Extraction Of Water-body Information In Urban Areas Based On Sentinel-2 SWIR Bands Sharpening

In recent years,as urban expansion and urbanisation continue to accelerate,the number and shape of surface water in urban areas are changing rapidly,and as one of the effective methods to extract water information about urban,remote sensing water index has been widely used.However,the fragmented distribution of features in the city,especially shadows,makes it difficult to extract water information using traditional remote sensing water indices.In this study,the cities of Nanjing and Fuzhou in China and the cities of Venice and Verona in Italy and their surrounding areas are used as the study area.The Landsat-8 and Sentinel-2A remote sensing images of the same day are used as the main data sources,and based on literature research,five sharpening algorithms,such as ATWT,Gram-Schmidt,HCS,HPF and NNDiffusion,which are more advanced in the existing literature,are applied to the 20 m spatial resolution SWIR band of Sentinel-2 data to obtain a SWIR band with higher spatial resolution(10 m).On this basis,by introducing LST data obtained based on Landsat-8 data and the HUTS spatial downscaling algorithm,a new water index,LAWI(LST Adjusted Water Index),that combines LST information is proposed.And using Landsat-8 and Sentinel-2A data,LAWI is compared,analysed and discussed with NDWI,MNDWI,AWEI.Qualitative and quantitative analyses in four different study areas revealed that the Gram-Schmidt and NNDiffusion algorithms provide the best quality SWIR sharpened images and the highest extraction accuracy in MNDWI.In particular,the NNDiffusion algorithm provides the best sharpening results.The PAN band selected using the correlation coefficient maximisation method provides a more stable sharpening result across the different sharpening algorithms and facilitates comparison between the different sharpening results.In addition,the NDWI,MNDWI and AWEI differ to different degrees in the water index images extracted based on Landsat-8 and Sentinel-2A images respectively,and the segmentation results of NDWI and MNDWI based on the ratio method can effectively reduce the appearance of anomalies.The LAWI proposed in this study can eliminate the extracted differences in the two different images to a greater extent and obtain water images with a higher degree of similarity,while effectively suppressing the shadows in the city,thus improving the extraction accuracy of water information.Compared with Landsat-8 data,the sharpened Sentinel-2A data is more suitable for urban water information extraction and has promising applications in the study of water and other geographical elements.The LAWI created in this study in combination with LST data provides a new approach to remote sensing extraction of water information in urban areas and suggests new ideas for the application of thematic index models across data sources.With the successful launch of Landsat-9,the number of Landsat and Sentinel-2 data available for joint applications on the same day will be greatly increased in the future.The adoption of the new water index proposed in this study can realise the complementary advantages of multi-source data and further improve its potential for integrated applications,and can also providea a useful idea for the construction and improvement of other types of remote sensing thematic indices.