Decision-making method for maize irrigation in supplementary irrigation areas based on the DSSAT model and a genetic algorithm

Supplementary irrigation area in north China is an important agricultural production area, but water shortages have limited the agricultural production considerably. Irrigation is required to ensure crop yield during growth seasons with variable rainfall. Most irrigation schedules are based on typical meteorological years. However, it is very difficult to predict the meteorological information of irrigation year in advance, which limits the planning of irrigation schedules. Therefore, taking the central part of Shanxi Province (supplementary irrigation area) as an example, this study focuses on the optimal irrigation decision-making method of supplementary irrigation area. The irrigation schedule of summer maize from 1970 to 2020 was firstly optimized by using DSSAT model and genetic algorithm. On this basis, a decision-making method based on the ratio of cumulative rainfall to cumulative ETc between plantation date to the current date to determine the irrigation time is developed. The results indicated that using P/ETc (the ratio of cumulative rainfall to cumulative crop water requirement) as the criterion leads to superior decision-making. Two supplementary irrigation events can meet the growth demand of summer maize. Irrigation events decisions were made on days 44 and 58 after summer maize sowing, the accuracy of this method for two irrigation events decisions is 82.35 % and 86.27 %, respectively, and the average accuracy of the two irrigation event decisions was 84.31 %. At the same time, a support vector machine algorithm is used to make decisions about irrigation, comparing the decision accuracy with the results of the decision method proposed in this paper. The accuracy of the support vector machine decisions for the first and second irrigation events was 82.35 % and 80.40 %, respectively, and the average accuracy of the two irrigation event decisions was 81.40 %. The decision model was also superior to that involving support vector machine algorithms for the irrigation decision of summer maize can be used for irrigation decision-making in similar supplementary irrigation areas with variable rainfall.