Avaliação de erros e de efeitos de escala para a rede de drenagem determinada do Modelo Digital de Elevação (MDE) do SRTM

• Ano de defesa: 2013
• Autor(a) principal: Saraiva, Alzira Gabrielle Soares
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraí­ba; BR; Engenharia Cívil e Ambiental; Programa de Pós-Graduação em Engenharia Urbana e Ambiental; UFPB
• Palavras-chave em Português: Rede de drenagem; Mudança de escala; Direção de fluxo; Drainage networks; Change in scale; Flow direction; ENGENHARIAS::ENGENHARIA CIVIL
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/tede/5497

The SRTM-DEM has been widely used for deriving drainage networks and for several different environmental studies. However, sometimes it is not possible to work with the original spatial resolution of the available DEM, mainly when the study area is large owing to the increase in computational cost, requiring the DEM to be resampled to a coarse resolution. According to the method used for deriving coarse-resolution drainage networks, the quality of the result can be quite different. The wrong choice of the method together with the inherent loss of information within this process may result in a river drainage network incoherent relative to an available network considered as correct. For this reason, this research aimed at identifying the errors present in flow paths and in the physical characteristics of four large-scale watersheds (Paraíba (20.000 km2), São Francisco (640.000 km2), Tapajós (500.000 km2) and Uruguai (207.000 km2)) and its major tributaries in both qualitative and quantitative ways. The comparisons between the drainage networks extracted from DEMs with different spatial resolutions and using different methods (DEM resampling, direct upscaling, multi-step upscaling) were carried out relatively to a high-resolution drainage network obtained from processing the original SRTM-DEM for the qualitative analysis, while for the quantitative analysis the drainage network manually digitized over ETM+/ LANDSAT 7 satellite images was used. The flow paths obtained through the methods of DEM resampling showed strong incoherencies mostly when the change in scale was large, and these errors influenced the area and shape of watersheds and also the quality of the derived drainage network, being stronger for the larger watersheds. The drainage networks obtained through the flow directions upscaling were of better quality because they reproduced more closely the high-resolution network. The multi-step upscaling method showed the better performance for three of the four study areas. The quantitative analysis relative to area and mean distance measured between the DEM-derived drainage network and that one obtained from LANDSAT images showed to be coherent to the qualitative analysis, indicating the multi-step upscaling method as the one of better performance, and the resampling procedure as the worst one. The results in terms of watershed areas comparison were coherent to the visual inspection, showing that the upscaling process reproduced quite well the values obtained in the high resolution, both quantitatively and qualitatively. The direct upscaling procedure obtained the best performance while the resampling method resulted in the largest inconsistencies regarding these two aspects, resulting in watershed delimitation totally different from the actual, mostly for Sao Francisco and Tapajos watersheds, in which the change in scale was larger. The influence of the methods used and of the change in scale over the length and sinuosity of river reaches was detected, sometimes resulting that the result was not very well presented. This can be highlighted when the numerical analysis is compared to the qualitative analysis for results obtained with the resampling procedure. However, although inconsistencies were presented for some river reaches, the upscaling procedure was able to have the largest number of river reaches with the best results. It can be concluded that independent on the characteristics of the watershed to be studied, the resampling procedure is not an adequate method for obtaining coarse-resolution drainage networks. The flow directions upscaling procedure is the most recommend method to be used in this situation, and the multi-step version of this method is the one with best performance when evaluating the flow paths for the study cases of this research.