Derivation of global vegetation biophysical parameters from EUMETSAT Polar System

Abstract

This paper presents the algorithm developed in LSA-SAF (Satellite Application Facility for Land Surface Analysis) for the derivation of global vegetation parameters from the AVHRR (Advanced Very High Resolution Radiometer) sensor on board MetOp (Meteorological-Operational) satellites forming the EUMETSAT (European Organization for the Exploitation of Meteorological Satellites) Polar System (EPS). The suite of LSA-SAF EPS vegetation products includes the leaf area index (LAI), the fractional vegetation cover (FVC), and the fraction of absorbed photosynthetically active radiation (FAPAR). LAI, FAPAR, and FVC characterize the structure and the functioning of vegetation and are key parameters for a wide range of landbiosphere applications. The algorithm is based on a hybrid approach that blends the generalization capabilities offered by physical radiative transfer models with the accuracy and computational efficiency of machine learning methods. One major feature is the implementation of multi-output retrieval methods able to jointly and more consistently estimate all the biophysical parameters at the same time. We propose a multi-output Gaussian process regression (GPRmulti), which outperforms other considered methods over PROSAIL (coupling of PROSPECT and SAIL (Scattering by Arbitrary Inclined Leaves) radiative transfer models) EPS simulations. The global EPS products include uncertainty estimates taking into account the uncertainty captured by the retrieval method and input errors propagation. A sensitivity analysis is performed to assess several sources of uncertainties in retrievals and maximize the positive impact of modeling the noise in training simulations. The paper discusses initial validation studies and provides details about the characteristics and overall quality of the products, which can be of interest to assist the successful use of the data by a broad user's community. The consistent generation and distribution of the EPS vegetation products will constitute a valuable tool for monitoring of earth surface dynamic processes.

Global Estimation of Biophysical Variables from Google Earth Engine Platform

“…The solution can be achieved by means of numerical optimization or Monte Carlo approaches which are computationally expensive and do not guarantee the convergence to an optimal solution. Recently, new and more efficient algorithms relying on Machine Learning (ML) techniques have emerged and have become the preferred choice for most RTM inversion applications [16][17][18][19]21]. In this work, we have followed the latter hybrid approach, combining radiative transfer modeling and the parallelized machine learning RFs implementation available in GEE to retrieve the selected biophysical variables.…”

“…Instead of using the common lookup tables available in the literature to parametrize the model [19,21,42], we exploit the potential of the TRY database to infer the distributions and correlations among some key leaf traits (leaf chlorophyll C ab , leaf dry matter C dm and water C w contents) required by PROSAIL. Table 2 shows some basic information about the considered traits extracted from the TRY.…”

“…The system's geometry was described by the solar zenith angle, the view zenith angle, and the relative azimuth angle between both angles, which in our case corresponded to illumination and observation zenith angles of 0 • . Sub-pixel mixed conditions (i.e., spatial heterogeneity) were tackled assuming a linear spectral mixing model, which pixels are composed by a mixture of pure vegetation (vCover) and bare soil (1-vCover) fractions' [21].…”

“…The leaf variables were randomly generated following the calculated kernel density distributions from the available leaf traits measurements, whereas distributions of the canopy variables as well as the soil brightness parameter, were similar to those adopted in other global studies [19,21]. Brown pigments were intentionally set to zero in order to account only for photosynthetic elements of the canopy (see Table 3).…”

“…On the other hand, hybrid methods couple statistical with physically-based approaches inverting a database generated by an RTM [16][17][18]. For example, CYCLOPES global products [19] were derived inverting the PROSAIL radiative transfer model [20] using neural networks, while the global EUMETSAT LAI/FAPAR/FVC products are being produced inverting PROSAIL with multi-output Gaussian process regression from the EUMETSAT Polar System (EPS) [21]. The use of RTMs implies modeling leaf and canopy structural and biochemical parameters.…”