Spatial Patterns Of The Crop Waterlogging Risk Areas Under The Impact Of Climate Change In The Middle And Lower Reaches Of Bayin River

Waterlogging is the damage to plants caused by water stress when soil water in the root zone of crops exceeding the maximum water capacity in the field,and is one of the main disasters affecting agricultural production.There are many factors that could cause crop waterlogging,such as climate change,floods and rising of groundwater level.However,due to the complexity of the hydrological and the related processes and the limitation of research methods,the mechanism of crop waterlogging caused by the rise of regional groundwater level is still unclear.The large area of land in the middle and lower reaches of Bayin River is an important food production base in Qaidam Basin.In recent years,under the impacts of climate warming and humidification,the groundwater level in the area rised obviously,and large areas of farmland has been submerged.That caused crop waterlogging.A large number of studies have focused on the response of crop growth status and yield to waterlogging at individual plant and field scale.Waterlogging,especially the damage caused by the rise of regional groundwater level,is related to the water cycle process in the basin.However,related studies are rare.Hydrological model is an important tool for simulating large regional scale hydrological processes.SWAT(Soil and Water Assessment Tool),a semi-distributed hydrological model,could predict the effects of different land use/cover,management practices,crop growth and climate on regional hydrological processes,and has satisfactory performance in arid inland river basins in China.But SWAT has weak capability to simulate groundwater processes.In addition,the crop growth module only considered the crop water stress caused by soil water shortage,but did not consider the water stress caused by excessive soil water.In this study,the crop species in the middle and lower reaches of Bayin River were extracted by using multi-temporal images with high spatial and temporal resolution combined with the vegetation index data set of time series.The SWAT model was coupled with a specialized groundwater model,MODFLOW(Modular Finite Difference Groundwater Flow Model),to accurately simulate regional hydrological processes.Moreover,the crop growth module of the SWAT-MODFLOW coupling model was improved,and the WLSWAT-MODFLOW model with the ability of identifying the waterlogging damage of crops was developed.We employed the remote sensing based LAI products,spring wheat yield data,high spatio-temporal resolution evapotranspiration data and point-scale groundwater level data to calibrate the WLSWAT-MODFLOW model,and R~2,NSE and PBIAS values were used to evaluate the the model performance.Finally,the observed climate data in 2001-2020and the climate change scenarios of SSP1-2.6,SSP2-4.5 and SSP5-8.5 in 2021-2100simulated by BCC-CSM2-MR model under CMIP6(Coupled Model Intercomparison Project Phase 6)were input to the WLSWAT-MODFLOW model to identify and predict the crop waterlogging risk areas,separately.The result showed that:(1)Based on Sentinel-2 images,Support Vector Machine(SVM),Artificial Neural Network(ANN)and Random Forest(RF)method were used to extract crop information in the middle and lower reaches of Bayin River irrigated areas,among which random forest method had the highest classification accuracy.The overall accuracy of Delingha irrigated area was 95.63%,Kappa coefficient was 0.95,and the overall accuracy of Gahai irrigated area was 94.75%,Kappa coefficient was 0.94.(2)WLSWAT-MODFLOW model had performance in mountain runoff,leaf area index,evapotranspiration,spring wheat yield and groundwater level in the middle and lower reaches of Bayin River.R~2and NSE values of mountain runoff simulation reached 0.84,and PBIAS value was 3.5%.The measured values of LAI were close to the simulated values.R~2values of simulated ET in all sub-basins were above 0.72,NSE values were all above 0.74,and the PBIAS values were all between-15%-15%except for a few subbasins.R~2of the simulated groundwater level in each well was above 0.91,and the error was within 0.5 m.The average error between actual spring wheat yield and simulated spring wheat yield was 4.73 kg/hm~2.(3)In the next 80 years,the precipitation of Bayin River Basin showed an increasing trend under different scenarios.The temperature showed a significant increasing trend over the next 80 years under SSP2-4.5 and SSP5-8.5 scenarios.In the next 80 years,the mountain runoff from the upper reaches of Bayin River under SSP1-2.6 scenario was substantially higher than in other scenarios after 2041.Under the SSP2-4.5,the discharge of mountain runoff decreased significantly.Under the SSP5-8.5 scenario,the mountain runoff showed initial decrease and then an increasing trend.(4)During the historical period from 2001 to 2020,the crop waterlogging risk area was 10.9 km~2.In the next 80 years,the maximum area of waterlogging risk area appeared in 2055 under SSP1-2.6 scenario,and the minimum area of waterlogging risk area appeared in 2042 under SSP2-4.5 scenario,respectively,with the maximum value of 11.52 km~2and the minimum value of 9.49 km~2.Under different scenarios,the area of waterlogging risk area in the middle and lower reaches of Bayin River were mainly affected by the runoff from the upper reaches(mountainous area).