Search

In the era of big data, making full use of remote sensing images to automatically extract surface water bodies (WBs) in complex environments is extremely challenging. Due to the weak capability of existing algorithms in extracting small WBs and WB edge information from remote sensing images, we proposed a new method—Multiscale Fusion SegFormer (MF-SegFormer)—for WB extraction in the Weihe River Basin of China using Landsat 8 OLI images. The MF-SegFormer method adopts a cascading approach to fuse features output by the SegFormer encoder at multiple scales. A feature fusion (FF) module is proposed to enhance the extraction of WB edge information, while an Atrous Spatial Pyramid Pooling (ASPP) module is employed to enhance the extraction of small WBs. Furthermore, we analyzed the impact of four kinds of band combinations on WB extraction by the MF-SegFormer model, including true color composite images, false color images, true color images, and false color images enhanced by Gaussian stretch. We also compared our proposed method with several different approaches. The results suggested that false color composite images enhanced by Gaussian stretching are beneficial for extracting WBs, and the MF-SegFormer model achieves the highest accuracy across the study area with a precision of 77.6%, recall of 84.4%, F1-score of 80.9%, and mean intersection over union (mIoU) of 83.9%. In addition, we used the determination coefficient (R2) and root-mean-square error (RMSE) to evaluate the performance of river width extraction. Our extraction results in an overall R2 of 0.946 and an RMSE of 28.21 m for the mainstream width in the "Xi'an-Xianyang" section of the Weihe River. The proposed MF-SegFormer method used in this study outperformed other methods and was found to be more robust for WB extraction. [ABSTRACT FROM AUTHOR]

The use of remote sensing to monitor surface water bodies has gradually matured. Long-term serial water change analysis and floods monitoring are currently research hotspots of remote sensing hydrology. However, these studies are also faced with some problems, such as coarse temporal or spatial resolution of some remote sensing data. In general, flood monitoring requires high temporal resolution, and small-scale surface water extraction requires high spatial resolution. The machine learning method has been proven to be effective against long-term serial surface water extraction, such as random forests (RFs). MODIS data are well suited for large-scale surface water dynamic analysis and flood monitoring because of its short return cycle and medium spatial resolution. In this paper, the Yangtze River Basin (YRB) in China was selected as the study area, and two MODIS products (MOD09A1 and MOD13Q1) and RF method were used to extract the surface water from 2000 to 2016. Considering the disadvantages of temporal or spatial resolution of these two MODIS products, this study also presents a data fusion method to combine them and get higher spatiotemporal resolution water results. Finally, 762 surface water maps from 2000 to 2016 are obtained, whose temporal and spatial resolution is every eight days and 250 m, respectively. In addition, water extent variation is analyzed and compared to observed precipitation data. The main conclusions are as follows: (1) this constructed approach for long-term serial surface water extraction based on the RF classifier is feasible, and a good fusion method is used to obtain the surface water body with higher spatiotemporal resolution; (2) the maximum area of the surface water extent is 48.53 × 103 km2, and seasonal and permanent water areas are 20.51 × 103 km2 and 28.01 × 103 km2, respectively; (3) surface water area is increasing in the YRB, such that seasonal water area decreased by 3450 km2, and the permanent water area increased by 3565 km2 in 2001–2015; (4) precipitation is the main factor causing variation in the surface water bodies, and they both show an increasing trend in 2000–2016. As such, the approach is worth referring to other remote sensing applications, and these products are very both valuable for water resource management and flood monitoring in the study area. [ABSTRACT FROM AUTHOR]