Sediment Source Analysis Of Intensive Land Use Watershed In Red Soil Region Of South China

In order to quantitatively study the erosion and sediment yield characteristics of different land use types in intensive exploitation watershed of Red Soil.This study selected an intensive exploitation watershed in the town of Gande,Anxi County as research area,and collected the soil samples(tea plantations,forest land,farmland and the mining area)and the river sediments samples.Analysing 41 kinds of geochemical characteristics of soil samples and sediment samples,selecting the best fingerprint factor by the composite fingerprint method,and calculating the sediment contribution rates by multivariate mixing model.The results showed:(1)Using the Kruskal-Wallis H test to screen out fingerprinting factors,which are significant differences(P<0.05)and can identify different sediment sources,as well.The result show that,in the Industrial and mining-agricultural leading area of A,34 factors pass the test of Kruskal-Wallis H.In the typical farming area of B,24 factors pass it.In the intensive land use area of C,35 factors are satisfied with the test.The stepwise discriminant analysis shows that,In the Industrial and mining-agricultural leading area,the best fingerprint combination is made up of 4 factors which are A-P,Magnetic susceptibility,Nd and La.In the typical farming area,the best fingerprint combination is made up of 3 factors which are In,Tm and Pb.In the intensive land use area,the best fingerprint combination is made up of 8 factors which are A-P,Cu,P,Y,Ca,Ga,Sn,and Nd.The total cumulative discriminant rates of A,B,and C reached more than 90%,and the correct classification rates of different groups samples in A,B,and C were 87.9%,100%,and 90.3%,satisfying the application requirements of combination fingerprinting methods completely.(2)The multivariate mixture model is applied to calculate the sediment contribution percentage in each potential sediment source.In the industrial and mining-agricultural leading area,the mean sediment contribution rate of the frontal rainy season is:mining area(61.18%),tea garden(12.16%),forest land(11.31%),farmland(15.35%);but in typhoon rainy season,mining area(64.47%),tea garden(16.50%),forest land(13.42%),farmland(5.61%).Under different rainfall characteristics,the mean sediment contribution rate and the unit area sediment contribution percentage of the mining area are the highest.That’s to say,mining area is the main sediment source of the basin.The goodness-of-fit test shows that the average GOF of the frontal rainy season in this area is 0.83±0.02,while the rainy season in the typhoon is 0.84±0.02,both of them are higher than 0.80,indicating that the multivariate mixture model is reasonable in fitting results.(3)In the typical farming area,the mean sediment contribution rate of the frontal rainy season is:forest land(40.98%),tea garden(40.94%),farmland(18.09%);but in typhoon rainy season,forest land(35.63%),tea garden(41.29%),farmland(23.08%).Under different rainfall characteristics,the forest land and tea garden are the main source of sediment in the basin.The goodness-of-fit test shows that the average GOF of the frontal rainy season in this area is 0.88±0.04,while the rainy season in the typhoon is 0.90±0.02,both of them are higher than 0.80,indicating that the multivariate mixture model is reasonable in fitting results.(4)In the intensive land use area,the mean sediment contribution ratesi of the frontal rainy season is:forest land(28.59%)>tea garden(28.20%)>mining area(27.59%)>farmland(15.61%),the unit area sediment contribution percentage of mining area is the highest.But in typhoon rainy season,tea garden(29.05%)>forest land(25.98%)>mining area(25.43%)>farmland(19.54%).Under the unit area,the sediment contribution percentage of mining area is 6.7 times,9.3 times,and 10.2 times higher than farmland,tea garden,and forest land.Namely,the mining area is the most affected by rainfall erosion.The goodness-of-fit test shows that the average GOF of.the frontal rainy season in this area is 0.85± 0.04,while the rainy season in the typhoon is 0.85±0.02,both of them are higher than 0.80,indicating that the multivariate mixture model is reasonable in fitting results.