Analysis And Prediction Of Time Scale Characteristics Of Reference Crop Evapotranspiration In Different Climatic Regions

The reference crop evapotranspiration is an important index to characterize the atmospheric evapotranspiration,which is an important part of crop water demand research.It is of great significance to the calculation of ecological water demand,the management of agricultural water resources and the rational utilization of water resources.Accurate prediction of future trend of reference crop evapotranspiration can provide important basis for crop planting structure adjustment,division and layout.Based on multifractal theory,wavelet theory and cloud model,this paper analyzes the non-linear and periodic characteristics of ET₀ time series,and constructs a cloud reasoning prediction model.The main results are as follows:?1?Based on the mathematical statistics and multifractal spectrum,the nonlinear characteristics of reference crop evapotranspiration time series are analyzed.The annual,seasonal and monthly ET₀ time series of the four climatic regions have multifractal characteristics,and the ET₀ time series show irregular high frequency oscillation and incomplete random distribution.The monthly scale shows long-range correlation and short-range variation,and the annual scale shows anti persistence and long-range variation,while the time scale and variation distance are irregular.?2?The periodic characteristics and multi-year change trend of the annual,seasonal and monthly ET₀ time series in four climatic regions are characterized by using the real isoline map of Morlet wavelet coefficient,the wavelet variance map and the real process line of characteristic time wavelet coefficient.The annual scale of the four climatic regions shows 28 long time periods,with a downward trend in the future;the seasonal scale shows 6,83 and 87 season time periods,with short time periods nested in the medium and long time periods,and the two cycles interact with each other;the monthly scale has 18 month short time periods.Different time scales have different periodic characteristics.The seasonal scale can show short,medium and long time cycles,and more accurately clarify the periodic characteristics of ET₀ time series.?3?Cloud model can measure the concept of uncertainty and visualize the process of uncertainty.With the reduction of time scale,the expectation and entropy of ET₀ decrease,and the uncertainty of ET₀ time series distribution decreases.Compared with other climate regions,ET₀ in temperate continental climate region is expected to change most acutely in each month,while in plateau mountainous climate region,the change of amplitude is the most gentle,and the change rule of super entropy in different climate regions is not clear on the annual,seasonal and monthly time scales.The accuracy of cloud reasoning prediction model is the highest in the temperate monsoon climate region,and the lowest in the plateau mountainous climate region.With the reduction of time scale,the relative error between the predicted value and the real value increases,and the dispersion decreases.Single condition single rule and single condition multi rule cloud reasoning models can accurately predict ET₀values at different time scales in different climate regions.Single condition multi rule model increases the characteristics of ET₀ periodicity and long-range correlation,so as to reduce the dispersion of predicted value and real value,and further improve the prediction accuracy.