João Maria de Andrade , John Cunha, João Nailson de Castro Silva, Iana Alexandra Alves Rufino, Carlos de Oliveira Galvão

Accurate information on the land cover is crucial for efficient monitoring and development of environmental studies in the Brazilian Caatinga forest. It is the largest tropical seasonal forest in South America, presenting high biodiversity and is under intense anthropogenic disturbance. Caatingas land cover is heterogeneous, and rainfall is its primary phenological regulator, presenting mainly deciduous species. Different land-cover patterns show distinct spatial responses to climate and soils changes and modify their physical properties over time. Rainfall is highly variable over time and space, but seasonally concentrated between 2 to 4 months. Therefore, distinguishing the different patterns of land cover through medium spatial-resolution remote sensing, such as the Landsat image series, is challenging, due to the particularities of the climate-vegetation interaction. Two remote sensing approaches have a high potential for efficient land-cover mapping in Caatinga: single and multi-date imagery. The heterogeneity of the land cover of this environment can contribute to a better performance of multispectral approaches, although it is normally applied for single-date images. In a land-cover mapping effort in Caatinga, the temporal factor gains relevance, and the use of time series can bring advantages, but, in general, this approach uses vegetation index, losing multispectral information. This manuscript aims to assess the accuracies and advantages of single-date multispectral and multi-date Normalized Difference Vegetation Index (NDVI) approaches in land-cover classification. Both approaches use the Random Forest method, and the results are evaluated based on samples collected during field surveys. Results indicate that land-cover classification obtained from multi-date NDVI performs better than single-date multispectral data. The lower performance observed for single-date multispectral classification is due to similarities in spectral responses: targets of deciduous vegetation lose their foliage and can be misread as non-vegetated areas. Meanwhile, an accurate classification by time series of plant clusters in seasonal forests allows incorporating seasonal variability of land-cover classes during the rainy and dry seasons, as well as transitions between seasons.