Prediction Of Soil Selenium Spatial Distribution In Shitai County Based On Multi-source Geographic Data And Random Forest Model

This study takes Shitai County,Chizhou City as the research area.Based on the sample points collected by Anhui Geological Experiment Research Institute in 2014,combined with Arc GIS,Geo Da,Sim DTA,R language and Google Earth Engine and other platform software,geostatistical and random The forest model was used to compare and explore the main influencing variables affecting the spatial distribution of soil selenium in Shitai County,including topography,vegetation factors,and soil types.Combined with important influencing variables,the spatial distribution of soil selenium in Shitai County was predicted.The spatial distribution map of soil selenium from township-village-grid in Shitai County was obtained by combining traditional geographic methods and machine learning methods.Finally,according to the results of the division of soil selenium in Shitai County,the opinions on the protection and development of selenium-enriched industry in Shitai County are put forward.The main findings are:(1)Descriptive analysis of soil selenium elements in Shitai County,and further spatial autocorrelation analysis of soil selenium content by Moran’s I index,the following conclusions are drawn:(1)Soil selenium content in Shitai County ranges from 0.120 to 8.704 mg /kg,the coefficient of variation was 10.5%,and the variation range was large,showing a moderate degree of variation;the average value was 0.601 mg/kg,and Shitai County had a strong ability to develop selenium resources.(2)There is a positive spatial correlation between soil selenium and various variables in Shitai County,and the global autocorrelation shows a clustering trend,and the correlation is relatively significant.(2)The relationship between soil selenium and 10 variables in Shitai County was analyzed by correlation,and the conclusions were as follows: terrain factors and soil types had significant effects on soil selenium content in Shitai County.The correlation coefficient between soil selenium content and elevation in Shitai County was 0.847(P<0.01),showing a very significant positive correlation;the correlation coefficient between soil selenium content and fluctuation was 0.349(P<0.01),showing a significant positive correlation Extremely significant positive correlation;the correlation coefficient between soil selenium content and terrain roughness index is 0.259(P<0.01),showing a very significant positive correlation;the correlation coefficient between soil selenium content and terrain moisture index is-0.138(P﹤0.05),showing a very significant negative correlation.After statistical analysis of the proportion of soil types in Shitai County,it was concluded that the content of selenium contained in different soil types from high to low is: coarse bone soil ﹥ red soil ﹥ lime soil ﹥ yellow soil ﹥ paddy soil.In terms of coefficient of variation,the variation degree of yellow soil is very small,that of red soil and coarse bone soil is moderate,and that of paddy soil and lime soil is very high.(3)Using the combination of geostatistics and random forest methods,the spatial distribution of soil selenium in Shitai County was predicted.First,a random forest regression model was established,and the variables were predicted to obtain the coefficient of determination R2=0.7274,and the importance of each variable was ranked.Combined with the prediction results of %Inc MSE,it was confirmed that the elevation,terrain roughness index,terrain humidity index,and fluctuation degree It has an important influence on the spatial distribution of selenium in soil.Secondly,according to these important variables,the random forest model is used to predict the selenium element value in the global range,and the interpolation and random forest model are combined to produce the spatial distribution map of soil selenium element in different ranges of township-village-grid,and the following conclusions are drawn:(1)1% of Shitai County is selenium deficiency area,2% selenium excess area,enough selenium area and selenium-enriched area up to 95%,most of Shitai County belongs to usable selenium-enriched area Values are sorted from high to low as Xianyu Town > Lilac Town > Xianyu Mountain Forest Farm > Huangshakeng Forest Farm > Hengdu Town > Renli Town > Zhonglong Mountain Forest Farm > Qidu Town > Dayan Township > Jitan Township > Xiaohe Town >Guniujiang Nature Reserve.(3)Among the 85 villages in Shitai County,there are 38 selenium-enriched and selenium-sufficient villages,and 11 villages whose selenium content is between 0.40-3.0 mg/kg are selenium-enriched areas,and the number of villages accounts for 13 of all villages in Shitai County.%,the area is 14%.The values of 27 villages are all between 0.40-0.175mg/kg,which are selenium-sufficient areas.The number of villages accounts for 32% of the county,and the area accounts for 35% of the county.(4)According to the above prediction results,corresponding opinions are put forward on the development and protection of soil selenium in Shitai County.It is divided into three different types of plans for selenium-deficient area and selenium-deficient area,sufficient selenium area and selenium-rich area,and selenium-excessive area.Establish a selenium resource database;(2)Using local selenium-rich resources to develop specialty products;(3) Strengthen industry-university-research cooperation to improve product innovation;(4)Conduct industrial upgrading around eco-tourism.