Cycle 8 (2019 Deadline)
Application of partitioned woody and herbaceous forage estimates in index-based livestock insurance, a better alternative to NDVI as a proxy for forage index
PI: Milkah Kahiu (firstname.lastname@example.org), International Livestock Research Institute, in partnership with Jomo Kenyatta University of Agriculture and Technology
U.S. Partner: Niall Hanan, New Mexico State University
Project dates: January 2020 - December 2022
Pastoralists inhabiting African rangelands primarily depend on livestock for their livelihoods but remain extremely vulnerable to droughts, the single largest cause of livestock mortality in the region. To mitigate the devastating impacts of drought, the International Livestock Research Institute initiated Index-Based Livestock Insurance (IBLI) to cushion pastoralists against the adverse impacts of drought. In 2015, the Kenyan government adopted IBLI through the Kenya Livestock Insurance Programme (KLIP), as part of a larger national social protection program. As of 2019, it provides coverage to nearly 20,000 households in eight Kenyan counties, with expansion plans including an additional six counties targeting more than 100,000 households by 2020. Due to the program’s success in Kenya, further expansion plans have focused on Ethiopia, where successful implementation is ongoing in Borana Region and a feasibility analysis in Somali Region has opened doors for further expansion. Other countries where feasibility analysis has promising results are Niger, Uganda, and Somalia, with further plans to cover the larger horn of Africa. Current IBLI contracts are based on an independent index derived from the Normalized Difference Vegetation Index (NDVI) as a proxy for forage scarcity, thus making it immune to manipulations by insurance clients or companies. NDVI is an indicator of vegetation vigor/greenness for all vegetation in a landscape. However, livestock in pastoral systems largely feed on herbaceous foliage, hence limiting the precision of aggregate NDVI for estimating forage availability, especially in areas with significant tree and tall shrub cover. Moreover, intended IBLI expansion into agrosilvopastoral systems poses a challenge in the current insurance contract design due to the mixture of woody cover, crops, and rangeland vegetation. These considerations necessitate review of the current index variables. It is noteworthy that the NDVI data used in the current IBLI is based on eMODIS, and with the MODIS Terra and Aqua instrument lifespans coming to an end, there is a need to explore alternative data sources for IBLI. Thus, the PI Dr. Kahiu and his team will carry out PEER-supported research exploring use of new satellite products for estimating forage index, contract design, and feasibility analysis. Their research will focus on expanding the analysis to include both leaf area index (LAI) and NDVI data from the Suomi National Polar-Orbiting Partnership (NPP) Visible Infrared Imaging Radiometer Suite (VIIRS) to ensure continuity in data flows for IBLI.
|The project team surveys the land (photo courtesy of Dr. Kahiu).|
IBLI is a tool that that addresses the devastating impacts of drought and fosters economic development and growth in the marginalized pastoral areas in Africa. In Kenya and Ethiopia, where a significant portion of the population still lives in poverty with limited access to basic services, affected by chronic droughts and food insecurity, IBLI is a welcome innovation. It has the potential to increase household food security and resilience for the pastoral communities, attract private sector investment and increased public and private capital flows, and improve financial institutions and infrastructures. The planned PEER project facilitates an improved forage index and better contract design, consequently fostering IBLI expansion toward broader impacts on pastoral livelihoods and sustainability across Africa. IBLI is fully operational in Kenya and Ethiopia, with government support contributing to livestock insurance policy in ongoing agricultural sector reforms towards smart agricultural practices. The current PEER-supported research will focus on Kenya and, if successful, could be replicated elsewhere in Africa.
Summary of Recent Activities
In the first quarter of 2022, Dr. Kahiu and her team continued to update their LAI products with new and improved data. This is important, as the index is the component used to monitor forage and to fit livestock mortality models for testing in Index-Based Livestock Insurance product design. The key adjusted elements are:
1. Generation of LAI partitioning data for Africa for 2002-2021 using MODIS LAI Collection: Due to an update to Collection 061, the team is regenerating an updated product for Africa. In early 2022, the team concluded filling data gaps and reviewed the performance of the produced LAI estimates for aggregate, woody, and herbaceous components. Through the U.S. partner Dr. Niall Hanan and his group at New Mexico State University (NMSU), Dr. Kahiu and her colleagues are implementing automated processing chains within the Google Earth Engine. These will be shared publicly once available.
2. Development of new wood cover estimates for eastern Africa to update the LAI partitioning for Africa: The NMSU team developed a Beta version of woody cover (2018) estimates for Eastern Africa. After validation assessment and comparison with other products, the Kenyan researchers agreed to review the product and should complete that process by September 2022.
3. Partitioning updated for Kenya and Ethiopia for 2015-2021: The Kenyan researchers ran a partitioning analysis for Kenya and Ethiopia for using the 40m resolution Beta woody cover product developed by NMSU. The results showed consistency with previous partitioning based on woody cover centered around year 2005. However, local variations in woody LAI and herbaceous LAI estimates are evident in some cases due to the differences in the input woody cover data for 2005 and the higher resolution product at 40m resolution estimated centered in 2018. These results suggested the need to improve the woody cover estimates.
During the spring of 2022, Dr. Kahiu and her colleagues also made a few adjustments to their paper on livestock mortality statistics for 2021. Data clean-up was completed and additional mortality statistics from Marsabit, Kenya, were incorporated into the models. The statistics generated from earlier analysis were based on linear regression models, which produced very poor results, so the team has continued to explore alternative models for improved performance. The results will be presented during the American Geophysical Union Meeting to be held in Chicago in December 2022.
Back to PEER Cycle 8 Grant Recipients