Measurement and analysis of regional agricultural water and soil resource composite system harmony with an improved random forest model based on a dragonfly algorithm

Abstract

To expand the field of harmony theory research, new paths for sustainable development of agricultural regions are explored. The uncertainty problem is selected to determine the optimal parameters of the random forest (RF) model. This paper uses the Dragonfly algorithm (DA) to calibrate the RF model parameters, proposes an improved model (DA-RF), and applies it to measure the agricultural water and soil resources composite system harmony (AWSRCSH) of the Jiansanjiang Branch of the Heilongjiang Great Northern Wilderness Agribusiness Group Corporation. The results show that the DA-RF model can effectively improve the simulation ability and stability with a running time that is longer than that of an empirical method. The annual variation in the AWSRCSH of the Jiansanjiang Branch presents an “N-type” trend. From 1997 to 2005, the branch had not yet entered a peak period of resource and agricultural development, while the AWSRCSH had steadily improved during this period. From 2005 to 2008, due to predatory operations, the AWSRCSH decreased, and the speed of this decrease increased. From 2008 to 2016, since local water and soil resource development was strictly controlled and agricultural modernization increased, the AWSRCSH gradually increased. Due to the advantages of water resources and topography, the AWSRCSH in the farms adjacent to the river was greater than that in inland farms. To address the slow growth rate of the AWSRCSH, some adaptive regulation strategies were proposed according to key driving factors, and the effect of harmonious regulation was predicted.