Spatial Multi-Scale Variation,Driving Mechanism,Prediction And Early Warning Of Soil Salinization In The Coastal Area Of The Yellow River Delta

The Yellow River Delta is a typical area of soil salinization in China’s coastal areas.Soil salinization is the main factor restricting the sustainable development of the region.Exploring its salinization variation characteristics and driving mechanism,and then realizing the prediction,prevention and early warning management of salinization is the premise of rational use of salinized soil,which is of great significance to ensure food security and achieve sustainable agricultural development in the Yellow River Delta.This study took Kenli District,a typical area of the Yellow River Delta,as the research area,and explored it in three scales:region,land type and profile.The regional scale corresponded to the whole study area.The land type scale was divided into cultivated land and wasteland.The profile scale was based on the soil configuration.In terms of regional and land type scale,this study conducted field sampling,used spatial interpolation to analyze the spatial variation law of regional and land type scale salinization,extracted the spatial distribution data of factors such as climate and elevation,screened and determined the main influencing factors,used regression analysis and geographic detector methods to analyze the driving force,explored its driving mechanism,and adopted random forest to build the driving model.In terms of the profile scale,the indoor soil column test method was used to find out the salt distribution law on the vertical scale,to explore the driving mechanism of soil configuration,groundwater salinity,groundwater level and coverage on the profile water and salt transport,and to build the profile scale salinization driving model by using multiple linear regression.Besides,the spatial distribution of soil salinization in autumn was predicted based on the driving model and historical data at various scales,and the regional land ecological use and crop yield are taken as the basis for early warning classification,so as to realize multi-scale salinization early warning and provide theoretical basis and technical support for the prevention and treatment of salinization disasters.The main results are as follows:（1）Multi-scale spatial variation of salinityThe spatial distribution of salt in the Regional Scale was the distribution pattern of high in the East and low in the west,high in the north and low in the south.From the southwest inland to the Northeast coastal area,the soil salt content gradually increased,the soil salinization gradually increased,and there was an obvious difference near the runoff of the Yellow River,which decreased locally.Land type scale:The salinization level of cultivated land was low,and most areas were 1.5-5.0 g·kg^-1.The overall distribution was light and moderate salinization from the west to the Huanghekou Town,and the eastern coastal area was distributed with strip-shaped severe salinization along the Northeast southwest parallel coastline.The soil salinity level of wasteland increased from west to East and from north to south,and its spatial distribution of salinity was consistent with the regional scale.Profile scale:The maximum influence thickness of mineralized groundwater on soil salinity above the water level was 130cm,and it could reach 140 cm near the ground.At the initial stage of water supply,the salt content in the soil raised from the surface to the simulated groundwater level,and the soil salt content had a linear positive correlation with the profile depth.With the increase of observation time,the salt accumulation in the surface was significant,the salt content in the surface was high,the middle was low,and it increased again near the water supply location.The salt content in the surface layer of the clay layer soil configuration was lower than that in the homogeneous loam configuration,and the salt content above the clay layer was lower than that below.（2）Driving mechanism of multi-scale salinizationThe driving forces of regional scale factors on salinization are as follows:coastal distance（0.7623）>Groundwater Depth（0.7266）>Cumulative Precipitation（0.7241）>Groundwater Salinity（0.6974）.The multi factor comprehensive driving force was greater than the single factor driving force,in which the core driving force of regional salinization was the decrease of coastal distance,the shallower underground depth,the increase of groundwater salinity and the decrease of cumulative precipitation.Meanwhile,the salinization driving model based on random forest had the highest accuracy,and R² reached 0.9541.The driving forces of different land types changed significantly.The driving forces of various factors of cultivated land on salinization are as follows:the Yellow River（0.6323）>Coastal Distance（0.6277）>Cumulative Precipitation（0.5933）>Groundwater Depth（0.5776）.Irrigation and drainage facilities in the characteristic factors effectively inhibited the salinization level of cultivated land.The driving force intensity of various factors of wasteland on salinization was:Cumulative Precipitation（0.8286）>Groundwater Depth（0.8216）>Coastal Distance（0.7895）>Groundwater Salinity（0.7625）.Among the characteristic factors,micro terrain had the greatest impact on wasteland salinization.The driving model constructed for different land types had high accuracy,R² of cultivated land salinization driving model was0.8792,and R² of wasteland salinization driving model was 0.5266.At the profile scale,the driving forces of various factors on the profile salinity distribution are as follows:Groundwater Salinity（0.6809）>Groundwater Level（0.3924）>Clay Layer Thickness（0.3268）>Clay Layer Depth（0.2925）.Beyond that,the salinity of groundwater and the depth of intercalated layer were positively correlated with the conductivity of the profile,while the depth of groundwater and the thickness of intercalated layer were negatively correlated with the salinity of the profile.There was a linear negative correlation between surface hay coverage and evaporation,which affected soil volume salt by affecting evaporation.It had a high accuracy to establish profile scale salinization driving model using multiple linear regression,and R² was 0.9517.（3）Salinization prediction and early warningThe multi-scale driving model was used to predict the autumn salinization in the study area at different scales.According to the results,the regional salinization level increased in autumn,in which the non-salinization and mild salinization areas decreased by 17.3%,and the moderate Salinization Soil and severe Salinization Soil increased by 10.3%and 3.45%respectively.As shown by the prediction result of cultivated land,the salinization in the East was obviously aggravated,and the area increased in a north-south strip.According to the profile prediction results,the soil salt content increased with the fluctuation of water level and reached the peak in August.Based on the prediction results,the early warning for regional land ecological use and crop yield was realized.As shown by the results,the early warning level of soil salinization in the northeast of the region was high,and the distribution range of each early warning level shifted and expanded from east to west.The early warning results of crop yield impact reveal that the red early warning area of cotton in autumn was small,the red early warning area of winter wheat was medium,and the red early warning area of corn was the widest.Through the systematic and integrated study of soil salinization in the coastal area of the Yellow River Delta,this paper better understands the multi-scale variation characteristics and driving mechanism of soil salinization,completes the detailed description of soil water and salt transport process at the profile scale,better reveals the control mechanism of soil water and salt dynamics under the influence of different driving forces in the soil,and provides a scientific basis for water and salt regulation and saline soil utilization management in the coastal saline area.