Research On Dynamic Monitoring Method Of Actual Irrigation Area Based On Multi-Source Remote Sensing Data

China is not only a large grain producer,but also a large grain consumer.Agriculture occupies an important position in China's national economy.However,in recent years,the rapid development of urbanization and industrialization in China has caused the continuous reduction of agricultural water availability,and food production safety and optimal allocation of water resources are facing huge challenges.Statistics from the Water Resources Bulletin for many years found that China's agricultural water consumption accounts for about 60%of total water consumption,and irrigation water accounts for about 90%of agricultural water consumption.However,the current agricultural water resource monitoring methods based on ground stations in China cannot fully achieve the goals of agricultural water conservancy informatization,nor can it be completely supporting the development of fine management of agricultural water resources.Compared with traditional ground monitoring methods,remote sensing has the characteristics of wide coverage,strong timeliness,objective data and high efficiency.This paper aims to improve the existing agricultural irrigation monitoring methods,and develops a dynamic monitoring method based on the integration of multi-source remote sensing data with actual irrigation area,and has achieved the following main results:Firstly,in order to achieve a certain precision of remote sensing monitoring of the actual irrigation area,it is necessary to quantitatively analyze the requirements of the actual irrigation area monitoring for the temporal and spatial resolution of optical remote sensing data and the accuracy of soil moisture inversion.Through the theoretical derivation of the remote sensing monitoring error function of the actual irrigated area from a single data source,it is obtained that the remote sensing monitoring error of the actual irrigated area is a function of the remote sensing data acquisition time interval and the accuracy of soil moisture inversion.The temporal and spatial coverage capabilities and soil moisture inversion accuracy of different optical remote sensing data sources of GF-1,HJ-1A/B,Landsat8,and MODIS were statistically evaluated,and compared with the theoretical deduction results of the allowable error of actual irrigation area.None of the remote sensing data sources can meet the demand for the accuracy of dynamic monitoring of the actual irrigation area,and the actual irrigation area monitoring research based on multi-source data fusion is needed.Secondly,in order to solve the problem of quantitative evaluation of error transmission in the process of multi-source data fusion actual irrigation area monitoring,the research is based on the principle of error transmission,and the theoretical derivation of the actual irrigation area monitoring error of multi-source data fusion is carried out.The function expression of the actual irrigation area monitoring error,the main error sources include the soil moisture error of a single data source,and the matching error of the corresponding relationship between the two data sources in the fusion process.The study simulates the monitoring error of the fusion soil moisture change under the conditions of different data source combination schemes and different error transmission coefficients.Comparing with the accuracy requirements of actual irrigation area monitoring,the results show that when GF-1 is combined with HJ-1A/1B and Landsat-8,it is difficult to meet the accuracy requirements of actual irrigation area monitoring due to the limitation of the revisit time interval.The combination of GF-1 and MODIS can Meet the accuracy requirements of actual irrigation area monitoring under certain error transmission coefficient conditions.In order to optimize the error transfer coefficient in the actual irrigation area monitoring error function based on multi-source data fusion,and improve the accuracy of the fusion data,the study based on the STARFM and CDSTARFM models proposed a method for simultaneous optimization of the time phase and spectrum correspondence functions.Thirdly,in order to improve the accuracy of soil moisture based on the fusion of multisource remote sensing data,the BPSTARFM fusion method is proposed based on the idea of simultaneous optimization of the time-phase and spectral correspondence functions.After this method is applied to the fusion of multi-source remote sensing data,it significantly improves the fusion result of multi-source soil moisture data and improves the theoretical application accuracy of the fused soil moisture data in actual irrigation area monitoring.The optimization of the spectral correspondence function requires high-precision dynamic crop pattern data as a reference,while the extraction of high-precision dynamic crop pattern data requires multisource data combination and collaboration.In this paper,based on SVM combined with variable fuzzy set method,the high-precision dynamic crop pattern extraction under the condition of multi-source remote sensing data collaboration is realized,and the inconsistency of the crop pattern extraction accuracy of different spatial resolution data is solved.The optimization of the time-corresponding relationship function needs to consider the spectrum change law of different ground features.Therefore,the study introduces the BP neural network,takes the spectral change characteristics of different ground object categories obtained from high spatial resolution remote sensing data as the prior conditions,and uses the advantage of the BP neural network to automatically solve the complex nonlinear mapping relationship between different elements.Finally,the actual irrigation area monitoring research was carried out using the high temporal and spatial resolution soil moisture data fused by BPSTARFM.Through noise reduction,interpolation and irrigation threshold selection of the fusion data,the actual irrigation area monitoring in the 2018 crop growth season in southern Hebei Province was realized.The accuracy of remote sensing monitoring of actual irrigation area is verified based on irrigation records and actual area samples.The verification results show that the fusion of soil moisture data can support the dynamic monitoring of the actual irrigated area and significantly improve the accuracy of a single data source for actual irrigated area monitoring.In summary,this paper analyzes the actual irrigation area monitoring and multi-source remote sensing data fusion principle,derives the actual irrigation area monitoring error function based on multi-source remote sensing data fusion,and quantitatively analyzes the actual irrigation area monitoring demand for remote sensing data.By proposing a BPSTARFM fusion method that optimizes the corresponding functions of time phase and spectrum at the same time,the soil moisture data with more continuous time and higher accuracy is obtained,and it is successfully applied to irrigation information monitoring.In the paper,it provides an effective method for the remote sensing dynamic monitoring of the actual agricultural irrigation area.