No abstract

Artigo recebido para revisão do evento em 22/04/2014, aceito para publicação em 11/05/2014 e recebido para publicação em 01/06/2014Resumo A exploração de recursos naturais é um dos grandes agentes modificadores de paisagens e quando realizadas de forma muito intensa ou inadequadas, modificam de forma substancial o local e seu entorno, fazendo com que muitas vezes a restauração de áreas exploradas se torne praticamente impossível e/ou muito onerosa, sendo assim, para que se reestabeleça um novo equilíbrio ao ambiente afetado, é preciso que haja uma série de ações voltadas à recuperação da qualidade ambiental do meio. O presente trabalho tem como objetivo aplicar técnicas de manejo do solo e de plantio de espécies de leguminosas, com vistas a recuperação da área degradada além de avaliar a eficácia destes no processo de recuperação a partir de parâmetros como o tempo de desenvolvimento da vegetação e as alterações nas condições físico/químicas do solo ao longo do processo. Palavras-chave: Processos erosivos; Leguminosas, áreas degradadas AbstractThe exploitation of natural resources is a major modifiers of landscapes and when performed in a very intense way or inadequate, substantially modify the site and its surroundings, causing often the restoration of mined areas practically impossible and / or very expensive, so in order to reestablish that a new balance to the affected environment, there must be a series of actions to restore environmental quality of the environment. This paper aims to apply techniques of soil management and planting of leguminous species, intending to recover the degraded area and to evaluate the effectiveness of the recovery process using parameters such as the time of development of vegetation and changes in physical and chemical conditions of the soil throughout the process.

Different uses of soil legacy data such as training dataset as well as the selection of soil environmental covariables could drive the accuracy of machine learning techniques. Thus, this study evaluated the ability of the Random Forest algorithm to predict soil classes from different training datasets and extrapolate such information to a similar area. The following training datasets were extracted from legacy data: a) point data composed of 53 soil samples; b) 30 m buffer around the soil samples, and soil map polygons excluding: c) 20 m; and d) 30 m from the boundaries of polygons. These four datasets were submitted to principal component analysis (PCA) to reduce multidimensionality. Each dataset derived a new one. Different combinations of predictor variables were tested. A total of 52 models were evaluated by means of error of models, prediction uncertainty and external validation for overall accuracy and Kappa index. The best result was obtained by reducing the number of predictors with the PCA along with information from the buffer around the points. Although Random Forest has been considered a robust spatial predictor model, it was clear it is sensitive to different strategies of selecting training dataset. Effort was necessary to find the best training dataset for achieving a suitable level of accuracy of spatial prediction. To identify a specific dataset seems to be better than using a great number of variables or a large volume of training data. The efforts made allowed for the accurate acquisition of a mapped area 15.5 times larger than the reference area.