Copernicus Global Land Operations - Scientific quality evaluation cross-cutting consistency - January-December 2019

The Copernicus Global Land Service (CGLS) is earmarked as a component of the Land service to operate a multi-purpose service component that provides a series of biogeophysical products on the status and evolution of land surface at a global scale. Production and delivery of the variables take place in a timely manner and are complemented by the constitution of long-term time series. The most advanced indirect validation technique consists in integrating the products into a land surface model (LSM) using a data assimilation scheme. The obtained reanalysis accounts for the synergies of the various upstream products and provides statistics which can be used to monitor the quality of the assimilated observations. Meteo-France develops the ISBA-A-gs generic LSM, able to represent the diurnal cycle of the surface fluxes together with the seasonal, inter-annual and decadal variability of the vegetation biomass. The LSM is embedded in the SURFEX modeling platform together with a simplified extended Kalman filter. These tools form a Land Data Assimilation System (LDAS). The current version of the LDAS (LDAS-Monde) is able to assimilate SPOT-VGT and PROBA-V Leaf Area Index (LAI) and ASCAT surface soil moisture (SSM) satellite products at a global scale at a spatial resolution of at least 0.25° x 0.25°. This permits the active monitoring of LAI and SSM variables. A passive monitoring of Surface Albedo (SA), Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) and Land Surface Temperature (LST) is performed (i.e., the simulated values are compared with the satellite products), as these quantities are not assimilated yet. The LDAS generates statistics whose trends can be analyzed in order to detect possible drifts in the quality of the products: (1) for LAI and SSM, metrics derived from the active monitoring (i.e. assimilation) such as innovations (observations vs. model), residuals (observations vs. analysis), and increments (analysis vs. model); (2) for SA, FAPAR and LST, metrics derived from the passive monitoring. In both cases, the Pearson correlation coefficient (R), the root mean square difference (RMSD), the standard deviation of difference (SDD), and mean bias skill scores are used. In this report, results are presented for the January-December 2019 period over Southern Africa and over the Murray-Darling basin. Note that the last data from SPOT-VGT were used on 13th May 2014. After this date, new LAI / FAPAR / SA products from PROBA-V are used. For LAI, over both Southern Africa and the Murray-Darling basin, the scores tend to present better values during the dry spells of 2019 than during previous years from 2010 to 2018. The RMSD scores of consolidated estimate of LAI Version 2 and LAI Version 1 are comparable in 2019. The impact on analyzed LAI of transitioning from SPOT-VGT to PROBA-V is neutral to positive. For FAPAR, over both Southern Africa and the Murray-Darling basin, the scores tend to present slightly better values during the dry spells of 2019 than during previous years from 2010 to 2018. Overall conclusions for FAPAR are similar to those for LAI. For SA, a striking result is that a large increase in the mean bias value is observed after the transition from SPOT-VGT to PROBA-V, of about 0.02 for both Southern Africa and for the Murray-Darling basin. There is a clear discontinuity in the SA time series, not observed for LAI nor for FAPAR. For SWI-001, the impact of the seasonal SSM CDF-matching performed prior the assimilation is particularly striking for Southern Africa. Without a seasonal CDF-matching, the original SSM information would be misleading over Southern Africa. For LST, the model tends to underestimate LST, especially at daytime. Over the Murray-Darling basin, the mean yearly bias is about -8°C in 2019 (a dry year), against -4°C in 2010 (a wet year). This result shows that daytime LST biases are more pronounced in dry conditions. Possible causes of the spatial, diurnal and seasonal patterns of the LST bias are hot-spot phenomenon (more sunlit than shaded elements are seen by the satellite), biases in the incoming solar and infrared radiation data used to force the model. LDAS analyses were also used to assess the accuracy of LAI and FAPAR observations, with respect to GCOS requirements. It is showed that small values of LAI observations tend to meet the GCOS requirements more often than large values of LAI observations and of FAPAR observations, for both Southern Africa and the Murray-Darling basin. Overall, low FAPAR values present more uncertainties than low LAI values