The SERVIR-Africa partnership with RCMRD promotes an increase in data sharing among RCMRD’s member nations. A flood modeling system that will help member states anticipate and prepare for flooding is in its pilot stages.
SERVIR joined with the Regional Centre for Mapping of Resources for Development (RCMRD) in 2008 with the aim of improving capacity building and mapping services in RCMRD’s 19 African member states. RCMRD had been formed when these member nations mutually decided that it would be more efficient to pool resources into a regional center, former Regional Director Andre Kooiman said in an interview with Earthzine.
ÛÏThe countries realized they had similar needs regarding capacity building and services for mapping,Û Kooiman said.
The center largely serves to facilitate data sharing among the member states. ÛÏWe establish who is using and co-producing data,Û he said. åÊÛÏ(We are) charting who’s doing what and how institutions can work together and share available spatial information.Û
Data sharing is a problem among nations and within governments, Kooiman says. He recounted how on one project, the crucial missing data was produced and available from another government agency. ÛÏThe offices were not 100 meters apart. The information they required was produced around the corner.Û
Kooiman said the novelty of open-source data is often a contributing factor to the dearth of data sharing. ÛÏInstitutions are not always used to open data policies, and it requires many discussions showing the benefits to convince an institution to take action,Û he said.
One of RCMRD’s current projects is a committee system for open data in the member states.
SERVIR-Africa is also supporting training in general geo-information management, as well as more technical training on geo-data development and meta-data management.
The center also focuses on putting data to good use in the GEOSS Societal Benefit Areas. Disaster management, biodiversity, and health are the primary areas of focus for SERVIR-Africa.
ÛÏWe are proud of having been able to establish these flood monitoring and warning systems to help save lives,Û he said.
Eric Kabuchanga, a technical specialist who works with the flood monitoring system, said the system is based on a predictive model developed in part by NASA’s Goddard Space Flight Center.
ÛÏGoddard and University of Oklahoma partnered to develop the CREST (Coupled Routing and Excess Storage) model,Û Kabuchanga said. åÊÛÏThe model simulates water and energy fluxes. We check for new data every hour and run the model — it generates stream flow and soil moisture predictions.Û
The model was first developed and implemented on a test case — a small watershed in western Kenya. ÛÏThe model was implemented, calibrated, and then validated in this watershed,Û he said. ÛÏThe domain was extended to include Kenya, Uganda, part of Ethiopia, Rwanda, Brunei, and Tanzania, where we have been running the model for the past year.Û
According to Kabuchanga, predictions from the CREST model have only been put to use in Kenya. The Rwandan, Ugandan, and Namibian governments have expressed interest in using the CREST model to predict flooding.
Kabuchanga said the main challenge with implementing the model is calibrating it with historical data. Missing or inaccurate historical water level data can affect calibration and validation of the CREST model.
ÛÏWe haven’t managed to do a calibration for the entire country (of Kenya) because of a lack of data,Û Kabuchanga said. ÛÏThere have always been problems with making water level readings when it was flooding or the water was rising.Û
Kabuchanga and his team are hopeful that records in Rwanda will be more complete and allow a more accurate calibration of the system, which is currently in the early stages of implementation in Rwanda.
He hopes that the model will be useful to other RCMRD member states in the future. ÛÏFloods affect most of the countries that are RCMRD members. We hope to expand to most other watersheds that flood frequently.Û
The CREST model’s output is used to create a Web-based visualization. ÛÏAt the end of the day, we create an average of soil moisture and stream flow and publish the data online in real time.Û
The SERVIR team is hoping to expand the use of the CREST model to include more rigorous flood prediction. ÛÏThese models don’t show in visual whether there is a flood or not,Û Kabuchanga said. ÛÏWe want to create a flood extent map, overlaid with other data sets (population, structures), so that we can do disaster extent reports.Û SERVIR will need to partner with other regional organizations before this goal can be realized.
Kooiman and Kabuchanga say that hesitance to share data is a barrier to SERVIR’s efforts.
ÛÏThe governments are not ready to share this data—they really want to know how you’ll use their data. It’s a problem that we run into with all governments,Û Kabuchanga said. ÛÏWe hope that the pilot we have done in Kenya will help us prove our point.Û