Retrieval of solar-induced chlorophyll fluorescence (SIF) from satellites is attractive due to the potential for monitoring and mapping photosynthetic activity, vegetation health, and carbon-fluxes, and is an earlier indicator of vegetation stress than other remote sensing vegetative health indices. The Orbiting Carbon Observatory-2 (OCO-2) has the ability to retrieve SIF from space, which can yield new information about productivity, eventually revealing information beyond a greenness factor about vegetative function under various conditions. Brazil is known for rich biodiversity, high endemism, and high degree of heterogeneity in vegetative cover, due to natural habitat variation and human-driven land-cover change for agriculture, thus providing an ideal environment to test for correlations between SIF and vegetative cover in a multiscale perspective. To better understand ecosystem-scale vegetation productivity in Brazil, OCO-2 data from June 2015 was obtained and subset for the country. The OCO-2 SIF measurements and ecosystem classes from the International Geosphere-Biosphere Programme (IGBP) were then extracted from the OCO-2 data and programs were developed in Interactive Data Language (IDL) to read, bound, and sort SIF measurements by ecosystem class. Descriptive statistics data were generated in IDL in order to identify and evaluate patterns that may exist in OCO-2 SIF within distinct ecosystem classes.
Results/Conclusions
Evergreen Broadleaf Forest had the greatest mean SIF with 0.86 (±0.49 S.D.) Wm-2μm-1sr-1 for retrieval at 757nm and 0.58 (±0.37) Wm-2μm-1sr-1 for retrieval at 771nm. This class also made up the largest percentage of our dataset at 60.6%. Savannah had the second highest mean SIF values of 0.58 (±0.51) and 0.39 (±0.38) Wm-2μm-1sr-1 at 757 and 771nm, respectively. Cropland/Natural Vegetation mosaic SIF was very similar to Savannah with 0.57 (±0.52) and 0.39 (±0.39) Wm-2μm-1sr-1 at 757 and 771nm, respectively. SIF estimates from Cropland/Natural Vegetation were 1.7 and 1.4 greater than Cropland, indicating that mixed land-use may be more photosynthetically productive than intense land-use/segmentation for farming.
Our results are a step toward disentangling the SIF signal from different ecosystems, understanding the variation in SIF within ecosystems, determining the ecological basis for variation, and ultimately allowing better characterization of photosynthetic activity for heterogeneous landscapes. Preliminary results suggest land cover variability manifests in SIF measurements from OCO-2. We are currently applying ground data and Landsat-8 data to improve spatial resolution of classification, evaluate accuracy, and make data useful for regional/sub-regional SIF estimates. It is through this fractional characterization that regional assessments of photosynthetic activity may be attainable and eventually monitored in specific biomes.