Description

Abstract: LEWS has developed a new forage forecasting technology that provides a comprehensive view of emerging forage conditions, as well as 90-day forecasts updated every 10 days. Predicting spatial forage availability will make it possible for pastoralists to assess impending livestock mortality by kind and class of animal and decline in milk production. With the new system, pastoralists will have more flexibility in decision-making, leading to timely destocking strategies and an assurance of greater ecosystem integrity. The primary Goal of the GL-CRSP Livestock Early Warning System Project (LEWS) is providing pastoral communities and supporting organizations with timely and high-value assessments of emerging forage conditions. Information on current trends in forage on-offer to livestock and the rate of change for conditions across East Africa can be provided by projecting forage conditions from computer models (driven by satellite-based weather data representing points on the ground) and coupling those projections with corresponding NDVI satellite forage greenness data. The forecasting procedure analyzes and projects equally spaced univariate time series data, transfer function data, and intervention data using an AutoRegressive Integrated Moving-Average (ARIMA) model. This approach predicts grazed standing crop of forage in a response time series as a linear combination of its own past values (modeled and NDVI data), past errors (shocks), and current and past values of other time series. Projections of 30, 60, and 90-day forage standing crops resulted in R2 greater than 0.93, 0.81, and 0.71 with standard errors of prediction of less than 141, 206, and 259 kg/ha of available forage, respectively. This methodology is a powerful new mechanism for decision makers to identify emerging hot spots that may be difficult to perceive, and determine if they are going to recover or worsen with a progressive 90-day analysis window. The forecasts are well within normal sampling error, indicating that this new tool is valuable for predicting near-term forage response.