Data-mining in Kenya

March 28, 2013 by Travis Franck

“The data might just be in filing cabinets at each border crossing.”

We always knew it was going to be tough to make a model of drought-induced population displacement in Kenya, and this comment in one of my meetings last month confirmed it.

I was in Nairobi to meet with experts and decision makers about our pastoralist climate risk and resiliency simulation, showing them the work we have done so far on a prototype simulation, and looking for leads on new data. It was a week full of meetings filled with learning about the complexities of rural life in Kenya. I was fascinated to hear about all the different factors – from where national borders lie to international aid priorities – that are affecting the resiliency of pastoralists.

Along with our partner from IDMC, I demoed our real-time simulation that combines rainfall/drought, pasture quality, livestock population, and international food assistance to explore the interplay of rainfall, livelihoods, and displacement.  In the real world and in the simulation, when rainfall changes, pastoralist livelihoods suffer and families can become displaced from their homes and grazing lands. The simulation shows the interplay of factors, and eventually we expect to be able to help decision makers test how different policies can help reduce the risk of drought-induced displacement.

During my trip I was able to meet many groups that are supporting Kenya’s pastoralists, including, International Livestock Research Institute (ILRI), UN Population Fund (UNFPA), pastoralist advocacy group REGLAP, the International Organization of Migration (IOM), professors at the University of Nairobi, and several pastoralists from northern Kenya. Reactions to the simulation were very positive and reinforced our sense of need for this work.

While some speculated that portions of the additional data we are looking for may be inaccessible (unless we find an intrepid grad student ready to do some pro bono fieldwork), we’ve still been able to uncover and connect data on important trends, for instance creating datasets that show the relationship between amount of rainfall and the distance livestock must travel to reach water.

One of the challenges of this project, not unlike some of our previous work, has been that we need to bring together research from many different fields. In this case we’re combining pasture research, livestock research, and pastoralism social science research. For now we’re continuing to integrate the different datasets and adjusting our simulation in light of newly discovered data. We are also expanding the model to include more policy interventions, like livestock restocking and changes in land tenure and access. Soon we’ll be expanding the framework to include cross-border flows of pastoralists in the Horn of Africa — between Kenya, Somalia, and Ethiopia.

With some unavoidable climate change already locked into the momentum of the climate system, we expect that more regions around the world are going to be needing risk reduction scenario planning tools, and we are hopeful that our style of interactive user-friendly simulations can help connect people with the data, analysis, and insight they will need.