The Research On Nonlinear Cointegration Relationship Of Supply And Demand Of Water Resources In Irrigation District

China is an immense country with a large amount of water resources,,but its water resources are unevenly distributed in time and space.From the perspective of the use of water resources,agricultural water plays a major role,and irrigation water use is a major component of agricultural water use.Therefore,rationally planning the use of irrigation water in agricultural areas and ensuring water safety in irrigation district are the fundamental conditions to realize the rational allocation of water resources and promote sustainable use of regional water resources.Under normal circumstances,precipitation and irrigation water are two major supply sources in the irrigation district.The crop water requirement is the demand for water use in irrigation districts,and the mismatch between water supply and demand is often the main threat to water security in irrigation districts.Affected by changes in human activities,underlying surface,etc.,the data series of water supply and demand in irrigation districts are characterized as uncertainties and ambiguities.Scientifically analyzing and quantifying the uncertain dependencies between the macroscopic tendency and microscopic details of water supply and demand in irrigation districts,and based on this,the non-stationary relationship between supply and demand timing of water resources in irrigation districts and the scientific prediction of irrigation water volume are found and studied to guarantee for rational allocation of water resources and promotion of water security in irrigation district.This paper aims to study the relationship of the data series between water supply and demand in irrigation district.Using multi-time scale analysis method to analyze the time sequence of water supply and demand to study the correlations between them under microscopic conditions.Based on this,combining the unsteady and non-linear characteristics of water resources,a nonlinear cointegration method in economics was used to quantify the relationship between time series.From the perspective of natural,artificial and natural-artificial water resources,the non-linearity cointegration relationship of two-variable(rainfall-crop water requirement,irrigation water-crop water requirement)and three-variable(rainfall-irrigation water-crop water requirement)of the water supply and demand in irrigation district was analyzed.The main conclusions are as follows:(1)Discrete wavelet transform method was used to decompose the original water supply and demand sequence.The water supply and demand time series show consistency in both macroscopic(primary sequence)and microscopic state(multi-time series).The rainfall data series showed a trend of rising at first and then decreasing.The main fluctuation periods were 4~5 years and 6~10 years;The crop water demand data series showed a trend of decreasing at first and then rising.The main fluctuation periods were 2~4 years and 4~ 8 years;The irrigation water data series showed a gradually upward trend.The main fluctuation periods were 2~6 years and 9~11 years.(2)Considering that the time series of water supply and demand in irrigation areas does not accord with the normal distribution characteristics,Spearman and Kendall are used as correlation coefficients for analysis.The results show that the relationship between irrigation water,rainfall,and crop water requirement at multiple time scales is consistent with the original sequence in terms of positive and negative correlation and its extent.(2)Considering that the time series of water supply and demand in irrigation areas does not accord with the normal distribution characteristics,Spearman and Kendall are used as correlation coefficients for analysis.The results show that the relationship between irrigation water,rainfall,and crop water requirement at multiple time scales is consistent with the original sequence in terms of positive and negative relations and the strength of the relationship.(3)Based on the results of multiple time scales,combined with wavelet neural network,two variable multiple time scale non-linear cointegration relations of water supply and demand in irrigation district under natural and artificial conditions were constructed,and multiple time scale BP neural networks and non-linear cointegration relations of the original data series were contrasted.The results show that: There is indeed nonlinear cointegration relationship between ‘irrigation water-crop water requirement’ and ‘rainfall-crop water requirement’,and multi-time scale nonlinear cointegration prediction model is superior to multi-time scale BP neural network prediction model and original sequence nonlinear cointegration prediction model;the magnitude and the positive and negative of the trend term coefficients in the nonlinear cointegration error correction equation are consistent with the results of the correlation analysis;multi-time scale decomposition and nonlinear co-integration theory have significantly improved prediction accuracy;low-frequency terms play a major role in multiple time-scale decomposition results.(4)Based on the results of multiple time scales and wavelet neural networks,a multi-time-scale nonlinear cointegration relationship between three variables under natural-artificial water supplyconditions in irrigation district is constructed.The results show that there is a nonlinear cointegration relationship among these three variables,and the three-variable multi-time-scale nonlinear cointegration prediction model is superior to the three-variable multi-time-scale BP neural network prediction model.and the original sequence nonlinear cointegration prediction model.The magnitude and positive and negative of the trend term coefficients in the nonlinear cointegration error correction equation are consistent with the results of the correlation analysis,multi-time scale decomposition and nonlinear co-integration theory have significantly improved prediction accuracy,and the three-variable prediction result is better than the two-variable prediction result;low-frequency terms play a major role in multiple time-scale decomposition results.