Cold Bias of ERA5 Summertime Daily Maximum Land Surface Temperature Over Iberian Peninsula

Abstract

Land surface temperature (LST) is a key variable in surface-atmosphere energy and water exchanges. The main goals of this study are to (i) evaluate the LST of the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim and ERA5 reanalyses over Iberian Peninsula using the Satellite Application Facility on Land Surface Analysis (LSA-SAF) product and to (ii) understand the main drivers of the LST errors in the reanalysis. Simulations with the ECMWF land-surface model in offline mode (uncoupled) were carried out over the Iberian Peninsula and compared with the reanalysis data. Several sensitivity simulations were performed in a confined domain centered in Southern Portugal to investigate potential sources of the LST errors. The Copernicus Global Land Service (CGLS) fraction of green vegetation cover (FCover) and the European Space Agency’s Climate Change Initiative (ESA-CCI) Land Cover dataset were explored. We found a general underestimation of daytime LST and slightly overestimation at night-time. The results indicate that there is still room for improvement in the simulation of LST in ECMWF products. Still, ERA5 presents an overall higher quality product in relation to ERA-Interim. Our analysis suggested a relation between the large daytime cold bias and vegetation cover differences between (ERA5 and CGLS FCocver) with a correlation of -0.45. The replacement of the low and high vegetation cover by those of ESA-CCI provided an overall reduction of the large Tmax biases during summer. The increased vertical resolution of the soil at the surface, has a positive impact, but much smaller when compared with the vegetation changes. The sensitivity of the vegetation density parameter, that currently depends on the vegetation type, provided further proof for a needed revision of the vegetation in the model, as there is a reasonable correlation between this parameter and the Tmax mean errors when using the ESA-CCI vegetation cover (while the same correlation cannot be reproduced with the original model vegetation). Our results support the hypothesis that vegetation cover is one of the main drivers of the LST summertime cold bias in ERA5 over Iberian Peninsula.

ECLand: The ECMWF Land Surface Modelling System

“…All 2D and 3D simulations are run at the Tco399 Gaussian octahedral grid ( 25km), the offline simulations are driven by the ERA5 atmospheric forcing while the coupled forecast simulations are initialized with corresponding offline simulations to avoid unstable surface conditions related to spinup problems. (i) at least 70% of the fraction of valid pixels and (ii) total cloud cover lower than 30% [82]. For each day and pixel, when at least 75% of the 3-hourly data was available (6 out of 8), the daily maximum and minimum LST are extracted to compute the diurnal range.…”

An All-Weather Land Surface Temperature Product Based on MSG/SEVIRI Observations

“…model skin temperature and satellite LST [10,11,90,91]. Namely, they show a tendency for model surface temperatures to have a lower daily amplitude, particularly over arid and semi-arid regionssee the strong negative daytime bias over the Iberian Peninsula (July), part/most of Northern Africa in July/January and in Southern Africa in January.…”

“…In an attempt to increase the use of LST for climaterelated studies, Good [8] explored relationships of LST and near-surface (2 m) air temperature over a number of standard weather stations worldwide. LST is also useful for assessing and improving parameters in surface schemes of Numerical Weather Prediction (NWP) and Earth System Models to provide more accurate surface and near-surface diagnostics [9][10][11]. It has been recognized that land surface models often struggle to correctly represent clear sky skin temperature ( ), particularly regarding its diurnal amplitude over arid and semi-arid regions with an underestimation of daytime and a small overestimation of night-time values [10][11][12][13].…”

“…LST is also useful for assessing and improving parameters in surface schemes of Numerical Weather Prediction (NWP) and Earth System Models to provide more accurate surface and near-surface diagnostics [9][10][11]. It has been recognized that land surface models often struggle to correctly represent clear sky skin temperature ( ), particularly regarding its diurnal amplitude over arid and semi-arid regions with an underestimation of daytime and a small overestimation of night-time values [10][11][12][13]. Johannsen et al [11] showed that proposed an algorithm using LST information from neighboring clear-sky pixels together with sensible, latent and longwave fluxes, which were parameterized as a function of skin temperature.…”

“…It has been recognized that land surface models often struggle to correctly represent clear sky skin temperature ( ), particularly regarding its diurnal amplitude over arid and semi-arid regions with an underestimation of daytime and a small overestimation of night-time values [10][11][12][13]. Johannsen et al [11] showed that proposed an algorithm using LST information from neighboring clear-sky pixels together with sensible, latent and longwave fluxes, which were parameterized as a function of skin temperature. The method was applied to data collected during the First International Satellite Land Surface Climatology Project (ISLSCP) Field Experiment (FIFE) and the Boreal Ecosystem-Atmosphere Study (BOREAS) (https://www.gewex.org/data-sets-international-satellite-land-surface-climatologyproject-islscp/).…”