Soil Salinization Digital Mapping Coupled With Vegetation Phenology And Soil Land Inference Model In The Oasis Of Xinjiang

Large-scale mapping of soil salinization is one of the most important frontiers in the field of soil informatization.At present,the scale requirements of soil environmental issue are gradually increasing.It is difficult for decision makers to grasp and understand the overall change trend of the large-scale soil environment with the extraction of plaque soil salinization information in small areas.The problem of large-scale mapping is due to the fact that there are too few effective sample points,which leads to greater uncertainty in the area data generated by remote sensing modeling and interpolation.In addition,single-temporal remote sensing parameters lack the influence of time information.Fortunately,the vegetation phenology information is extracted from the remote sensing images throughout the year to reflects the dynamic change trend of vegetation driven by time factors,which solves the shortage of the vegetation index.However,phenological characteristics rarely participate in salinization digital mapping in current salinization research.Therefore,there is an urgent need to build better modeling solutions and research ideas to overcome the bottleneck of large-scale digital mapping.Based on the above problems,we aim to obtain soil-environment relations and draw soil salinization maps from 2010 to 2019 in the oasis of Xinjiang.First of all,Xinjiang is divided into four major eco-geographical divisions.Within the framework of soil genetics theory,we selected 11 kinds of vegetation phenological parameters and 36 kinds of non-phenological land surface parameters,Boruta feature selection is used to screen out the environmental parameters of soil salinity distribution in the oasis of Xinjiang.Discuss the simulation of soil salinization distribution map with Soil Land Inference Model under the scheme with or without phenological parameters from 2010 to 2019 in the oasis of Xinjiang.The main conclusions are as follows:(1)Due to the large spatial heterogeneity in Xinjiang,we divided Xinjiang into relatively homogeneous regions and extracted oasis areas that are more seriously affected by salinization.It is divided into Eastern Xinjiang,southern Xinjiang,northern Xinjiang and Yili region such four regions and 11 sub regions in total,considering the sampling points and subsequent modeling needs,north Xinjiang and Yili region were merged into zone I,and east and south Xinjiang were merged into zone II for follow-up research.(2)The three types of parameter combination schemes of all parameters of Xinjiang,no phenological parameters of Xinjiang and only phenological parameters of Xinjiang are designed to test the contribution of phenological parameters.Taking the years with sampling points as an example,the environmental parameters under the specific combination scheme are selected through the Boruta feature selection.The annual vegetation and salinity index accounted for more than 45%,and the remaining parameters were evenly distributed under the no phenological parameters scheme.However,the annual contribution of vegetation and salinity index decreased significantly,and the proportion of phenological parameters was second only to vegetation and salinity index after coupling the phenological parameters.The climatic parameters such as precipitation and temperature have different degrees of participation in each year.The topographic parameters,especially hydrological factors,are relatively high but the contribution of the overall topographic parameters is relatively small,which plays a role in balancing and stabilizing the subsequent modeling.Through the Boruta feature selection,7 years of environmental parameters were selected.After statistical analysis,the contribution was greater than 5 as a node,a total of 16 and 18 environmental parameters were selected under all parameters and in Xinjiang and no phenological parameters in Xinjiang program,respectively.(3)Using SoLIM and adopting three types of modeling schemes(Full parameter of Xinjiang,No phenological parameter of Xinjiang,Only phenological parameter of Xinjiang)inferencing of soil salinity from 2010 to 2019in the oasis of Xinjiang.The prediction effect of all parameters in Xinjiang is the best and the prediction accuracy of different years varies greatly.Among them,the effect of reasoning accuracy is excellent in 2014(R²=0.7041,RMSE=4.8104,?_c=0.7861).Except for 2013 and 2015,R² fluctuated slightly between 0.67-0.70 and?_c was also above0.75 in other years.In cross-regional prediction,its accuracy will decrease due to the interference of the sampling point area and the number of sampling points,the RMSE increase fluctuates between 1 and 3.(4)The soil salinization showed a continuous“W shape”which first increased and then fluctuated and declined in the oasis of Xinjiang from 2010 to 2019.On the whole,the average salinity reached the highest salinization degree in 2012 and 2013 in Xinjiang,the salinization was the slightest in 2019,the average salinity is 8.26 g·kg^-1 in Xinjiang.The salinization degree of the four major subregions is ranked from slight to heavy in order is Yili Plain,the oasis of North Xinjiang,the oasis of South Xinjiang and the oasis of East Xinjiang,respectively.(5)The average distribution of salinization is mainly non-salinized soil in the oasis of Xinjiang from 2010to 2019.The salinization in Region I is generally lower than that in Region II.The average non-salinized soil accounted for more than half of Xinjiang(52.80%)in the past 10 years.The proportions of slightly salinization,moderately salinization,highly salinization and extremely severe salinization were 7.49%,16.70%,10.80%and 12.21%,respectively.The salinization of the four major eco-geographical zones,Yili Plain,the oasis of North Xinjiang,the oasis of South Xinjiang,and the oasis of East Xinjiang showed a gradual upward trend,while non-salinity gradually decreased,and salinity above severe levels gradually increased.The average soil salt content of Yili Plain,the oasis of North Xinjiang,the oasis of South Xinjiang and the oasis of East Xinjiang were 5.73 g·kg^-1,8.19 g·kg^-1,10.77 g·kg^-1 and 14.53 g·kg^-1,respectively.In summary,the SoLIM inference mapping results are used to clarify the spatiotemporal dynamics of soil salinization in the past 10years in the oasis of Xinjiang.Providing relevant theoretical support to mitigate the harm of salinization,to avoid uneven distribution of water and soil resources,and to prevent the continuous degradation of land ecosystems in arid areas.