Recurrent Neural network

A common set of drivers explain variation of monthly GW anomaly across the WCS, but there are regional differences. Model without pumping and Land use might be used in areas with deeper GW compared to shallow GW.

An algorithm is developed that can accurately predict and forecast farm-scale regional daily out to 3 days. Daily ET forecast (3 days) model based on random forest (RF) has R2 and RMSE of 0.72 mm and 0.76mm respectively while recurrent neural network (RNN) ensemble forecast model was able to forecast 3 days ET with R2 and RMSE of 0.71 mm and 0.78 mm respectively.