| Jianshui,Yunnan Province is located in the Karst Fault Basin.Moisture is an important bottleneck restricting vegetation restoration in the area.This area is in Karst area.It has characteristics such as thin soil,poor water holding capacity,serious leakage of water,and sensitive soil moisture to meteorological factors.This article takes Jiubiao small watershed in Jianshui County of Yunnan Province as an experimental monitoring area.I collect weather data from 1960 to 2016 in Jianshui County and comprehensively analyze meteorological elements in Jianshui County.Set up uninterrupted soil moisture monitoring system and small weather station in the test area.Collected three kinds of artificially restored vegetation with the same terrain and climate:Pinus massoniana,Cupressus duclouxiana,Eucalyptus maideni,and three kinds of shrub vegetation with the same terrain:low-coverage Dodomaea viscosa,high-coverage Dodomaea viscosa and natural secondary forest:Thorn thorn shrub soil moisture data.In this paper,the horizontal vertical variation of soil moisture content under different vegetation types and the response law of soil moisture content to rainfall were studied.The linear and nonlinear analysis of the relationship between the soil moisture and the average temperature relative humidity and the atmospheric pressure of seven meteorological factors was conducted by stepwise regression to BP neural network.The main results are as follows:(1)The analysis of weather data in Jianshui County is known,the main types of rainfall in Jianshui County are moderate rain and light rain.The frequency of rain accounts for about 17%,and the frequency of light rain accounts for about 78%.In Jianshui County,there were inter-decadal fluctuations in precipitation from the 1960s to the 2000s,and evaporation decreased significantly after the 2000s.Dryness was greater than 3.0 in the 1960s and 1980s,and between 2-3 in the 1970s and 1990s.(2)Study on the coefficient of variation of soil moisture under different vegetation types found that the coefficient of variation of soil moisture in each layer of arborous forest is:Cupressus duclouxiana>Eucalyptus maideni>Pinus massoniana.The coefficients of variation of 0-10cm and 10-20cm layers of shrubbery land were as follows:low-coverage Dodomaea viscosa>high-coverage Dodomaea viscosa>Thorn thorn shrub;Coefficient of variation in the 20-30cm layer of the shrubbery:low-coverage Dodomaea viscosa>Thorn thorn shrub>high-coverage Dodomaea viscosa.(3)In the same forest land,the depth of the impact of rainfall on the soil increases with the increase of rainfall.The soil moisture content has a rapid response to heavy rain and heavy rain,and the soil moisture content in each layer has been significantly affected.The medium-rain and light-rain conditions have great influence on the soil moisture of 0-10cm in the surface layer of each forest,and have little effect on the soil moisture below 10cm.The response of different types of forestland soil moisture to precipitation changes is significant.(4)Through linear regression analysis,the soil moisture content of different soil layers in different vegetations has different fitting accuracy.The goodness of fit of the three arbor-like regression equations was higher than the three shrub lands.In bush lands,linear regression analysis is not suitable for fitting due to severe rocky desertification.(5)Non-linear fitting of soil moisture through BP neural network relationship model found that the accuracy of simulated soil moisture content and the sensitivity to various meteorological elements are not the same.The Pinus massoniana plots and the high coverage mulberry plots are all the most sensitive to rainfall.The simulation accuracy of the two plots is high,and the simulation accuracy shows the law of decreasing with the deepening of the soil depth.Comparing regression equations and BP neural network model for soil moisture simulation results,it was found that the simulated effect of BP neural network model on soil moisture content was significantly better than the linear regression fitting model. |
PDF Full Text Request