Evaluation of the Sentinel-3A Gross Dry Matter Productivity (GDMP) product for evergreen forests

The recent consolidation of global carbon monitoring systems has been coupled with notable advances in methods for estimating the carbon cycle components in terrestrial ecosystems. These developments include a comprehensive assessment of the uncertainty and biases associated with these estimations. Gross primary productivity (GPP) is the most significant component of the terrestrial carbon cycle. Accurate estimation of GPP and its fluctuations over space and time is crucial not only for assessing ecosystem functioning and carbon balance but also for evaluating the resilience of ecosystems to adapt, survive and thrive in response to climate changes. Unfortunately, direct measurement of GPP through remote sensing (RS) signals is not feasible. Several RS signals associated with vegetation pigments and canopy structures can indeed function as proxies for GPP. These signals can be effectively integrated with various modelling approaches, considering different types and levels of complexity, to generate an estimation of global GPP at high spatio-temporal resolution.

This study aims to explore how Sentinel satellites can improve the remote global GPP estimation. Specifically, we evaluated the quality of the 10-daily GDMP product of Sentinel-3, ensuring its reliability and credibility, with a specific focus on evergreen forests, particularly Aleppo pine stands. The outcomes of this study are expected to contribute to refining and calibrating GDMP algorithms for improved accuracy.

The first aspect of our methodology involves the selection of several Aleppo pine forests across South-East Spain, where eddy covariance towers were installed, to study inter-site variability including soil characteristics, vegetation dynamics and forest management. Then, direct cross-comparisons between eddy covariance measurements and satellite observations were conducted for 4 independent study sites covering different periods (i.e., 2015-2018 and 2019-2023) to quantify uncertainties and biases in the GDMP product.

The results revealed that the GDMP product exhibits improved performance during wet periods, ranging across sites from the highest R² of 0.83 to the lowest R² of 0.71, but it encounters challenges in accurately simulating Gross GPP under drought conditions. This funding was expected because some potentially important factors such as drought stress among others were omitted in the current computation model of Copernicus Global Land Service. Therefore, it is suggested that the product could be more accurately labelled as a potential GDMP. Similarly, our analysis showed that Aleppo pine demonstrates high plasticity in response to local conditions that is not adequately captured by this GDMP model.

To address the challenges encountered in accurately simulating the GDMP product, we are working on developing a potential solution. This involves incorporating drought stress factors to enhance the model by integrating relevant physiological and environmental variables that influence specific responses of Aleppo pine to water shortage.