Comparative analysis of normalized difference water index (NDWI) methods in continental wetland

Abstract

The remote sensing data, like radiometric indices, represent alternative data generation sources that can help us understand the natural processes that occur in the Pantanal. The Normalized Difference Water Index (NDWI) is a partially known method in the scientific community which makes geotecnologies use. Although, its use and application still causes doubts and confusions, mainly because there are two different methods for calculating NDWI, both with different objectives. In this sense, it was analyzed which NDWI method has better performance to represent the Pantanal wet and flooded areas. The indices were calculated 4x, initially in the McFeeteres method (1996), followed by 3 repetitions in the Gao (1996) method. The repetition for the Gao method was due to the medium infrared (SWIR) band availability of the MODIS product used. The results comparison evidenced that using the McFeeters NDWI method generated greater spectral confusion in wet or exposed soil and in undergrowth targets, whereas water and tree vegetation targets were easier characterized. In the Gao NDWI method results, an improved relationship between near infrared (NIR) and SWIR was noticed, due to band 2 (NIR) and 5 (SWIR), what, in this situation, revealed that each target had a more focused spectral behavior at certain intervals. In bands 6 (SWIR) and 7 (SWIR) the spectral confusion was considered high, mainly in moist soil and water targets.