Intelligent irrigation water requirement system based on artificial neural networks and profit optimization for planting time decision making of crops in Lombok Island

Cropping pattern is a scheduling for farming time on a certain land in a definite period (e.g. 1 year), including unfilled area. In arranging crop planting patterns, hydrological (rainfall), climatological (temperature, humidity, wind speed, and sunshine), crop (crop coefficient value, productivity and price) and land area data are required. Therefore, a method that can be applied to predict the hydro climatological data is needed. The appropriate method for such prediction is Back Propagation Neural Network (BPNN). Prediction result of BPNN will be used to determine minimum crop water requirements, and it will be associated with planting time (age) of each crop for making cropping pattern. The design of most favorable cropping pattern will obtain the maximum profit and reduce fail harvest problem, which in turns it can contribute to national food resilience. Based on the simulation result, it was known that the BPNN with two hidden layers is able to predict hydro climatological data such as of rainfall, temperature, humidity, wind speed, and sunshine data with an average accuracy rate of 95.72% - 96.61%. Meanwhile, validation of predictions obtained an average percentage error of 1.12% with an accuracy of 99.76%. The results of the optimization of the cropping pattern in Lombok in March 2013-February 2014 revealed an accurateness of profit in each district/city in East Lombok, Central Lombok, West Lombok, North Lombok, and Mataram increased 2.02%, 16.88%, 20, 23%, 21.89%, and 5.58%, respectively. Over all, the increasing average was found to be 13.3% from the previous year.