Inventário e classificação de pequenas áreas úmidas para a gestão e manejo sustentável com uso de sensoriamento remoto

Abstract

This work gave rise to the first inventory of small wetlands in terrestrial matrix of Cerrado in Brazil, considering small wetlands to be smaller than 4 hectares. Wetland inventories are essential, according to the Ramsar Convention, to identify, classify, assess the ecological integrity and monitor them for the conservation and rational use of these ecosystems. Thus, it was proposed in this work to carry out the inventory and classification of small wetlands in the Brazilian Cerrado matrix using remote sensing. A priori, the steps for using remote data in inventory planning were established, which were, (a) define the purpose of the inventory, (b) select the appropriate sensor and (c) choose the remote technologies and software to perform the inventory. Criteria and key questions for selection of Remote Sensing data and software for wetland inventory, assessment and monitoring were tabulated and can be used as instruments for training managers in their management routines and their planning. The inventory, in a case study in the Cerrado matrix in the catchment of the Quilombo River in the municipality of Chapada do Guimarães - MT, identified and delimited 8810 small wetlands occupying a total area of 196.7 km2 within the hydrographic basin with 1669 km2 which corresponds to a density of 5.28 wetlands km−2 . Thus, approximately 12% of the basin area was covered by small wetlands, which had 180 first order, 39 second order, 13 third order and 4 fourth order streams and the Quilombo river as a fifth order river. Regarding size, the majority of the small wetlands (90.8%) were < 4 hectares. Seven types of natural wetlands (paths, Palm forest of Mauritia flexuosa, Wet meadows on hill slopes, Termite mound savanna, Humid meadows of Axonopus purpusii, Riparian wetlands along small rivers and Rainwater-fed wetlands in small depressions) and four anthropic types (weirs, mining pits, stream impoundments by the construction of roads and wetlands on the edge of the Manso dam) were classified, according to the Brazilian Classification System for Wetlands, and characterized. Proving the effectiveness of the sensor and the methodology with GEOBIA in identifying and classifying small wetlands with a Kappa index of 0.968, the features indicative of small AUs were tabulated. For each feature, the type of macrohabitats it has the greatest capacity to recognize was indicated. Thus, six indicator features were recommended for mapping small wetlands in order to reach the macrohabitat level, Asymmetry, Rectangular Fit, Mean NIR Band, Compactness, Density and Length. The spectral difference algorithm applied after the initial image segmentation was essential to establish the real limits and regions with greater or lesser waterlogging. Therefore, the results obtained from this work confirm the possibility of carrying out an inventory of small wetlands with the use of remote sensing through GEOBIA. The inventory of small wetlands carried out will serve as a basis for the identification of small wetlands to be restored, risk and vulnerability assessment. This encourages the elaboration of public policies and scientific studies on these small but highly reputed ecosystems.