Detection Of Inland Water Bodies Based On Spaceborne GNSS-R

Inland water bodies are an important part of the global water cycle.Rapid and accurate acquisition of information on the spatial distribution of inland water bodies over time is important for understanding the interaction mechanism between regional hydrology and climate change,investigation and planning of water resources,and prevention of flood disasters.The spaceborne GNSS-R(Global Navigation Satellite System-Reflectometry)is an emerging remote sensing technology tool.It has the advantages of low cost,high spatial and temporal resolution,and large coverage,which can make up for the shortage of existing remote sensing technology in detecting inland water bodies.The aim of this paper is to investigate the feasibility of spaceborne GNSS-R for detecting inland water bodies,which can be divided into mapping inland permanent water bodies and monitoring the time-series changes of inland floods.The main work of the paper is carried out as follows:(1)An improved coherence method is proposed to detect inland water bodies.Based on the existing coherence detection method DPSD(DDM Power-Spread Detector),we analyze the causes of its false detection phenomenon in inland water body detection and propose an improved coherence method.The method distinguishes land and water bodies by calculating the power ratio between the peak area of DDM(delay-Doppler Map)and the horseshoe-shaped area to establish the connection with the water body.The improved method does not require thermal noise removal by an empirical function.The method proposed in this paper reduces the uncertainty in the detection results brought by the DPSD method due to the use of empirical functions.(2)Mapping inland permanent water bodies using coherence methods.Use DPSD and improved methods to map permanent water bodies in the Amazon Basin and Congo Basin.The generated water body masks are compared with the available optical water body data GSW(Global Surface Water)to calculate the accuracy.The overall detection accuracy of the improved coherence method in the Amazon Basin is 94.48% compared to the GSW data.Among them,the detection accuracy of water bodies is 92.23%.The corresponding accuracies in the Congo Basin were 96.12% and 93.16%,respectively.Compared with the DPSD method,the false detection rate and the missed detection rate of the improved method in the Amazon Basin were reduced by 17.1% and 9.1%,respectively.The false and missed detection rates in the Congo basin were reduced by 10.2% and 22.0%,respectively.(3)An improved method is used to monitor the temporal variation of flooding in inland transient water bodies.Taking the "7.20 Henan rainstorm" as an example,a 0.05° inundation area map was generated every three days using spaceborne GNSS-R to monitor the temporal variation of water bodies at the time of flooding since the onset of the extreme rainfall event.The generated inundation area maps are consistent with the rainfall data,SMAP(Soil Moisture Active/Passive)areas of soil moisture change.A 0.01° inundation area map was generated using data from about 15 days before and after the rainfall to detect all the affected areas.The obtained water body areas are basically consistent with the water bodies recorded by MODIS(Moderateresolution Imaging Spectroradiometer)optical images and SAR(Synthetic Aperture Radar)images in terms of location and shape.