Comparação entre bases de dados para mapeamento e modelagem em bacias hidrográficas

Abstract

The elaboration of maps related to urban and environmental planning, such as director plans, zoning agricultural, risk areas, etc., generally follow modeling techniques. For the development of these models are required use of databases and different methods applications. However, a significant portion of available databases do not have adequate quality but, as are the enabling be acquired involuntarily leads the professional to use them. In this sense, aiming to verify the influence of database and methods in modeling underwent the following procedures: i) compare quantitatively the use and land cover between two medium and high spatial resolutions maps; ii) compare quantitatively soil loss areas from six topographic factor scenarios (LS), generated from two digital elevation models (DEM) of medium and high spatial resolution and two methods of flow direction (simple and compound); and iii) compare quantitatively and qualitatively areas of soil loss - especially those who are more vulnerable - from twelve scenarios universal soil loss equation (USLE) modeling. The study areas include João Leite river and Bandeira stream. To compare the maps of land cover were used two orbital images: Landsat-8 OLI 2014 sensor (30 m spatial resolution) and Theos-2011 sensor (2m spatial resolution). These two orbital images originated three land cover mappings: Theos/2011 (not prepared by the author), Landsat/2014 and Theos/Atualizado-2014 (generated updating the mapping Theos/2011 from the visual interpretation of high resolution spatial images available on Google Earth software). In relation to USLE, the R factor (erosivity) was obtained from records of 5 rainfall stations; the K factor (erodibility) was obtained by values determined in the literature; were generated two scenarios of CP factor (cover and management and support practice) using the Landsat/2014 and Theos/Atualizado-2014 mappings, which was adopted soil loss values determined in the literature; and to the LS factor (topographic) were generated six scenarios, using the DEM Topodata (~30m spatial resolution) and DEM2006 (~4m spatial resolution); and the direction of methods of D8 flow (single) and D∞(compound), for the two watersheds. When comparing land cover maps Theos/Atualizado-2014 and Landsat/2014, it was found due to the difference in spatial resolution and classification method, Landsat/2014 overestimated vegetation classes and underestimated pasture, agriculture and forestry classes. As for the modeling of topographic factor (LS) was observed that the scenarios composed by DEM Topodata and D8 flow direction method was inefficient at modeling the runoff, and underestimate the most vulnerable classes. The scenarios originated by DEM DEM2006 and D∞ method allowed the best modeling, considering the flow as much as convergent divergent. In relation to soil loss modeling (USLE), different databases and methods employed in the LS and CP factors influenced the identification of most vulnerable areas of the twelve generated scenarios. The scenarios composed by Topodata and D8 method DEM beyond underestimate the risk areas classified as Strong to Very Strong up to 95%, made it possible to identify the exact location of these areas. The scenarios composed by Topodata DEM and D∞ method had a better performance than DEM Topodata and D8 method by allowing the location of the region where the most vulnerable areas were located. However, the setting allowing better detail relief as well as the accuracy of the identification of the most vulnerable areas was composed of the LS factor DEM DEM2006 D∞ and method, and CP Theos factor. It was found that, for the viability of land cover mapping and modeling through USLE in identifying the most vulnerable areas applications of robust methods and high resolution spatial database are required.