Determining optimal seasonal integration times of NDVI series for index-based livestock insurance in East Africa

Coarse-resolution NDVI time series, aggregated in space and time, are used as a seasonal forage scarcity index for an existing livestock insurance scheme in East Africa. Payouts are made to pastoral households if the index drops below a specific threshold, corresponding to drought conditions. This paper's aim is to improve the seasonal definitions used in the scheme and to evaluate options for further anticipating the payout in time. To achieve this, we first performed a phenological analysis of SPOT-VGT FAPAR series resulting in location-specific season start- and end-dates. The resulting seasonal definitions were then used for calculating a forage scarcity index from eMODIS NDVI. Subsequently we evaluated if high end-of-season index predictability was maintained when bringing end-dates of the temporal integration further forward in time. Cross-validated statistics showed that the payout time after drought can be advanced by 2-3 months, allowing pastoralists to take protective measures to safe their livestock.