Remote Sensing Area Change Detection Of Typical Reservoirs In Beijing-Tianjin-Hebei Region During 2014-2019

Resources,as one of the carriers of surface water,have the functions of water supply,flood control and power generation.As an artificial water conservancy facility,the area change of reservoirs can not only reflect the change of regional ecological environment,but also the change of human activities.The change detection of reservoirs can not only provide decision-making basis for regional flood and drought control,but also provide important guidance for regional sustainable development.Based on the 361 GF-1 WFV images,six typical large reservoirs in Beijing-Tianjin-Hebei region,Miyun reservoir,Guanting reservoir,Yuecheng Reservoir,Gangnan Reservoir,Huangbizhuang reservoir and Panjiakou pivotal reservoir,were detected monthly.NDWI adaptive threshold method is used for water extraction;For reservoir with water bloom,NDVI adaptive threshold method and support vector machine supervision classification method are used to extract the water bloom range;For frozen reservoir,blue,green,red,near-infrared four bands adaptive threshold method and support vector machine supervision classification are used to extract the frozen region;For the reservoirs which are confused with the mountain shadow,the DEM slope is used to segment the threshold value,and the mountain shadow region is obtained to eliminate the mountain shadow.Finally,the areas of the reservoirs are obtained.This paper establishes modlels about the areas of the reservoirs and the water level data during 2014-2019.The conclusions are as follows:（1）NDWI adaptive threshold method can achieve efficient extraction of water,which has a good effect on distinguishing vegetation and water,and is suitable for the extraction of reservoirs in spring and summer;NDVI adaptive threshold method and support vector machine supervised classification method can achieve effective extraction of reservoirs where water bloom occurs in summer;sum of four bands adaptive threshold method and supervised classification method can effectively extract the frozen reservoirs,and the slope data obtained by DEM can effectively eliminate the mountain shadow of the reservoirs.（2）The area change trend of Guanting reservoir,Gangnan Reservoir and Huangbizhuang Reservoir in 2014-2019 is similar,while the area change of other reservoirs in 2014-2019 have no obvious regularity.The area of Guanting reservoir and Miyun Reservoir increased greatly from 2014 to 2019,the area of Huangbizhuang Reservoir increased slightly,the area of Gangnan Reservoir and Panjiakou Reservoir decreased slightly,and the area of Yuecheng Reservoir decreased greatly.（3）There is a significant correlation between the area and water level data of each reservoirs.The correlation coefficients of the linear,exponential,logarithmic and power models for each reservoirs are high,and the accuracies of the models are high.The best model between the area and water level data of Miyun reservoir is linear model:y=0.209x+119.310（R²=0.993）,the best model between the area and water level data of Guanting Reservoir is logarithmic model y=0.664ln（x）+44.596（R²=0.921）,the best model between the area and water level data of Yuecheng Reservoir is linear model y=0.721x+117.390（R²=0.960）,the best model between the area and water level data of Gangnan Reservoir is linear model y=0.588x+170.570（R²=0.970）,the best model between the area and water level data of Huangbizhuang reservoir is power model y=88.778x^0.073（R²=0.972）,the best model between the area and water level data of Panjiakou reservoir is power model y=94.566x^0.205（R²=0.875）,and the best models of each reservoirs can realize the estimation of reservoir areas by using water level data,providing decision basis for regional disaster assessment.