Publications

Biudes, MS; Vourlitis, GL; Velasque, MCS; Machado, NG; Danelichen, VHD; Pavao, VM; Arruda, PHZ; Nogueira, JD (2021). Gross primary productivity of Brazilian Savanna (Cerrado) estimated by different remote sensing-based models. AGRICULTURAL AND FOREST METEOROLOGY, 307, 108456.

Abstract
Gross primary production (GPP) is the total amount of fixed carbon and depends on vegetation health and water and energy availability. GPP has been monitored worldwide by flux towers and models that are coupled with remotely sensed data, such as by Moderate Resolution Imaging Spectroradiometer (MODIS). However, models have not been evaluated for tropical savanna, which presumably represent a challenge because of large spatial and seasonal variation in GPP. Thus, our goal was to evaluate the Vegetation Photosynthesis Model (VPM), Temperature and Greenness Model (TG), Vegetation Index Model (VI), and the MOD17A2 product of MODIS in a tropical mixed woodland-grassland. The Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), Land Surface Water Index (LSWI), Land Surface Temperature (LST) and Photosynthetically Active Radiation Fraction (fPAR) derived from the MODIS sensor were used as model inputs and integrated with ground-based micrometeorological variables. GPP varied significantly between the wet and dry seasons and was positively correlated with seasonal variations in soil volumetric water content (VSWC) and precipitation, and negatively correlated with the vapor pressure deficit (VPD). Satellite vegetation indices (NDVI, LSWI, and to a lesser extent, the EVI) and derived quantities (fPAR and LUE) also exhibited similar correlations with VSWC and precipitation. Thus, there were strong positive correlations between the SVIs and GPP. All of the models were able to simulate the seasonal variations in GPP; however, VPM had the best performance with the highest correlation and smallest errors. TG, VI, and MOD17A2 models performed similarly, except for the VI model based only on the EVI. Given their ability to capture seasonal dynamics, remote-sensing based models, such as those tested here, will likely be an important tool for assessing how climate variability alters C cycling dynamics in these spatially and temporally heterogeneous landscapes.

DOI:
10.1016/j.agrformet.2021.108456

ISSN:
0168-1923