Abstract

DURANTE, Martin  and  DI BELLA, C.M.. A MODIS based tool to assess inundation patterns: an example for the Paraná Delta River. RIA. Rev. investig. agropecu. [online]. 2020, vol.46, n.1, pp.96-107. ISSN 1669-2314.

Wetlands are one of the most productive yet altered ecosystems due to human activities across the world. They rely largely on their water flow to provide several ecosystem functions. Then, to develop a land use plan that allows a productive use maintaining the ecological integrity it is critical to understand the flooding patterns. In that sense, the capability of remote sensors to estimate water cover for large areas at detailed spatial and temporal scales can help to develop managerial decision tools. However, the temporal and spatial variation of water components may alter it spectral properties. We studied the capability of different MODIS derived spectral indices to estimate water cover or water presence-absence. The study region was the Paraná Delta River, which is a 2 Mha wetland area. Between all the models evaluated, one based on the spectral index NDWI1 ((Red – SWIR) / (Red + SWIR)) was the most accurate. A NDWI1 = -0.2 threshold allowed to separate those pixels with less than those with more than 60% of water cover with an accuracy of 91%. By this model, we described the flooding patterns of different landscape units of the region during the last 12 years and classified the region according to the impact of ordinary and extraordinary flooding events. We consider this information can help to improve the knowledge about the hydrodynamics, monitor the impact of some activities and develop a more sustainable regional planning.

Keywords : water surface detection; spatial and temporal dynamics; unsupervised classification; water cover changes; normalized difference water index (NDWI).