Utilização de câmeras hiperespectrais para caracterização fenológica de dosséis florestais: um estudo de caso na Floresta Nacional do Tapajós Pará
We investigate how canopy spectral signals are linked with phenological processes during the 2012 dry season in the Tapajós National Forest, Pará, Brazil. We disentangled the spectral components of canopy phenology processes and analyzed their variations over time using six images acquired by a tower-mounted hyperspectral camera. We applied two techniques over the data (spectral mixture analysis - SMA and principal component analysis - PCA) and calculated two vegetation indices (Enhanced Vegetation Index - EVI and Normalized Difference Vegetation Index - NDVI). The results showed that the differences in timing and intensity of leaf flushing and leaf shedding between the species produced different green vegetation (GV) and non-photosynthetic vegetation (NPV) SMA-fractions. Differences in canopy phenology between the species were also confirmed by PCA. Temporal variations in the described patterns of canopy phenology were detected by the EVI and NDVI, presenting statistical differences (0.05 significance level) between the species and across the dates. Canopy phenology patterns of leaf flushing increased the EVI values toward the end of time series due to the observed changes in the NIR reflectance from mature to young leaves. NDVI was more sensible to canopy phenology patterns of leaf shedding, decreasing over time due to the reduction in chlorophyll content and the resultant increase in the red reflectance with deciduousness. We conclude that tower-based networks of phenological observations can provide new opportunities for better understand tropical ecosystem phenology as well as for scaling these processes across the landscape with the use of moderate spatial resolution satellite data.