Irrigation Area Extraction Based On Temporal Stability Theory

China is a traditional agricultural country in the world,and the shortage of water resources is a major basic situation in China.Nowadays,with the rapid progress of urbanization and industrialization,agricultural water consumption is restricted in some areas,and food security as well as efficient use of water resources are severely constrained.Irrigation,as a major agricultural water user,is an important part of the implementation of the strictest water resource management system to achieve strong supervision of irrigation resources.Rapid and accurate information on actual irrigation on farmland to improve the efficiency of irrigation water use has become an urgent problem for the relevant departments.However,the existing irrigation zoning map has a long history and ground monitoring stations cannot achieve full regional coverage,making it difficult to achieve the goal of fine management of agricultural water use.The powerful timeliness and wide coverage of remote sensing technology bring a new perspective for irrigation area monitoring.This study uses atmospheric data to simulate soil moisture under no irrigation conditions with the help of land surface model,combines the data of multi-source remote sensing soil moisture products before and after downscaling,compares and analyzes the spatial and temporal changes of soil moisture,evaluates the ability of remote sensing products to detect irrigation signals based on the theory of temporal stability,and uses the unsupervised clustering method to extract the area of irrigated areas and validates the results based on this method.The main research contents and results are as follows:(1)The temporal and spatial dynamics of soil moisture before and after downscaling are compared with those simulated by SURFEX.The SURFEX land surface model is used to simulate soil moisture without irrigation,and on this basis,it is compared with the soil moisture time series of remote sensing products before and after downscaling using the Disaggregation Base On Physical And Theoretical Scale Change(DISPATCH)method.The study shows that the soil moisture inversions of the Advanced Microwave Scanning Radiometer(AMSR)and Soil Moisture Active and Passive(SMAP)have highly similar trends,and the two products are more capable of capturing soil moisture trends than European Space Agency Climate Change Initiative(ESA CCI).The difference in soil moisture is increased by the scale transformation,which makes areas with high soil moisture during precipitation and irrigation periods wetter and areas with low soil moisture during non-irrigation periods drier.The difference in soil moisture was not significant in the SURFEX simulations,and there were differences in the direction and magnitude of changes in the wheat irrigation period compared with the remote sensing products.Spatially,the spatial distribution of soil moisture in various remotely sensed soil moisture products as well as SURFEX simulated products is approximately the same,and lower soil moisture exist in south-central Jinan and at the border between Binzhou and Dongying.(2)The ability of each remote sensing soil moisture product to detect irrigation signals is evaluated.Based on the temporal stability theory,both spatial anomalies and temporal anomalies are calculated to identify the causes of soil moisture anomalies through comparison with SURFEX variation trends,and to evaluate the ability of three series of remote sensing soil moisture products to detect irrigation signals.The study shows that the spatial relative differences of soil moisture obtained from the un-descaled products within 3 years are small,among which SMAP shows the strongest spatial anomaly ability during the wheat irrigation period,followed by AMSR,and ESA CCI performs the best only in April and mid-May 2020.After downscaling,each product magnifies the spatial anomaly values during the soil wetting period,and the ability of SMAP and ESA CCI products to express spatial relative differences is greatly improved compared with AMSR products.SMAP with 1 km resolution become the best product in expressing spatial anomalies.Combining the spatial relative difference expression ability and its accuracy,it is determined that the SMAP product with a spatial resolution of 1 km is the best in expressing spatial anomalies.When the scale is not downscaled,the effect of temporal relative differences in the SMAP product expression study area is the best in 2018-2020,followed by ESA CCI,and ESA CCI performs poorly in the irrigation periods except for May and November 2020 when the effect is better than SMAP.however,the temporal anomalies of AMSR and ESA CCI are greater than0 in the dry season,which is not consistent with the real situation.The trend of temporal anomalies do not change significantly after downscaling,and the irrigation period slightly decreases,which may be caused by the downscaling method of the constructed first-order linear equation.Combining the ability of relative temporal difference expression and its accuracy,it is determined that the SMAP product with a spatial resolution of 9 km expresses temporal anomalies best.(3)The irrigated area is extracted and the results are validated.Using the theoretical index of temporal stability and the Pearson correlation coefficient between the remotely sensed soil moisture products and the simulated soil moisture of the SURFEX model as the parameter input for clustering,three clustering methods,K-means,ISODATA and neural network,are used to cluster the most suitable products for detecting irrigation signals.After the validation of both area quantity and spatial distribution,it is obtained that the SMAP products based on 1 km spatial resolution with K-means clustering method with the classification of mean time relative difference and mean spatial relative difference as input parameters are the most effective in identifying irrigation areas with relative errors of 1.4866%,-0.2118% and 2.6411% for 2018-2020,respectively.