Comparative Analysis of Automatic Water Identification Method Based on Multispectral Remote Sensing

摘要

In this paper, from the water spectral reflectance characteristics, I select five multi-spectral remote sensing water information extraction methods, issue comparative analysis and applicability assessment in five typical experiment areas based on Landsat data. I have made a comprehensive comparison and algorithm applicability analysis on the five kinds of methods. The main conclusions are the following: MNDWI and The spectral relationship method between multi-band have the most widely applicability, MNDWI always use in the building shadow and bare land. The spectral relationship method between multi-band has the special effects on the mountain shadow areas. NDVI has an apparent superiority in the dense forest and grassland, the low-cost of single-band threshold make it have more advantages in the great plains areas, compromise and low noise of NDWI can be used in complicated areas.