Analysis Of Multi-source Soil Moisture Fusion And Spatial-Temporal Variation In The Yellow River Basin And Its Surrounding Areas

Soil moisture is important in agriculture,forestry,animal husbandry,fishery,ecological environment and climate change as one of the quantitative expression factors of terrestrial dry and wet changes.Accurate spatio-temporal continuous soil moisture acquisition is a prerequisite for its application in different fields.The traditional ground observation method,remote sensing inversion,and model simulation methods all have different degrees of deficiencies in acquisition;therefore,how to accurately estimate soil moisture data is a serious problem facing current soil moisture research.The fusion of remote sensing soil moisture data from multiple sources provides us with a new idea to accurately estimate soil moisture,which can well fuse the advantages of different soil moisture data and thus produce high quality soil moisture data.To this end,this thesis evaluates Climate Change Initiative Soil Moisture(CCI),Soil Moisture Active and Passive(SMAP),and the fifth generation ECMWF reanalysis-Land(ERA5-Land)soil moisture data with actual measured data at the site as a reference,and on this basis,the Triple-Collocation fusion method(TC)and Wavelet Transform fusion method(WT)are applied to carry out research on multi-source soil moisture data fusion methods,and the fusion results of soil moisture data calculated by applying the optimal fusion method are analyzed for trend,mutation,and period characteristics,and the cross-wavelet and wavelet coherence methods are applied to analyze the mutual response relationships between soil moisture and surface temperature,precipitation,and vegetation leaf area index,and the main conclusions are as follows:(1)CCI soil moisture data have the highest accuracy but insufficient spatial coverage,SMAP soil moisture data are severely biased dry,and ERA5-Land soil moisture data have the best correlation with same-site observations.The soil moisture data were evaluated using the station observation data as the reference.Although the CCI soil moisture data were high compared with the measured data,the spatial distribution pattern in the Yellow River basin and its surrounding areas was the closest to the measured data and precipitation data,and the time series trend was the most similar to the measured data at the stations.The disadvantage is that the spatial coverage is low and there are missing data.The SMAP soil moisture data in the Yellow River basin and its surrounding areas are dry in general,and there is a gap between the data accuracy and CCI soil moisture data,but the correlation between the two and the measured data is not much different.However,ERA5-Land soil moisture data,as model simulation data,performed well in terms of spatial and temporal continuity and coverage,and had the best correlation with the measured data at the site.(2)TC method can well obtain soil moisture data with better quality.The TC method is applied to obtain the weights of three kinds of soil moisture data,and the fusion of multisource soil moisture data is finally realized through weight distribution,and the TC soil moisture fusion results are obtained.The results show that the root mean square error and mean deviation of TC soil moisture data are smaller than those of the pre-integration soil moisture data,have good correlation with the measured sites,and the data quality is closer to the actual values,which are consistent with the precipitation variation characteristics in different agro-climatic zones.However,the spatial correlation of TC soil moisture data with the measured data at the sites is lower than that of ERA5-Land soil moisture data.(3)Compared with the TC fusion method,the WT method has better fusion effect and more accurate soil moisture data estimation.The WT fusion method is applied to achieve the fusion of multi-source remote sensing soil moisture data by decomposition-fusionreconstruction,and the WT soil moisture fusion results are obtained,and the WT soil moisture data are compared with the actual measured site data,soil moisture data before fusion and TC soil moisture data.The results show that the WT soil moisture data has better consistency and similar spatial and temporal variation characteristics with less dispersion value and significantly improved accuracy.The root mean square error,mean deviation and correlation with the measured sites in different agro-climatic zones are better than the prefusion data and TC soil moisture data,and the accuracy comparison is the best,which can better represent the spatial and temporal variability of the study area.(4)Soil moisture in the Yellow River basin and its surrounding areas showed a weak upward trend,with an abrupt change in mid-July 2017,producing periodic oscillations on a128-d scale,with positive resonant changes with precipitation and low vegetation leaf area index,and negative resonant changes with surface air temperature and high index of leaf area.The better quality WT soil moisture fusion results were applied to explore the day-byday trends,abrupt changes,periodic characteristics of soil moisture and the response of soil moisture to different influencing factors in the Yellow River basin and its surrounding areas from June to September 2016-2018 through distance level analysis,M-K abrupt change test,wavelet analysis,cross wavelet,and wavelet coherence analysis.The results showed that the soil moisture in the study area showed a weakly increasing trend,with a decreasingincreasing-decreasing variation characterized by approximately 14 d cycles,which was consistent with the rainy season variation in the study area.Soil moisture in the study area showed a bimodal trend when it increased,with a significant increase in soil moisture around July 2017,producing a sudden change.Soil moisture oscillates periodically at 128 d scale and is centered in mid-July 2017,further demonstrating a significant increase in soil moisture in July 2017.Cross wavelet and wavelet coherence analysis showed that soil moisture has obvious positive covariance cycle and resonance frequency with precipitation,negative covariance cycle and resonance frequency with surface temperature,negative resonance frequency and resonance change cycle with high vegetation leaf area index,and positive resonance frequency and resonance change cycle with low vegetation leaf area index.