Analysis of segmentation techniques for improved geomorphometric data from open source tools

The objective of this paper is to demonstrate the practicality of using segmenting images for obtaining geomorphometric data for the environmental management of watershade. This study was conducted on the Bandeirinha Stream basin (BHCB), located in Goias State in the municipality of Formosa-GO. Shuttle Radar Topography Mission (SRTM) was used, with data 1" x 1" (one second of arc) or 30m spatial resolution. The ODR_Obia algorithm was applied for SRTM segmentation in order to separate the classes and identify the research targets. This algorithm was adapted from the TA Baatz Segmenter using the QGIS open source platform. The results from the segmentation and extraction of pre-defined indices resulted in a new segmented raster image. To obtain geomorphometric data the ODR_Hidro model was used, acccessed via the QGIS platform, developed in Python language. The mathematical calculations, along with its adaptation, were compiled within the ODR_Hidro model through the use of intermediary hydrologic algorithms used in softwares such as QGIS, GrassGis and SagaGIS, which extract geomorphometric indices, the delineation of the basin, and the extraction of drainage channel networks. The results were evaluated by testing the hypothesis in the Mann-Whitney statistic test to check if the average discrepencies of the coordinates for E and N bifurcation points were equal or similar to those produced by the ODR_Hidro model, for both the SRTM segmented image and the original SRTM. The Mann-Whitney test was applied to six samples, four of which accepted the hypothesis and only two rejected. The results suggest that applying the ODR_Hidro model for characterizing and defining geomorphometric indices using SRTM image segmentation is quite efficient as the field validation showed the effectiveness the model has for proving the existence of drainage in the 1st order which was unmapped in the DSG card. We can thus conclude that image segmentation for geomorphometric characterization of watershade will be able to efficiently contribute towards providing morphometric data.