Evaluating the quality of satellite-based photosynthetic activity products to estimate diurnal and short-term CO2 trends within the Amazon rainforest; An evaluation of different Vegetation Indices and the influence of clouds to estimate photosyntheti

By Coen Rouppe van der Voort

Abstract
The Amazon rainforest is a key driver in the global carbon cycle due to its capacity to absorb atmospheric CO₂ through photosynthesis. However, ongoing deforestation and climate change are weakening this carbon sink, prompting the need for improved understanding of vegetation–atmosphere interactions. This thesis evaluates the potential of satellite-based vegetation indices (VIs) to estimate short-term and diurnal photosynthetic activity trends in the Amazon, by comparing them with in situ CO₂ assimilation data collected during the Cloudroots campaign at the ATTO site in August 2022. Two satellite systems, PlanetScope and GOES-16, were assessed for their ability to capture spatial and temporal dynamics in photosynthetic activity. Four vegetation indices (EVI, NIRv, PRI, and Red-Edge chlorophyll index) were analysed in relation to measured net canopy assimilation (A_n). PlanetScope's high spatial resolution enabled the detection of short-term variability, while GOES-16 provided high temporal resolution suitable for diurnal analysis. Results indicate that NIRv displayed the strongest correlation with A_n for short-term trends (R² = 0.48), while EVI significantly improved diurnal An predictions when combined with solar zenith angle (SZA) data (P value = 0.0066). Short term NIRv results aligned with previous research, whereas diurnal measurements were influenced by additional factors causing differentiation from previous research. Clouds and atmospheric conditions were found to substantially affect VI performance. PRI was particularly sensitive to atmospheric optical effects, while Red-Edge remained relatively stable. These findings demonstrate that satellite-based VIs can provide valuable insights into photosynthetic dynamics within the Amazon rainforest, but their effectiveness varies depending on temporal scale and environmental conditions. Future research should focus on applying improved atmospheric correction methods, extending observational periods, and integrating sun sensor target variables to enhance future short-term and diurnal remote sensing assessments of carbon uptake in tropical ecosystems.