Back in March 2014, I had the opportunity to be part of a World Bank team supporting the Tongan government to develop a reconstruction policy after Tropical Cyclone Ian hit earlier this year. To implement the policy, the Ministry of Infrastructure led a series of surveys to inform housing reconstruction. This post, which does not intend to be scientific or exhaustive, is to share some of the key lessons I learned from this experience.
Damage assessments are routine in the aftermath of disasters, but they differ depending on their objectives (Hallegatte, 2012 - pdf). A rapid survey in the wake of a disaster event could help to estimate grossly the direct human and economic losses and damages. This type of survey is best to capture the amplitude and the severity of the disaster. However, such survey could present some flaws, often because the survey will be conducted in a very short time frame with minimal design. On the other hand, a survey conducted a few months after the event is best to understand better the context of the disaster. It also allows a better design and better preparation. But, equally, such survey could include biases. For instance, the time lag between the event and the survey itself could create some level of challenges. Most likely, people would have started to fix their houses or have moved away from the affected area, and that will add a layer of complexity to the survey.
Housing damage assessment for reconstruction
The first thing was to develop Terms of Reference (ToR) for the assessment. The ToR included the objectives of the assessment, the means to conduct it, the methodologies to be used, the expected results and the timeline. In Ha’apai, we mixed traditional and modern survey instruments for the housing damage assessment. The Ministry of Infrastructure hired engineers to conduct the structural assessment of houses and to report whether a house was repairable or not, habitable or not. For a practical reason, this damage level survey was done using paper forms. Working in a pair with the engineers, SOPAC, a regional science research organization in the Pacific, led the building assessment. Based on experience, the SOPAC team used Trimble—a handheld device with integrated GPS—which allowed faster and more efficient data collection. The team prepared in advance a survey questionnaire and entered the form into the Trimble with existing spatial building points collected by previous projects (including PCRAFI). Each building point was given a unique identifier (ID) using Geographic Information System (GIS). This ID will be the link between the paper forms and the data collected with the device. The SOPAC team captured general information that describes houses—such as the roof and wall types. Such data would inform the houses’ structural vulnerability when combined with the damage level data from the engineers.
Household survey for reconstruction
A household survey was conducted in parallel with the damage assessment. A partnership between the Department of Statistics and the Tongan Ministry of Education allowed the involvement of teachers from Ha’apai as enumerators. The objective of the survey was then to identify the socio-economic vulnerability of households in the affected area. Some households had their houses totally destroyed by the cyclone; some had only partial damage, and some households were not affected at all. The decision to conduct the household survey was based on the year of the last census. The last census for Tonga was conducted in 2011, and it was decided that such information needed an update given the dynamic of households on the islands (the affected island was losing population by emigration to Tongatapu, the most populated island, or abroad); and also because, ultimately, the team was interested in working toward increasing household resiliency and not just reconstructing buildings. Nevertheless, the existing census was used as a baseline to generate a pre-list of households that the enumerators updated in the field during the survey. In addition to the usual demographic survey, households were asked about the impacts of Tropical Cyclone Ian on their assets as well as their self-recovery efforts.
Like any data collection, data quality control is the key in whatever type of survey instruments used. Such mechanism needs to be set up at the design level. One of the main challenges for the Tonga damage housing assessment was to link the three data sets from the building survey, the structural damage assessment and the household survey. Therefore, it is important to define in advance a reliable link between data sets and to set up redundant identification between the surveys. For Tonga, we used the “Building ID” and the “Name of Head of Household," but beware of names because “texts” are always tricky when it comes to statistical analysis. It is more advisable to use numerical IDs, which are easier to manipulate. The use of GPS-integrated devices also helped to directly geo-locate the surveyed houses and link them to the matching households. In the future, I would encourage exploring more the use of smart devices and new technology for damage and loss assessment surveys. I believe this type of data collection could save time and reduce systematic errors in surveys.
What is your experience in conducting surveys? I would be very interested in hearing your perspective on the topic.