Forecasting vegetation greenness with satellite and climate data

A new and unique vegetation greenness forecast (VGF) model was designed to predict future vegetation conditions to three months through the use of current and historical climate data and satellite imagery. The VGF model is implemented through a seasonality-adjusted autoregressive distributed-lag function, based on our finding that the normalized difference vegetation index is highly correlated with lagged precipitation and temperature. Accurate forecasts were obtained from the VGF model in Nebraska grassland and cropland. The regression R² values range from 0.97-0.80 for 2-12 week forecasts, with higher R² associated with a shorter prediction. An important application would be to produce real-time forecasts of greenness images.