Soil Erodibility Characteristics Of Main Monitoring Points In Hubei And Applicability Evaluation Of Estimation Model

Soil erodibility K is one of the important factors of regional soil and water loss evaluation model.At present,there are many calculation methods of K value at home and abroad,but its applicability is still unclear.In this study,the soil at the main monitoring stations of 8 water and soil conservation zones in Hubei Province is taken as the research object.The runoff plot observation,soil physical and chemical properties determination and indoor artificial rainfall simulation test are carried out.Based on different soil erodibility estimation models,the soil erodibility K values of the main water and soil conservation monitoring points in Hubei Province are compared and analyzed,and combined with relevant analysis methods,This paper expounds the physical and chemical properties of different soil and water conservation monitoring points and the influencing factors of aggregate stability,reveals the characteristics of slope erosion and erodibility K value of soil at different soil and water conservation monitoring points under indoor simulated rainfall,discusses the applicability analysis of different soil erodibility K value estimation models,selects the key factors of soil erodibility measurement method K value in field runoff plots,and constructs and evaluates the prediction model.The main results of this paper are as follows:（1）The physical and chemical properties of soil at the main monitoring points in different water and soil conservation zones are quite different,and soil organic matter and clay are closely related to the stability of aggregates.Three treatments（fast wetting,slow wetting and mechanical oscillation）were carried out on soil aggregates by LB method.The results showed that rapid wetting had a great impact on aggregate crushing.Pearson correlation analysis showed that soil clay,soil particle density,organic matter,soil bulk density,alkali hydrosazable nitrogen,available potassium and available phosphorus were closely related to the stability of soil aggregates;Among the stability indexes of soil aggregates studied by RDA ranking analysis,the first axis contribution rate of water stable aggregates R_>0.25 is the largest（85.21%）,followed by the contribution rate of average weight diameter（MWD）（78.02）,and the contribution rate of relative dissipation index（RSI）and relative mechanical failure index（RMI）was 74.61%.The forward selection analysis of soil aggregate stability in environmental factors showed that the contribution rate of organic matter to R_>0.25 was 57.3%,and the contribution rate of clay to MWD and RSI/RMI was the largest,61.5%and 55.1%respectively.（2）Based on the indoor simulated rainfall,the runoff rate,sediment yield,cumulative runoff,cumulative sediment yield,flow velocity and soil erodibility K value of the same soil increase with the increase of rainfall intensity.Under different rainfall intensities,there is a good linear and exponential function relationship between cumulative sediment yield and rainfall duration,and the determination coefficient R² between the models is above0.950.Under different soil and water conservation monitoring points,the soil erodibility K value under the same rainfall intensity is quite different.Under the 40 mm/h rainfall intensity,Ⅷ（Zigui chenjialing monitoring point）is the largest,andⅣ（Huanggang Shiqiaopu monitoring point）is the smallest,with a difference of 0.0215t·hm²·h/（hm²·MJ·mm）;Under 60 and 80 mm/h rainfall intensity,the maximum value ofⅡ（Enshi Changyan monitoring point）and the minimum value ofⅢ（Jingzhou Yanqiao reservoir monitoring point）are shown,and the difference between the maximum value and the minimum value is 0.0238 and 0.0245 t·hm²·h/（hm²·MJ·mm）respectively.（3）Based on the conditions of field runoff plots,the annual average soil erodibility of different water and soil conservation monitoring points is as follows:VI（Xiangyang Xionghe reservoir monitoring point）is the largest,VIII（Zigui chenjialing monitoring point）is the second,and I（Caidian West Lake monitoring point）is the smallest.In the accuracy test of five soil erodibility K value estimation models,the average absolute error（0.0169）,average relative error（0.0901）and root mean square error（0.0197）of Shirazi model are closer to 0 than those of other models,followed by Torre model,which are 0.0171,1.0091and 0.0203 respectively,while Nomo method is the largest,which are 0.0183,1.0987 and0.0230 respectively;The accuracy factor of Torre model is close to 1 compared with other models.Therefore,the K value of 8 soil and water conservation monitoring points in Hubei Province estimated by Shirazi model and Torre model is the best,which is suitable for the quantitative evaluation of soil erosion in this area.（4）The key factors affecting the measured value of soil erodibility were selected through grey correlation analysis.The soil factors with a correlation degree greater than 0.5include 12 factors,such as silt,clay,R_>0.25（FW）,MWD（SW）,R_>0.25（SW）.Pearson correlation analysis of key factors showed that clay particles were significantly positively correlated with organic matter,total porosity,average weight diameter under slow wetting and mechanical shock treatment（P<0.05）,and highly significantly positively correlated with water stable aggregates R_>0.25 under rapid wetting treatment（P<0.01）.The soil erodibility K value is predicted by BPNN for four key factors(silt,clay,R_>0.25（FW）and MWD（SW）).The results showed that the 4-3-1 structure has the best effect on predicting the soil erodibility K value.The relationship formula is:K_{measured value}=0.578 K_{predicted value}+0.010,R²=0.652.Therefore,the prediction data of soil erodibility and four key factors based on BP neural network are effective.