No abstract

Um mapeamento da cobertura florestal e dos diferentes usos da terra proporciona informações fundamentais para a gestão territorial visando ao desenvolvimento social e econômico, planejamento e controle ambiental e proteção dos recursos naturais. Neste artigo, é apresentado um novo mapeamento, valendo-se de sinergias entre os dados de campo do Inventário Florístico Florestal de Santa Catarina (IFFSC) e o uso de recursos de sensoriamento remoto. Imagens do satélite Landsat-8 OLI do ano de 2017 foram classificadas, utilizando o algoritmo Random Forest. A legenda é composta por 12 classes temáticas; a área mínima mapeada é de 0,5 hectare. O mapa tem acurácia geral de 95%, com intervalo de confiança de 1,0% (alfa=0,05). A acurácia média por classe varia entre 90% (agricultura) e 97% (restinga).  Para a classe floresta, o mapa apresentou coincidência de 96,2% com os pontos amostrais do IFFSC. A cobertura florestal nativa (florestas a partir do estágio médio de regeneração) está presente em 38,05% do território, reflorestamentos em 10,46%, agricultura em 16,73% (incluídos 1,77% de culturas de arroz irrigado), pastagens e campos naturais em 29,24%. A área da extensão original da restinga foi determinada em 1.773km², dos quais 814,5km² (ou 45,9%) cobertos por remanescentes naturais, praias e dunas. O mapeamento constitui a base para a tomada de decisão de agentes públicos envolvidos em atividades de planejamento e gestão territorial e servirá como linha-base para o monitoramento contínuo da extensão da cobertura florestal do estado.

The mapping of the use of the land in a certain region allows theidentification of the available resources in it. In that way, the areaof remote sensoring searches for ways of processing large amountsof data with reliability in its results. This paper compares the automatedprocess of image, along with a spacial filter, with the imageedited manually. The overall accuracy for each categorized imageresulted in 71,3% for simple category, 72,6% for filtered categoryand 94,5% for manually edited category.

In natural regeneration, vegetation goes through different stages over time, each of them with different conditions to drain the precipitated water to the soil surface. The aim of this study was to evaluate interception loss, throughfall, stemflow and surface runoff in early and advanced stages regeneration at evergreen rainforest subtype sites within the Atlantic Forest, located in the Serra do Itajaí National Park, in southern Brazil. Rainfall was sampled by three rain gauges and throughfall was measured using "U" type gutters. The stemflow was measured in 24 trees per stage (i.e., 48 trees in total). The interception loss was calculated as the difference between rainfall and the sum of the throughfall plus stemflow. The surface runoff was evaluated using metal rails. Vegetation in the initial stage is a composition of 7 species with a basal area of 5.01 m².ha-1 and advanced stage site consists of 28 species with a basal area of 34.7 m².ha-1. Throughfall records were 81.7% in the initial and 74.1% in the advanced regeneration stages. The stemflows registered were 5.93% in the initial stage and 0.54% in the advanced stage while the interception losses were respectively 13.2% and 25.8%. Surface runoffs observations were 11.1% in the initial stage and 10.7% in the advanced stage. Statistically significant differences were found between the regeneration stages for the stemflow and for the interception loss parameters. The study showed that, for some hydrological processes, the behavior of the precipitated water differs for the stage of vegetation regeneration. The regeneration stage does not influence the surface runoff, demonstrating that after a few years of vegetation regeneration, this hydrological process is equivalent to vegetation in an advanced stage.

Backdating de pixels invariantes para detecção de mudanças de uso e cobertura da terra na Mata Atlântica subtropical no Brasil: uma comparação de algoritmos