Spatial And Temporal Variation Of Soil Properties Based On Digital Soil Mapping

Soil is the product of climate,parent material,biology,terrain,time and other natural factors.It is the basic resource condition supporting human activities and plant and animal growth,and it is very important for people’s production and life.With the influence of natural factors and human activities,soil properties have complex spatial variability and non-stationarity.Soil p H is one of the important indicators of soil quality.The study of temporal and spatial distribution characteristics of soil p H and its influencing factors is of great significance for soil quality management and sustainable utilization of soil nutrients.In this study,196 and 140 soil samples were collected in Anhui Province based on the data of the second National soil census(1980s)and Soil Series of China·Anhui Volume(2010),respectively.A variety of characteristic variable processing methods and machine learning models were adopted,combined with terrain factors,vegetation indices,climate and other environmental covariables.This study predicted the spatial distribution of soil p H in Anhui Province in 1980 and 2010,selected the best fitting model for this region,and explored the spatio-temporal variation characteristics and driving factors of soil p H in this region in the past 30 years.The conclusions are as follows:(1)Boruta algorithm and recursive feature elimination are the best among the four feature processing methods.According to the characteristics of the filter as a result,in1980 the importance of soil p H distribution variables including latitude,mean annual precipitation,mean annual temperature,multiple resolution valley bottom flatness six environment variables.The most important environmental variables affecting soil p H distribution in 2010 were latitude,mean annual precipitation,mean annual temperature,multiple resolution valley bottom flatness,vegetation enhancement index,multiple resolution valley ridge top flatness,longitude and topographic wetness index.(2)Considering the model performance and mapping effect,the best fitting models of soil p H in Anhui province in 1980 and 2010 were recursive feature elimination-gradient boosting decision tree model and Boruta algorithm-random forest model,respectively.The prediction coefficient R~2 of the two models were 0.42 and 0.69,respectively.The results showed that the distribution pattern of soil p H in Anhui province was similar to that in the whole country,which showed a distribution pattern of southern acid and northern alkali.(3)The mean values of soil p H in 1980 and 2010 in Anhui province were 6.41±1.32and 6.37±1.16,respectively.In the past 30 years,soil acidification occurred in some areas of Anhui Province,and the spatial variation of soil p H had obvious local characteristics.The soil area with p H decreasing accounted for 18.54%of the total area,and the decreasing unit ranged from 0.3 to 1.5,mainly distributed in the southern Huaibei plain,the Yangtze River plain and the hilly mountainous area of southern Anhui Province.51.19%of the total area was the soil whose p H was basically unchanged.The area of soil with p H rising accounted for 29.86%of the total area,and the units of p H rising ranged from 0.3 to 1.8,which were distributed in the north of Huaibei Plain,the hilly land of Jianghuai River and the northern dabie Mountains of western Anhui.(4)In the past 30 years,the p H of different soil types showed different trends.The p H of paddy soil increased as a whole,while that of sandy ginger black soil and aquic soil showed acidification trend.The p H of red soil is basically unchanged.Under different land use patterns,the mean value of soil p H was in the order of dry land>paddy field>grassland>forest land.The area of acid and alkaline soil in dry land and paddy field decreased,while the area of neutral soil increased,the area of acidic soil in forest land increased,and the area of neutral and alkaline soil in grassland increased,but the area of acid soil decreased.(5)Geodetector were used to detect the important factors affecting the spatial variation of soil p H,and the results of factor detection were basically consistent with the results of feature algorithm screening important variables.Average annual rainfall was one of the most important driving factors of soil p H change,and was negatively correlated with p H.In 1980,the interaction of environmental factors showed that the comprehensive explanation power of annual precipitation and multi-scale valley floor flatness was 55.7%,and in 2010,the comprehensive explanation power of latitude and other factors was more than 70%.In addition,combined with the statistical data of the yearbook of Anhui Province from 2000 to 2010 and meteorological research data,acid rain and long-term excessive fertilization and other human factors are also important factors affecting the temporal and spatial variation of soil p H.Figure[31]Table[25]Reference[92]...