Forecasting seasonal rainfall patterns for crop production in Juba County, South Sudan using the Artificial neural networks

Abstract

A simple Multi-Layered Feed Forward (MLF) training process of the Artificial Neural Networks model with a learning back-propa-gation algorithm was applied to forecast rainfall data of Juba County, Central Equatoria State in South Sudan from 1997-2016. Annual rainfall data were aggregated into three seasons MAMJ, JAS and OND and later trained for best predictions for the period 2017-2034 using the Alyuda Forecaster XL software. Best training was attained once the minimum error or cost function of the weight  was attained during gradient descent and expressed as Mean Square Error (MSE) and AE of the input variable. The results showed that for MAMJ and JAS months, the number good forecasts were over 97% whereas this was between 60-80% for OND months. The Seasonal Kendal (SK) test on future rainfall forecasts as well as the Theil-Sen slope showed a declining mono-tonic trend in the mean amounts in all three seasons with MAMJ, JAS at OND at 100, 150 and 80 mm respectively towards the end of 2034. Forecast of the Standardized Precipitation Index (SPI) showed that the MAMJ months for the years 2019 to 2027 will be mo-derately wet with near to normal drought except in April 2021 which will experience some severe wetness. Interdecadal severe drought is expected between 2028 to 2033 after almost two decades. The SPI of JAS and OND seasons will remain near normal to moderate drought during the same period. Declining onset of MAMJ rains is expected to significantly affect the timing for land preparation and crop planting. The forecast accuracy of the MLFFNN can be used as a vital tool for decision makers in projecting future rainfall and drought events.