River Discharge Derived From Sentinel-1 SAR Data

Surface water resources is one of the most important natural resources and occupy an important position in economic development and ecosystem protection.In recent years,with the global change of water resources the problem of spatiotemporal distribution inequality is getting increasingly serious,the international community has paid more and more attention to the problem of surface water monitoring.River discharge is an important parameter to represent the reserves of surface water resources.Conventional discharge data is obtained through hydrometers or ADCP measurements at a hydrological observation site.However,due to economic,technical,and political constraints,many regions still lack long-range and large-scale measured data.The decreasing number of survey sites and the uneven distribution of site space further restrict the discharge data of some important hydrological units.With the development of space information technology,remote sensing technology has the advantages of low measurement cost and wide coverage,which can provide a new and effective method for discharge data monitoring.The current remote sensing parameters used in the discharge inversion model mainly include the average river width,water surface area and river island area.However,these models are usually based on the parameters of a single reach without considering the hydrological or hydraulic relationships between upstream and downstream regions.Therefore,the accuracy of river discharge inversion in complex hydrological conditions is poor.And since the reaches used to calculate the average river width or area are mostly randomly selected,there is no uniform selection criterion.In view of the above problems,this paper is mainly about studying the inversion method of discharge considering the relationship between upstream and downstream hydrology and analyzing the suitability of the reaches for discharge inversion.The main conclusions and results obtained in this study include:(1)Based on the in situ data,the measured water level and discharge relationships at the five major hydrological stations along the Gan River Basin were analyzed,and it was found that the measured water level and discharge relationship at the Waizhou Station was unstable.By comparing the measured during flood season and dry season,it was found that,in flood season when the water level of Poyang Lake is high,potential backwater effect may be the main causes of the instability of the measured water level and discharge relationship at the site.Therefore,in combination with the measured water level data of the downstream Nanchang station,multiple regression was performed during the flood season,and the RMSE decreased from 1735.273m3/s to 721.835m3/s,and the accuracy increased by 58.4%.(2)Based on the Sentinel-1 SAR image,the average river width of some reaches was extracted,and a one-variable regression analysis was performed with the measured discharge data of the corresponding hydrological station.Based on the monitoring data of average river width and area greater than 2000km2 of Poyang Lake,multivariate regression analysis was performed for the discharge at the Waizhou Station.The results show that the accuracy of the multivariate regression model combined with upstream and downstream hydrological information is better than that of the one-variable model.Therefore,from the perspective of remote sensing monitoring,it is proved that there is a backwater effect at the Waizhou station when the downstream water level is high.(3)The impact on discharge monitoring suitability of average slope in the maximum surface area,the average slope in the 90m buffer of maximum surface area,the average slope in the 180m buffer of maximum surface area,the average river width in the fluctuation area and the area ratio of the fluctuation area was analyzed.The research shows that the average slope has a certain trend influence on the modeling effect.The bigger the slope is,the worse the modeling effect.The average river width in the fluctuation area has a positive correlation with the modeling effect and with a larger average river width in the fluctuation area,the better the effect.A strong linear positive correlation was shown between the area ratio of the fluctuation area in the modeled reaches and the modelling effect.And the area ratio of the fluctuation area has a great influence on the modeling effect,the R2 is 0.9182,the larger the area ratio of the fluctuation area is,the better the modeling effect.