
The primary beneficiary are the earthquake-affected victims who were desperately waiting for reconstruction funds in order to have a roof over their heads again. CBS, and along with it the GoN, are other beneficiaries who now have a proven, successful, replicable and scalable ICT intervention that can be used for many other data related endeavors.
The survey data have been made open by the CBS and the NPC, thereby, opening doors for wider public sector innovation using this rich dataset.
Results, Outcomes & Impacts
Total Surveyors Deployed 3,000
Nos. of Districts Surveyed 31
Buildings Assessed 1.05 Million
People Reached 5.08 Million
Photographs Captured 9.34 Million
Database size 10 Terabytes
(A) Beneficiaries identified; reconstruction funds disbursed to the beneficiaries quickly and judiciously.
(B) Realizing that the survey data contains valuable socio-economic data also, CBS released the anonymized survey data for wider public use. This has opened doors for use of this data beyond the immediate reconstruction only.
(C) The NPC commissioned the development of eq2015.npc.gov.np web portal which is powered by the data collected during this survey. This portal was commissioned to increase the usefulness and usability of data by making it palatable to both kinds of audience, novice and experts.
(D) This proven system can be used for other data-related projects in Nepal and abroad.
Challenges and Failures
In the beginning, not everyone was confident about mobile survey. Afterall, it had never been tried before.
Bulk of the total survey data size was due to photographs. Therefore, two technical solutions were devised to store and upload them. First, captured photo size was compressed by ten times without visible degradation in quality. Second, the mobile data collection app was designed to upload text and photos separately. Later, the app would automatically match photos to corresponding records upon upload.
The third challenge was to create a central data server system that was robust and scalable to handle half a million rows of data. Since there were about 2,000 engineers on field uploading data every day, the system had to be fail-safe also.
Since this was a time-sensitive project, course-corrections if any, had to be made in time. Therefore, a progress tracking system was developed that would give near real-time statistics of the completed survey to the decision-makers.
Conditions for Success
Coordination among stakeholders: This intervention had over 10 direct and indirect stakeholders (NRA, NPC, CBS, Kathmandu Living Labs, UNOPS, HERD, Real Solutions, World Bank, DFID, MoFAGA, DUDBC, and Department of Vital Registration).
Leadership and guidance: The leadership and guidance shown by the CBS and the NRA helped steer this project to success. UNOPS’ overall logistics management was also played a key role.
Motivation of stakeholders: This is the key. Motivated stakeholders meant that we were all ready to make necessary changes in our plans and approach to achieve the project goals.
Technical knowledge: This was essentially a data project using ICTs. Hence, technical knowledge of both data and ICTs are indispensable parts of this project. Data expertise was provided by CBS while Kathmandu Living Labs delivered the technical know-how.
On the back of the success of this intervention, the NRA is using the same system for its grievance handling survey (new survey).
CBS is planning to conduct a digital census using similar digital system in at least a few areas for its 2021 census.
Poor/no internet connectivity, difficult topography, dispersed human settlements are the realities of Nepal and it will remain so for the foreseeable future. The features of the system (such as offline mobile data collection, efficient data compression and transfer, near-real time progress monitoring system, and efficient server infrastructure that can handle tens of millions of rows of data) are now a proven method of efficiently collecting and processing vast amount of data even in such adverse conditions. Hence, this system can be confidently put in use in many other data-related projects in Nepal and elsewhere.
Communication and coordination is the key. In a multi-stakeholder project like this, clear communication across stakeholders is absolutely vital. Relaying relevant information quickly to everyone concerned is crucial.
Stakeholders should be steadfast on project goal - no doubt - but they should be flexible in their approaches towards achieving that goal. Situations evolve, priorities shift, original plans may not work. In addition, this was a crisis intervention - not everything works according to the plan. All stakeholders should understand and internalize this in order to steer the innovation to success despite changing scenarios.
Often times, impulsive reaction to crises such as this earthquake is to start with the action (in this case, the survey) immediately. While this may give the illusion of speed at the beginning, in the long run, lack of proper planning may actually cause long delays later in the project - sometimes, it may even lead to total failure. Therefore, in this innovation, we deliberately took the time up front in planning the mobile data collection aspect of the survey to the level we could. Careful planning helped us identify potential challenges such as slow internet connections, dispersed surveyors, bandwidth congestion issues at the central server, etc. Consequently, we could anticipate much of the challenges that would later surface as the survey progressed, and could work to mitigate them in time.