• Albergel, C., , Calvet J.-C. , , Gibelin A.-L. , , Lafont S. , , Roujean J.-L. , , Berne C. , , Traullé O. , , and Fritz N. , 2010: Observed and modeled gross primary production and ecosystem respiration of a grassland in southwestern France. Biogeosciences, 7, 16571668.

    • Search Google Scholar
    • Export Citation
  • Baldocchi, D. D., and Coauthors, 2001: FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Amer. Meteor. Soc., 82, 24152434.

    • Search Google Scholar
    • Export Citation
  • Béziat, P., , Ceschia E. , , and Dedieu G. , 2009: Carbon balance of a three crop succession over two cropland sites in South West France. Agric. For. Meteor., 149, 16281645.

    • Search Google Scholar
    • Export Citation
  • Calvet, J.-C., and Coauthors, 1999: MUREX: A land-surface field experiment to study the annual cycle of the energy and water budgets. Ann. Geophys., 17, 838854.

    • Search Google Scholar
    • Export Citation
  • Chen, T. H., and Coauthors, 1997: Cabauw experimental results from the project for intercomparison of land-surface parameterization schemes. J. Climate, 10, 11941215.

    • Search Google Scholar
    • Export Citation
  • Courtier, P., , Freydier C. , , Geleyn J.-F. , , Rabier F. , , and Rochas M. , 1991: The Arpège project at Météo France. Proc. ECMWF Seminar on Numerical Methods in Atmospheric Models, Vol. 2, Reading, United Kingdom, ECMWF, 193–232.

  • Diak, G. R., , Bland W. L. , , Mecikalski J. R. , , and Anderson M. C. , 2000: Satellite-based estimates of longwave radiation for agricultural applications. Agric. For. Meteor., 103, 349355.

    • Search Google Scholar
    • Export Citation
  • Douville, H., , Royer J. F. , , and Mahfouf J. F. , 1995: A new snow parameterization for the Météo-France climate model, Part I: Validation in stand-alone experiments. Climate Dyn., 12, 2135.

    • Search Google Scholar
    • Export Citation
  • Durand, Y., , Brun E. , , Mérindol L. , , Guyomarc’h G. , , Lesaffre B. , , and Martin E. , 1993: A meteorological estimation of relevant parameters for snow models. Ann. Glaciol., 18, 6571.

    • Search Google Scholar
    • Export Citation
  • Durand, Y., , Giraud G. , , Brun E. , , Merindol L. , , and Martin E. , 1999: A computer-based system simulating snowpack structures as a tool for regional avalanche forecasting. J. Glaciol., 45, 469484.

    • Search Google Scholar
    • Export Citation
  • Frouin, R., , Lingner D. W. , , Gautier C. , , Baker K. S. , , and Smith R. C. , 1989: A simple analytical formula to compute clear sky total and photosynthetically available solar irradiance at the ocean surface. J. Geophys. Res., 94, 97319742.

    • Search Google Scholar
    • Export Citation
  • Gandin, L. S., 1965: Objective Analysis of Meteorological Fields (in Russian). Isreal Program for Scientific Translations, 242 pp.

  • Geiger, B., , Meurey C. , , Lajas D. , , Franchistéguy L. , , Carrer D. , , and Roujean J.-L. , 2008: Near real-time provision of downwelling shortwave radiation estimates derived from satellite observations. Meteor. Appl., 15, 411420.

    • Search Google Scholar
    • Export Citation
  • Gibelin, A.-L., , Calvet J.-C. , , Roujean J.-L. , , Jarlan L. , , and Los S. O. , 2006: Ability of the land surface model ISBA-A-gs to simulate leaf area index at the global scale: Comparison with satellites products. J. Geophys. Res., 111, D18102, doi:10.1029/2005JD006691.

    • Search Google Scholar
    • Export Citation
  • Granier, A., , Bréda N. , , Longdoz B. , , Gross P. , , and Ngao J. , 2008: Ten years of fluxes and stand growth in a young beech forest at Hesse, North-eastern France. Ann. For. Sci., 65, 704, doi:10.1051/forest:2008052.

    • Search Google Scholar
    • Export Citation
  • Henderson-Sellers, A., , Yang Z.-L. , , and Dickinson R. E. , 1993: The project for intercomparison of land–surface parameterization schemes. Bull. Amer. Meteor. Soc., 74, 13351349.

    • Search Google Scholar
    • Export Citation
  • Ineichen, P., , Barroso C. , , Geiger B. , , Hollmann R. , , Marsouin A. , , and Mueller R. , 2009: Satellite Application Facilities irradiance products: Hourly time step comparison and validation over Europe. Int. J. Remote Sens., 30, 55495571.

    • Search Google Scholar
    • Export Citation
  • Josey, S. A., , Pascal R. W. , , Taylor P. K. , , and Yelland M. J. , 2003: A new formula for determining the atmospheric longwave flux at ocean surface at mid-high latitudes. J. Geophys. Res., 108, 3108, doi:10.1029/2002JC001418.

    • Search Google Scholar
    • Export Citation
  • Kucharik, C. J., , and Twine T. E. , 2007: Residue, respiration, and residuals: Evaluation of a dynamic agroecosystem model using eddy flux measurements and biometric data. Agric. For. Meteor., 146 (3–4), 134158.

    • Search Google Scholar
    • Export Citation
  • Kucharik, C. J., , Barford C. , , El Maayar M. , , Wofsy S. C. , , Monson R. K. , , and Baldocchi D. D. , 2006: A multiyear evaluation of a Dynamic Global Vegetation Model at three AmeriFlux forest sites: Vegetation structure, phenology, soil temperature, and CO2 and H2O vapor exchange. Ecol. Modell., 196, 131.

    • Search Google Scholar
    • Export Citation
  • Le Moigne, P., 2002: Description de l’analyse des champs de surface sur la France par le système SAFRAN (Description of the analysis of near-surface atmospheric fields over France with Safran system). CNRM/GAME Tech. Note 77, 30 pp.

  • Le Moigne, P., Ed., 2009: SURFEX scientific documentation. Centre du Groupe de Météorologie à Moyenne Echelle Tech. Note 87, 211 pp. [Available online at http://www.cnrm.meteo.fr/surfex/.]

  • Martin, E., and Coauthors, 2007: Le code de surface externalisé SURFEX de Météo-France. Extended Abstracts, Ateliers de Mod´elisation de l’Atmosphère, Toulouse, France, 4 pp.

  • Masson, V., , Champeaux J. L. , , Chauvin F. , , Meriguet C. , , and Lacaze R. , 2003: A global database of land surface parameters at 1-km resolution in meteorological and climate models. J. Climate, 16, 12611282.

    • Search Google Scholar
    • Export Citation
  • Michelot, A., , Eglin T. , , Dufrêne E. , , Lelarge-Trouverie C. , , and Damesin C. , 2010: Comparison of seasonal variations in water-use efficiency calculated from the carbon isotope composition of tree rings and flux data in a temperate forest. Plant, Cell Environ., 34, 230244, doi:10.1111/j.1365-3040.2010.02238.x.

    • Search Google Scholar
    • Export Citation
  • Noilhan, J., , and Planton S. , 1989: A simple parameterization of land surface processes for meteorological models. Mon. Wea. Rev., 117, 536549.

    • Search Google Scholar
    • Export Citation
  • Noilhan, J., , and Mahfouf J. F. , 1996: The ISBA land surface parametrisation scheme. Global Planet. Change, 13, 145159, doi:10.1016/0921-8181(95)00043-7.

    • Search Google Scholar
    • Export Citation
  • Otkin, J., , Anderson M. C. , , Mecikalski J. R. , , and Diak G. R. , 2005: Validation of GOES-based insolation estimates using data from the U.S. Climate Reference Network. J. Hydrometeor., 6, 460475.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., , Zhang B. , , and Dutton E. G. , 2005: Do satellites detect trends in surface solar radiation? Science, 308, 850854.

  • Prata, A. J., 1996: A new long-wave formula for estimating downward clear-sky radiation at the surface. Quart. J. Roy. Meteor. Soc., 122, 11211151.

    • Search Google Scholar
    • Export Citation
  • Quintana-Seguí, P., and Coauthors, 2008: Analysis of near surface atmospheric variables: Validation of the SAFRAN analysis over France. J. Appl. Meteor. Climatol., 47, 92107.

    • Search Google Scholar
    • Export Citation
  • Rambal, S., , Joffre R. , , Ourcival J. M. , , Cavender-Bares J. , , and Rocheteau A. , 2004: The growth respiration component in eddy CO2 flux from a Quercus ilex Mediterranean forest. Global Change Biol., 10, 14601469.

    • Search Google Scholar
    • Export Citation
  • Ritter, B., , and Geleyn J. F. , 1992: A comprehensive radiation scheme for numerical weather prediction models with potential applications in climate simulations. Mon. Wea. Rev., 120, 303325.

    • Search Google Scholar
    • Export Citation
  • Salgado, R., , and Le Moigne P. , 2010: Coupling of the FLake model to the Surfex externalized surface model. Boreal Environ. Res., 15, 231244.

    • Search Google Scholar
    • Export Citation
  • Sarrat, C., and Coauthors, 2009: Mesoscale modeling of the CO2 interactions between the surface and the atmosphere applied to the April 2007 CERES field experiment. Biogeosciences, 6, 633646.

    • Search Google Scholar
    • Export Citation
  • Simmons, A., , Uppala S. , , Dee D. , , and Kobayashi S. , 2006: ERA-I: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, No. 110, ECMWF, Reading, United Kingdom, 25–35.

  • Szczypta, C., , Calvet J. , , Albergel C. , , Balsamo G. , , Boussetta S. , , Carrer D. , , Lafont S. , , and Meurey C. , 2011: Verification of the new ECMWF ERA-Interim reanalysis over France. Hydrol. Earth Syst. Sci., 15, 647666, doi:10.5194/hess-15-647-2011.

    • Search Google Scholar
    • Export Citation
  • Trigo, I. F., , Barroso C. , , Viterbo P. , , Freitas S. C. , , and Monteiro I. T. , 2010: Estimation of downward longwave radiation at the surface combining remotely sensed data and NWP data. J. Geophys. Res., 115, D24118, doi:10.1029/2010JD013888.

    • Search Google Scholar
    • Export Citation
  • Trigo, I. F., and Coauthors, 2011: The Satellite Application Facility for Land Surface Analysis. Int. J. Remote Sens., 32, 27252744.

  • Twine, T. E., and Coauthors, 2000: Correcting eddy-covariance flux underestimates over a grassland. Agric. For. Meteor., 103, 279300.

    • Search Google Scholar
    • Export Citation
  • Vidal, J.-P., , Martin E. , , Franchistéguy L. , , Baillon M. , , and Soubeyroux J.-M. , 2009: A 50-year high-resolution atmospheric reanalysis over France with the Safran system. Int. J. Climatol., 30, 16271644, doi:10.1002/joc.2003.

    • Search Google Scholar
    • Export Citation
  • Wielicki, B. A., , Barkstrom B. R. , , Harrison E. F. , , Lee III R. B. , , Smith G. L. , , and Cooper J. E. , 1996: Clouds and the Earth’s Radiant Energy System (CERES): An Earth Observing System experiment. Bull. Amer. Meteor. Soc., 77, 853868.

    • Search Google Scholar
    • Export Citation
  • Wilson, K. B., and Coauthors, 2002: Energy partitioning between latent and sensible heat flux during the warm season at FLUXNET sites. Water Resour. Res., 38, 1294, doi:10.1029/2001WR000989.

    • Search Google Scholar
    • Export Citation
  • Zhao, W., , Kuhn W. R. , , and Drayson S. R. , 1994: The significance of detailed structure in the boundary layer to thermal radiation at the surface in climate models. Geophys. Res. Lett., 21, 16311634.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Map of France and location of FLUXNET stations.

  • View in gallery

    Yearly 2006 mean statistics for (left) DSSF and (right) DSLF (W m−2): (top to bottom) mean (SAF), bias (SAFRAN − SAF), and SDD. The means and standard deviations for the 8602 grid points listed correspond to differences of daily averaged values for the whole year.

  • View in gallery

    Diurnal cycle of SAFRAN and LSA SAF (left) DSSF and (right) DSLF corresponding to quarterly average values: summertime (JJA) and wintertime [January–March (JFM)] in 2006.

  • View in gallery

    LSA SAF − SAFRAN difference in incoming solar radiation (expressed in W m−2) as a function of the altitude (m) for 2005 and 2006. I would like to reduce a the size of the figure 4 (20% smaller).

  • View in gallery

    The 2006 statistical comparison in of solar radiation between in situ measurement and LSA SAF (solid line) and SAFRAN (dotted line) for the six FLUXNET stations located in France. Vertical bars represent standard deviations.

  • View in gallery

    Impact assessment on (left) surface temperature (TG1) and on (right) soil temperature (TG2) due to using the LSA SAF radiative forcing (DSSF + DSLF) vs of SAFRAN. Statistics correspond to differences of daily averaged values for 2006. (top to bottom) Mean value with LSA SAF forcing, bias (SAFRAN − LSA SAF), and SDD (K).

  • View in gallery

    As in Fig.6, but for surface soil moisture (WG1) (m3 m−3).

  • View in gallery

    Seasonal variations in 2006 of the daily RN over the FLUXNET station of Aurade from ground measurements (dots) and ISBA using SAFRAN radiative forcing (line).

  • View in gallery

    As in Fig. 8, but for based on ISBA simulations using either SAFRAN or LSA SAF radiative forcing: between SAFRAN and ground measurement (gray line), and between LSA SAF forcing (DSSF + DSLF) and ground measurements (black dotted line).

  • View in gallery

    Variation of monthly absolute bias of RN for six FLUXNET stations based on ISBA simulations using various radiative forcings: SAFRAN forcing (Ref.), LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

  • View in gallery

    As in Fig. 10, but for SDD.

  • View in gallery

    Variations in 2006 of the correlation (r) of daily RN over six FLUXNET stations based on ISBA simulations using various radiative forcings: SAFRAN forcing (Ref.), LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

  • View in gallery

    Bias (absolute value) and SDD statistical scores between FLUXNET measurement and simulated energy fluxes with ISBA. Reduction (y axis) of the statistical scores due to the use of LSA SAF satellite forcing are compared to statistics obtained with SAFRAN forcing (x axis). Monthly averaged daily bias and SDD of RN, LE, and H. One point represents 1 of the 12 monthly statistical scores for one of the six FLUXNET stations (see Figs. 11 and 12 for RN). Satellite forcings are LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

  • View in gallery

    Correlation (r) between FLUXNET measurement and simulated energy fluxes with ISBA. Improvement (y axis) of the R statistical score due to the use of LSA SAF satellite forcing is compared to the R obtained with SAFRAN forcing (x axis). The daily R of RN, LE, and H are monthly averaged. One point represents one of the 12 monthly statistical scores for one of the six FLUXNET stations (see Fig. 13 for RN). Satellite forcing are LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 8 8 4
PDF Downloads 4 4 1

Incoming Solar and Infrared Radiation Derived from METEOSAT: Impact on the Modeled Land Water and Energy Budget over France

View More View Less
  • 1 CNRM/GAME, Météo-France/CNRS, Toulouse, France
  • 2 Instituto de Meteorologia, Lisbon, Portugal
© Get Permissions
Full access

Abstract

The Land Surface Analysis Satellite Applications Facility (LSA SAF) project radiation fluxes, derived from the Meteosat Second Generation (MSG) geostationary satellite, were used in the Interactions between Soil, Biosphere, and Atmosphere (ISBA) land surface model (LSM), which is a component of the Surface Externalisée (SURFEX) modeling platform. The Système d’Analyze Fournissant des Renseignements Atmosphériques à la Neige (SAFRAN) atmospheric analysis provides high-resolution atmospheric variables used to drive LSMs over France. The impact of using the incoming solar and infrared radiation fluxes [downwelling surface shortwave (DSSF) and longwave (DSLF), respectively] from either SAFRAN or LSA SAF, in ISBA, was investigated over France for 2006. In situ observations from the Flux Network (FLUXNET) were used for the verification. Daily differences between SAFRAN and LSA SAF radiation fluxes averaged over the whole year 2006 were 3.75 and 2.61 W m−2 for DSSF and DSLF, respectively, representing 2.5% and 0.8% of their average values. The LSA SAF incoming solar radiation presented a better agreement with in situ measurements at six FLUXNET stations than the SAFRAN analysis. The bias and standard deviation of differences were reduced by almost 50%. The added value of the LSA SAF products was assessed with the simulated surface temperature, soil moisture, and the water and energy fluxes. The latter quantities were improved by the use of LSA SAF satellite estimates. As many areas lack a high-resolution meteorological analysis, the LSA SAF radiative products provide new and valuable information.

Corresponding author address: Dominique Carrer, Météo-France, 42, Avenue Gaspard Coriolis, 31057 Toulouse CEDEX, France. E-mail: dominique.carrer@meteo.fr

Abstract

The Land Surface Analysis Satellite Applications Facility (LSA SAF) project radiation fluxes, derived from the Meteosat Second Generation (MSG) geostationary satellite, were used in the Interactions between Soil, Biosphere, and Atmosphere (ISBA) land surface model (LSM), which is a component of the Surface Externalisée (SURFEX) modeling platform. The Système d’Analyze Fournissant des Renseignements Atmosphériques à la Neige (SAFRAN) atmospheric analysis provides high-resolution atmospheric variables used to drive LSMs over France. The impact of using the incoming solar and infrared radiation fluxes [downwelling surface shortwave (DSSF) and longwave (DSLF), respectively] from either SAFRAN or LSA SAF, in ISBA, was investigated over France for 2006. In situ observations from the Flux Network (FLUXNET) were used for the verification. Daily differences between SAFRAN and LSA SAF radiation fluxes averaged over the whole year 2006 were 3.75 and 2.61 W m−2 for DSSF and DSLF, respectively, representing 2.5% and 0.8% of their average values. The LSA SAF incoming solar radiation presented a better agreement with in situ measurements at six FLUXNET stations than the SAFRAN analysis. The bias and standard deviation of differences were reduced by almost 50%. The added value of the LSA SAF products was assessed with the simulated surface temperature, soil moisture, and the water and energy fluxes. The latter quantities were improved by the use of LSA SAF satellite estimates. As many areas lack a high-resolution meteorological analysis, the LSA SAF radiative products provide new and valuable information.

Corresponding author address: Dominique Carrer, Météo-France, 42, Avenue Gaspard Coriolis, 31057 Toulouse CEDEX, France. E-mail: dominique.carrer@meteo.fr

1. Introduction

Land surface models (LSMs) used to monitor biophysical variables in hydrological and/or environmental applications require detailed information on the characteristics of the incoming solar or infrared irradiance. Together with atmospheric variables like air temperature, air humidity, precipitation, and wind speed, reliable hourly estimates of downwelling surface shortwave (DSSF) and longwave (DSLF) radiation fluxes are needed to drive the LSMs. These two quantities can be provided by interpolated in situ measurements, atmospheric simulations, or derived from high-temporal-resolution satellite observations.

At a global scale, atmospheric reanalyses, such as the European Centre for Medium-Range Weather Forecasts (ECMWF) Interim Re-Analysis (ERA-I), provide meteorological data for such applications (Simmons et al. 2006). In France, a mesoscale atmospheric analysis system devoted to mountainous regions was built in the 1990s to provide the atmospheric forcing to a snow model for the forecast of avalanche hazards (Durand et al. 1993, 1999): Système d’Analyze Fournissant des Renseignements Atmosphériques à la Neige (SAFRAN). More recently, the SAFRAN analysis was extended to cover the whole of France in order to drive the Interactions between Soil, Biosphere, and Atmosphere (ISBA) LSM model (Noilhan and Planton 1989) with meteorological data (Le Moigne 2002).

Geostationary satellites provide information that enables frequent, routine, and cost-effective estimates of both DSSF and DSLF radiations for use by agricultural and other applications, at a resolution that cannot be practically matched by ground-based measurements (Diak et al. 2000). Satellite determination of shortwave radiation essentially depends on the solar zenith angle, on cloud coverage, and to a lesser extent on atmospheric absorption and surface albedo (Geiger et al. 2008). Thermal– infrared irradiance is the result of atmospheric absorption, emission, and scattering within the entire atmospheric column, although the radiation reaching the surface is essentially emitted within the first hundred meters of the atmosphere (Zhao et al. 1994). Under clear-sky situations, DSLF depends on the vertical profile of temperature and gaseous absorber concentration—mainly water vapor and CO2. The cloud contribution mainly occurs in the atmospheric windows (8–13 μm) and mainly depends on cloud-base properties (height, temperature, and emissivity). The resolution scale of a remote sensing instrument is relevant because it allows a better identification, for instance, between broken clouds, which cast shadow and edge effects, and thin clouds like cirrus.

Narrow-band International Satellite Cloud Climatology Project (ISCCP) and Moderate Resolution Imaging Spectroradiometer (MODIS) as well as broadband Clouds and the Earth’s Radiant Energy System (CERES) data have been used to estimate surface radiation fluxes using a variety of techniques (Pinker et al. 2005; Wielicki et al. 1996). Also, the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) Land Surface Analysis Satellite Applications Facility (LSA SAF) project ensures the operational dissemination of DSSF and DSLF products from the Meteosat Second Generation (MSG) geostationary satellite (http://landsaf.meteo.pt). The particularity here is that DSSF and DSLF are fully consistent with other products such as land surface albedo, land surface temperature, snow cover, leaf area index, and vegetation cover fraction (Trigo et al. 2011). The target accuracy, as specified within the EUMETSAT network for the DSSF and DSLF products, is 10%; in the case of DSSF values below 200 W m−2, an absolute error of 20 W m−2 is used instead. These values are considered for both clear and cloudy conditions, despite the differences in respective algorithm performance. Note that an optimum precision of 5% is targeted, which is closer to Global Climate Observing System (GCOS) specifications. Nevertheless, such specifications may be stringent for cloudy conditions, triggering a significant DSSF decrease.

The objective of this study is to compare the incoming solar and longwave radiations from SAFRAN with the corresponding satellite-derived LSA SAF DSSF and DSLF estimates, respectively, and to assess the impact on LSM simulations of the uncertainties of the radiative atmospheric forcing. Land surface models are designed to simulate the exchanges of energy, matter, and momentum between the land surface and the atmosphere. The LSA SAF products, as a result of the short repeat cycles of the MSG geostationary satellite, are particularly well suited for such impact studies over Europe.

The LSM used in this study is the three-layer version of the ISBA model (Noilhan and Mahfouf 1996). ISBA is imbedded into Surface Externalisée (SURFEX) (Martin et al. 2007; Le Moigne 2009; Salgado and LeMoigne 2010), which is the surface modeling platform of Météo-France. SURFEX can be used offline, forced by atmospheric analyses (or by local meteorological observations), or online, coupled with an atmospheric model. SURFEX is used for operational and research applications, either offline (e.g., Gibelin et al. 2006) or online (e.g., Sarrat et al. 2009). In this study, SURFEX is used offline and there is no feedback from the surface to the atmosphere. The radiative atmospheric forcing comes from the SAFRAN system or from the LSA SAF satellite data.

The paper is organized as follows: section 2 presents the datasets and the ISBA simulations. An intercomparison of radiation products and an evaluation of their impact on ISBA simulations are made in section 3. The conclusion section summarizes the added value of the use of LSA SAF products for LSM applications.

2. Presentation of atmospheric forcing datasets

a. SAFRAN analysis

SAFRAN was primarily designed to provide consistent atmospheric forcings in the context of a marked topography with the presence of snow (e.g., Durand et al. 1993, 1999). SAFRAN was extended to low-altitude zones in order to develop hydrological applications over France. SAFRAN considers 615 homogeneous climatic zones usually smaller than 1000 km2. For each zone, SAFRAN estimates one value of the key meteorological variables (2-m air temperature and air humidity, 10-m wind speed, incoming radiation, and cloudiness) at several altitude levels sliced by 300 m, which means an adjustment of atmospheric profiles according to the topography. SAFRAN also performs a 6-h analysis of the synoptic network. All variable fields follow an optimal interpolation between the initial guess, conventional observations, and metrics from the geographic distribution (e.g., Gandin 1965). The first guess comes from the large-scale operational weather prediction model Action de Recherche Petite Echelle Grande Echelle (ARPEGE) (Courtier et al. 1991) or from the ECMWF operational archives. First-guess vertical profiles are extrapolated from the lowest atmospheric level to the surface in order to prepare continuous profiles for each 300-m slice where analysis is performed. All vertical profiles (air temperature, air humidity, and cloudiness) and surface wind are linearly interpolated to an hourly time step. Finally, a vertical interpolation of the analyzed variables is performed to account for the surface orography computed on the SAFRAN output grid. SAFRAN output grid has an 8-km horizontal resolution and uses a Lambert-II projection. It is also at this stage that solar radiation and longwave radiation fluxes are calculated using a radiative transfer scheme (Ritter and Geleyn 1992). Hitherto, no satellite information was considered.

The SAFRAN incoming solar radiation was compared by Vidal et al. (2009) to ground stations distributed over France. The daily scores in terms of bias and root-mean-square error (RMSE) were around −3.7 and 41 W m−2, respectively, over the 1986–2007 period. These daily scores are consistent with the hourly scores given by Quintana-Seguí et al. (2008). At the hourly time step, the RMSE for 2004/05 was assessed to 93 W m−2, which corresponds to 60% of its mean value. Vidal et al. (2009) noticed the lack of observations available to the radiation transfer scheme in the southern part of Massif Central, which leads to an overestimation of solar radiation and an underestimation of infrared radiation. Coastal areas present some biases too. SAFRAN also appears to underestimate the daily maximum of incoming solar radiation (Quintana-Seguí et al. 2008). Over France, only two well-instrumented stations are equipped with an IR sensor. The RMSE for DSLF at the hourly time step in these two stations is moderate (33 and 42 W m−2) and bias important (−8 and −32 W m−2) relative to the shortwave (Quintana-Seguí et al. 2008).

b. LSA SAF operational products

The LSA SAF project is supported by EUMETSAT and by National Meteorological Services. The LSA SAF consortium is led by the Portuguese Meteorological Institute. DSSF and DSLF estimates are derived from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) instrument on board MSG. They are operationally disseminated since September 2005 (http://landsaf.meteo.pt) with a temporal frequency of 30 min at the full SEVIRI spatial resolution over the whole MSG disk. The SEVIRI sensor has a nadir resolution of 3 km in the shortwave channels VIS0.6, VIS0.8, and SWIR1.6, centered on 0.6, 0.8, and 1.6 μm, respectively.

The method used to estimate DSSF is derived from Frouin et al. (1989). Main inputs of DSSF are cloud characteristics (cloud mask, cloud type, and cloudiness) and the total precipitable water. Cloud characteristics are determined using a method developed by the Nowcasting and Very Short Range Forecasting Satellite Application Facility (NWC-SAF) (http://www.meteorologie.eu.org/safnwc/) and implemented in the LSA SAF processing chain. Water vapor is provided by the ECMWF numerical weather prediction (NWP) model. Climatology is currently used for ozone concentration and aerosol properties. Two distinct algorithms are handled to treat clear- and cloudy-sky situations. With the occurrence of clouds, the downwelling radiation reaching the ground is considerably reduced. DSSF is thus strongly anticorrelated with top-of-atmosphere (TOA) reflectance values: the brighter the clouds, the more radiation is reflected in direction of the satellite and the less radiation reaches the surface. In this case, TOA albedo is first calculated from the observed directional reflectance values by applying a broadband conversion and an angular dependence model. In the clear-sky method, DSSF is estimated with an empirical parameterization for the effective transmittance of the atmosphere as a function of the concentration of atmospheric constituents. A detailed description of the algorithm, and validation results, are given in Geiger et al. (2008). These authors report a standard deviation of differences (SDD) for instantaneous DSSF measurements between satellite and six European ground measurement stations in the order of 40 W m−2 (clear sky), 110 W m−2 (cloudy sky), and 85 W m−2 (both cases) for the 2004/05 period. The mean bias between instantaneous satellite and in situ radiation fluxes is less than 10 W m−2. Daily averaged estimates of DSSF present SDD less than 45 W m−2 for all situations. A product evaluation by Ineichen et al. (2009) led to similar results.

DSLF estimates make use of a combination of satellite information on clouds and of NWP fields. The adopted algorithm to compute DSLF for the period under analysis in this paper consists of a hybrid method based on two different bulk parameterization schemes (e.g., Prata 1996; Josey et al. 2003). Inputs are the ECMWF forecasts for the 2-m temperature, dewpoint temperature, and the total column water vapor. Two cloud products (cloud mask and effective cloudiness) from NWC-SAF are used, thereby making fully consistent the segregation of cloudy pixels between DSSF and DSLF derivations. The verification of DSLF product is a difficult task as this quantity is not extensively measured by weather stations. The validation of this product (Trigo et al. 2010 and also in the report available on the LSA SAF website: http://landsaf.meteo.pt) is based on 10 ground observations. The comparison against in situ measurements suggests that the LSA SAF algorithm generally underestimated DSLF. The instantaneous bias is often close to −10 W m−2 but could be up to −30 W m−2. The RMSE is about 30 W m−2. The underestimation of DSLF product is particularly apparent for clear-sky cases, with biases of the order of −10 to −20 W m−2. Cloudy pixels also exhibited negative biases, but higher dispersion than in clear cases. The accuracy of LSA SAF DSLF product compares well with performance of Geostationary Operational Environmental Satellite (GOES)-based DSLF obtained by Diak et al. (2000) (RMSE of 20 W m−2 for half-hourly average from two pyrgeometer measurements).

To compare the LSA SAF products with SAFRAN (section 3a), they were reprojected and averaged from the original SEVIRI grid to the regular 8-km grid of SAFRAN in Lambert-II.

c. Ground measurements

Flux Network (FLUXNET; http://daac.ornl.gov/FLUXNET/fluxnet.shtml) is a global network focusing on micrometeorological measurements at tower-equipped sites. Water, carbon, and energy fluxes are measured with the eddy-covariance method (Baldocchi et al. 2001).

In situ observations in 2006 were extracted for six French sites ranging from single vegetation types to a mixture of species (see Fig. 1): Auradé (maize; Béziat et al. 2009), Fontainebleau (deciduous broadleaf trees; Michelot et al. 2010), Hesse (deciduous broadleaf trees; Granier et al. 2008), Lamasquère (maize; Béziat et al. 2009), Lusignan (temperate grassland), and Puechabon (deciduous broadleaf trees; Rambal et al. 2004).

Fig. 1.
Fig. 1.

Map of France and location of FLUXNET stations.

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

These FLUXNET stations include observations of incoming solar radiation and are used, in this study, to provide the ground truth reference to assess the LSA SAF products. Moreover, the consistency of the impact assessment on LSM simulations could be verified from these six independent ground stations by comparing modeled and measured energy fluxes.

d. Model simulations

The representation of the soil–vegetation–atmosphere exchanges in the SURFEX platform uses the ISBA parameterization scheme (Noilhan and Planton 1989). The goal in implementing SURFEX is to gather all developments conducted in surface modeling at Météo-France for the generic surfaces units (soil–vegetation, urban areas, sea surface, and lakes) (http://www.cnrm.meteo.fr/surfex/). Besides, SURFEX yields the necessary interface between the atmospheric and hydrological modeling. The ISBA parameters, and the fraction of surface types, are mapped using the ECOCLIMAP database (Masson et al. 2003), which includes a land cover classification in association with sets of surface parameters that are primarily useful for the achievement of climate studies.

At each time step, an ISBA surface grid box is forced by surface atmospheric variables (air temperature, air specific humidity, horizontal wind components, atmospheric pressure, liquid and solid precipitation, and incoming direct and diffuse shortwave and longwave radiation). In addition to surface radiation variables, ISBA simulates the momentum and sensible and latent heat fluxes. These fluxes serve as lower boundary conditions to calculate the atmospheric radiative and turbulent properties.

The main advantage of the ISBA parameterization is that it is capable of accurately reproducing the energy and water budgets with a simple set of equations as confirmed by the Project for the Intercomparison of Land-Surface Parameterization Schemes (PILPS; Henderson-Sellers et al. 1993; Chen et al. 1997) or Calvet et al. (1999).

In this study, we consider the three-layer version of the ISBA model in which the deepest soil layer may provide water to the root zone through capillary rises only. The thickness of the top layer is 0.01 m. The two other layers have a varying thickness over France that depends on the soil water holding capacity. The depths of second and third layers in ECOCLIMAP (Masson et al. 2003) have maximum values of 2.5 and 3 m, respectively. ISBA simulations will concern the ground temperature and water content of the top layer (TG1 and WG1) and the second layer (TG2 and WG2). Although a multilayer scheme for snow is available in ISBA, a one-layer configuration (see Douville et al. 1995) is retained here as it is appears appropriate for the time period and area of the study. From 1 January to 31 December 2006 with a time step of 15 min (or 900 s), no fewer than four ISBA simulations were performed over France on a Lambert-II grid having a 8-km resolution: 1) forced by SAFRAN (hereinafter referred to as “reference”), or forced by SAFRAN except for MSG-derived 2) DSSF, 3) DSLF, or 4) DSSF and DSLF (hereinafter referred to as DSSF, DSLF, and DSSF + DSLF, respectively). It is worth mentioning that the ISBA model was spun up twice in offline mode with the SAFRAN atmospheric forcing of 2005; after that the soil reaches an equilibrium state that can serve for initialization.

3. Comparison of the radiation products and impact assessment

a. LSA SAF vs SAFRAN

A comparison between the LSA SAF and SAFRAN radiation products was performed over France for the whole year 2006. In what follows, all statistics are daily scores on a yearly or monthly basis between daily averaged values.

Table 1 displays quantitative statistics of the comparison between the LSA SAF and SAFRAN radiation products over France in 2006. On a daily basis, the mean bias is 3.75 W m−2 for DSSF and 2.61 W m−2 for DSLF, which represents 2.5% and 0.8% of their mean value, respectively. Even though these statistics included nighttime values, the discrepancies for the quantity DSSF (which is equal to zero during nighttime) appeared to be more important than for DSLF. In terms of cumulative DSSF, such a difference represents 2.5% of the 52 541 W m−2 of cumulative solar radiation that reached the France territory in 2006. The SDD of daily averaged values represent 34 and 17 W m−2, respectively. For DSSF, the standard deviation is high on a daily time scale, while the bias is low (3.75 W m−2) on a yearly basis. This reveals that LSA SAF and SAFRAN are able to estimate the yearly accumulated DSSF quantities, while large discrepancies may occur for cumulative daily incoming flux values. Somewhat interestingly, this latter feature is not the fact of a particular time of the year and hence SDD is not linked to the season.

Table 1.

Mean, bias, and SDD of daily incoming solar (DSSF) and infrared (DSLF) radiation between SAFRAN and LSA SAF over France in 2006.

Table 1.

Bias and SDD values averaged over France are shown in Fig. 2. This set of figures enhances the climate zoning used in the SAFRAN analysis, which explains the spatial discontinuity of the statistical scores. For instance, the incoming longwave radiation from SAFRAN shows values 20 W m−2 larger than the LSA SAF DSLF along the western coastline of France. On the other hand, SAFRAN DSLF values lower than the LSA SAF DSLF, of about 10 W m−2, are observed at the foothills of Massif Central, over the Pyrenees, and over the Alps. The DSSF biases are often negative over coastal areas (−15 W m−2). Positive biases are observed over Massif Central (up to 20 W m−2) in the center of France. These findings are consistent with the Quintana-Seguí et al. (2008) results (see section 2a). SDD of DSLF values over the Alps exceed 20 W m−2 at some locations. These same areas have low values of bias. Such results indicate day-to-day sizeable deviations between LSA SAF and SAFRAN radiation fluxes, whereas a common yearly trend is well depicted.

Fig. 2.
Fig. 2.

Yearly 2006 mean statistics for (left) DSSF and (right) DSLF (W m−2): (top to bottom) mean (SAF), bias (SAFRAN − SAF), and SDD. The means and standard deviations for the 8602 grid points listed correspond to differences of daily averaged values for the whole year.

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Although biases on DSSF and DSLF are very low on a daily basis (see Table 1), significant discrepancies occur during the daily cycle. Figure 3 shows average diurnal cycles at summertime and at wintertime. At summertime [June–August (JJA)], a difference of about 50 W m−2 is observed between the two DSSF products at midday. This is consistent with the results of Quintana-Seguí et al. (2008) and Szczypta et al. (2011) regarding the SAFRAN underestimation of the daily maximum DSSF. As far as DSLF is concerned, Fig. 3 reveals that the two products are anticorrelated, with a LSA SAF DSLF maximum at 1500 UTC, whereas SAFRAN indicates a minimum at the same time. In summer, convective clouds usually appear at 1500 UTC, and this process tends to trigger the DSLF maximum seen by the LSA SAF product. Nevertheless it would not explain the behavior during the winter months. Finally, Fig. 4 shows the difference between the LSA SAF and the SAFRAN DSSF averaged over the years 2005 and 2006 as a function of the elevation. It appears that the bias increases sharply from sea level up to 1000 m, and stabilizes for higher elevations. Although the bias is higher for 2006 than for 2005, a similar response of the bias to the elevation is observed for the two years.

Fig. 3.
Fig. 3.

Diurnal cycle of SAFRAN and LSA SAF (left) DSSF and (right) DSLF corresponding to quarterly average values: summertime (JJA) and wintertime [January–March (JFM)] in 2006.

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Fig. 4.
Fig. 4.

LSA SAF − SAFRAN difference in incoming solar radiation (expressed in W m−2) as a function of the altitude (m) for 2005 and 2006. I would like to reduce a the size of the figure 4 (20% smaller).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Although SAFRAN and LSA SAF are generally quite consistent, significant discrepancies can be noticed over certain regions and on an hourly basis. A part of the discrepancies can be due to the hourly interpolation of the 6-h analysis of SAFRAN fields.

b. LSA SAF and SAFRAN vs FLUXNET

The FLUXNET stations used in this study only measure the shortwave radiation components. Figure 5 presents time series of statistics (bias and SDD) between ground measurements and SAFRAN or LSA SAF data for DSSF over the six FLUXNET sites (see Fig. 1). These statistics between FLUXNET daily observations and daily averaged SAFRAN data have been calculated on a monthly basis in order to display possible seasonal trends. Except for Puechabon, no seasonal effects are evident. LSA SAF seems to be slightly less biased than SAFRAN and, also, presents lower SDD values. For example, a difference of about 30 W m−2 between the two SDD is observed for the Fontainebleau site from February to May 2006. On a yearly basis, for the six FLUXNET stations in 2006, the bias (SDD) scores of daily averaged SAFRAN and LSA SAF data are 0.88 (35) and 5.31 (18) W m−2 (Table 2). Since negative and positive values of the bias in different stations may balance each other (−13.71 W m−2 in Fontainebleau and 14.39 W m−2 in Puechabon for SAFRAN), it is relevant to consider the bias of absolute values. The mean absolute biases of SAFRAN and LSA SAF data are 9.22(6.6%) and 5.31(3.8%) W m−2, respectively. On average in 2006 for the six FLUXNET stations, the LSA SAF scores with respect to SAFRAN are better by 3.91(2.8%) W m−2 and 17(12.1%) W m−2 for the absolute bias and for the daily SDD, respectively.

Fig. 5.
Fig. 5.

The 2006 statistical comparison in of solar radiation between in situ measurement and LSA SAF (solid line) and SAFRAN (dotted line) for the six FLUXNET stations located in France. Vertical bars represent standard deviations.

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Table 2.

Bias and SDD of daily incoming visible radiation (DSSF) between SAFRAN and in situ measurements and between LSA SAF and in situ measurements in 2006 for six FLUXNET stations over France.

Table 2.

In Otkin et al. (2005), the mean bias errors at 23 stations are plotted versus the station elevation. The comparison suggests that satellite-derived surface insolation is systematically overestimated for stations at lower elevations (300 m) and underestimated at higher-elevation sites. Such a dependency could not be assessed in this study, as all the investigated stations had an elevation lower than 300 m. Nevertheless, such dependence could exist with regard to the shape of the difference between both fluxes (SAFRAN and LSA SAF) as a function of the elevation (Fig. 4).

c. Impact on ISBA simulations

In this section the impact on LSM simulations from using either SAFRAN or LSA SAF radiation forcings is examined with two diagnostic variables (surface and soil temperature and soil water content). The simulated energy balance with SURFEX is also analyzed: net radiation (RN), sensible heat flux (H), and latent heat flux (LE). The model simulations were compared with the ground measurements from the six FLUXNET French sites described above.

1) Temperature

Figure 6 shows the mean values of surface temperature (TG1) and soil temperature (TG2) over France produced by ISBA using the LSA SAF forcing for 2006. For a grassland site in southwestern France, Albergel et al. (2010) found that TG2 was a good estimation of soil temperature at a depth of 20 cm. Also shown are the mean bias and SDD using either SAFRAN or LSA SAF as radiative forcing (DSSF + DSLF). It comes out that the average impact over the year of using the LSA SAF incoming radiation can locally exceed 1 K. Higher temperature values are obtained with SAFRAN over western France. Lower values are found over the Pyrenees and also in Corsica. The impact at the surface (TG1) is similar to the impact in the soil (TG2). This is consistent because TG2 characterizes the soil diurnal temperature cycle driven by TG1 fluctuations. Although SDD is lower for TG2 because of the thermal inertia of the soil, the TG1 SDD reaches 3 K over mountainous areas. A comparison of instantaneous temperatures (not shown) also revealed significant differences mainly due to differences in cloudiness: maximum differences fell within the ranges (−10 K to 15 K) for TG1 and (−3 K to 7 K) for TG2.

Fig. 6.
Fig. 6.

Impact assessment on (left) surface temperature (TG1) and on (right) soil temperature (TG2) due to using the LSA SAF radiative forcing (DSSF + DSLF) vs of SAFRAN. Statistics correspond to differences of daily averaged values for 2006. (top to bottom) Mean value with LSA SAF forcing, bias (SAFRAN − LSA SAF), and SDD (K).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

2) Soil water content

Figure 7 displays statistical results for the surface soil water content (WG1) using the same protocol as for temperature. The impact of using the satellite forcing is less than 5% over the major part of France. Establishing a close link between difference on radiative forcing and impact on water content or temperature is not straightforward. Impact on the root-zone soil moisture was negligible. Table 3 summarizes the impact on TG1, TG2, and WG1 variables. On average for 2006 over France, the use of the LSA SAF forcing (instead of SAFRAN) induces a difference of 0.38 K/0.36 K/0.001 m3 m−3 for the TG1/TG2/WG1 daily values, respectively. Consequently, a mean difference of 2.5% on DSSF and 0.8% on DSLF forcing (section 2a) affects the averaged values over one year of TG1/TG2/WG1 by 3.5%/3.3%/0.5%.

Fig. 7.
Fig. 7.

As in Fig.6, but for surface soil moisture (WG1) (m3 m−3).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Table 3.

Difference statistics between simulations using SAFRAN or LSA SAF radiative forcing (DSSF +DSLF) for estimation of surface temperature (TG1), soil temperature of the root zone, and surface water content (WG1). Statistics are averaged daily differences for 2006 over France domain.

Table 3.

3) RN

The seasonal cycle of the daily averaged RN is shown in Fig. 8 for the reference ISBA simulation over the Aurade site in 2006. Simulated RN values with ISBA and in situ measurement indicate that RN reaches a maximum in summer, as could be expected. The wintertime RN is small or is even slightly negative. The large bias during summer in Aurade station could have multiple causes, amongst which is the uncertainty on the atmospheric forcing. Cedilnik et al. (2011, manuscript submitted to J. Appl. Meteor. Climatol.) examined the impact on short-range forecasts of daily satellite-derived albedos in a limited-area NWP model over Europe. From comparisons against three measuring flux tower stations in France, it appears that model biases in surface net radiation are significantly reduced when using the surface albedo observed by satellite for the current day. Figure 9 shows the temporal evolution of the bias for RN based on ISBA simulations using SAFRAN forcing, and using the LSA SAF radiative forcing (DSSF and DSLF), with respect to the in situ observations of RN. A visual inspection permits us to notice lower bias values when the LSA SAF radiation forcing is used by ISBA. The magnitude of the reduction of this bias often exceeds 50% for days having a significant daily bias (not shown).

Fig. 8.
Fig. 8.

Seasonal variations in 2006 of the daily RN over the FLUXNET station of Aurade from ground measurements (dots) and ISBA using SAFRAN radiative forcing (line).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Fig. 9.
Fig. 9.

As in Fig. 8, but for based on ISBA simulations using either SAFRAN or LSA SAF radiative forcing: between SAFRAN and ground measurement (gray line), and between LSA SAF forcing (DSSF + DSLF) and ground measurements (black dotted line).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Figure 10 depicts the monthly bias (in absolute units) of RN between in situ observations and the ISBA simulations using four radiative forcings: SAFRAN, LSA SAF DSSF and SAFRAN DSLF, LSA SAF DSLF and SAFRAN DSSF, and both LSA SAF DSLF and DSSF. The use of the satellite DSSF improves the RN simulations at all sites except for Hesse, where the bias increases. Improvements are significant for Lamasquere in the spring as the monthly bias decreases from 50 to 30 W m−2. The satellite DSLF product has a small positive or ever so slightly negative impact on simulated RN. The combination of the two satellite forcings gives slightly better results than the use of DSSF alone, even if the main factor for the bias reduction is the use of DSSF.

Fig. 10.
Fig. 10.

Variation of monthly absolute bias of RN for six FLUXNET stations based on ISBA simulations using various radiative forcings: SAFRAN forcing (Ref.), LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Figure 11 presents the SDD score of daily differences of RN, averaged on a monthly basis. There is a seasonal evolution of the SDD that can rise up to 40 W m−2 in summer and is around 20 W m−2 in winter. The use of the LSA SAF DSSF product tends to reduce the summertime high SDD values, and the SDD is relatively stable all along the year—around 20 W m−2.

Fig. 11.
Fig. 11.

As in Fig. 10, but for SDD.

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

The monthly averaged correlation scores (r) of the daily RN values are presented in Fig. 12. Similar conclusions as before can be drawn: if winter months are discarded, the use of the LSA SAF DSSF product induces a significant improvement of r, which is often greater than 0.8. For example, the correlation jumps from 0.5 to 0.8 from March to August at Aurade because of the LSA SAF DSSF. The impact of the DSLF product is less marked and can be negative in some cases. The low correlation obtained during winter months is simply due to RN values observed at this season—close to zero or slightly negative (see Fig. 8).

Fig. 12.
Fig. 12.

Variations in 2006 of the correlation (r) of daily RN over six FLUXNET stations based on ISBA simulations using various radiative forcings: SAFRAN forcing (Ref.), LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Figures 13 and 14 summarize these results. If the LSA SAF DSSF product is used, monthly averaged daily SDD values can decrease up to 20 W m−2, and 10 W m−2 on average for 2006 (see Table 4). Also, the correlation is higher, particularly in situations where the correlation values were poor in the reference case (Fig. 14). Table 5 presents yearly statistics. It should be noticed that the use of the DSSF and DSLF products induces a 0.88 to 0.96 increase of the RN correlation score.

Fig. 13.
Fig. 13.

Bias (absolute value) and SDD statistical scores between FLUXNET measurement and simulated energy fluxes with ISBA. Reduction (y axis) of the statistical scores due to the use of LSA SAF satellite forcing are compared to statistics obtained with SAFRAN forcing (x axis). Monthly averaged daily bias and SDD of RN, LE, and H. One point represents 1 of the 12 monthly statistical scores for one of the six FLUXNET stations (see Figs. 11 and 12 for RN). Satellite forcings are LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Fig. 14.
Fig. 14.

Correlation (r) between FLUXNET measurement and simulated energy fluxes with ISBA. Improvement (y axis) of the R statistical score due to the use of LSA SAF satellite forcing is compared to the R obtained with SAFRAN forcing (x axis). The daily R of RN, LE, and H are monthly averaged. One point represents one of the 12 monthly statistical scores for one of the six FLUXNET stations (see Fig. 13 for RN). Satellite forcing are LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF).

Citation: Journal of Hydrometeorology 13, 2; 10.1175/JHM-D-11-059.1

Table 4.

Bias and SDD for 2006 of measured and simulated RN flux on a daily basis at six FLUXNET sites over France. Two forcings are used for the simulations: SAFRAN forcing and LSA SAF DSSF forcing.

Table 4.
Table 5.

The 2006 mean bias, SDD, and correlation coefficient (r) of simulated energy fluxes compared on a daily basis with in situ measurement over the six stations. Radiative forcings are SAFRAN (reference), LSA SAF solar incoming flux (DSSF), LSA SAF infrared incoming flux (DSLF), and LSA SAF solar and infrared incoming flux (DSSF + DSLF). No closure correction is applied.

Table 5.

4) LE and H heat fluxes

Figures 13 and 14 also show the impact of the use of satellite forcing on LE and H. The impact is positive, even if it is less significant than for RN. The SDD of the simulated H flux decreases by up to 15 W m−2 and the correlation markedly improves when the LSA SAF DSSF product is used (Fig. 13 and 14). Considering the six FLUXNET stations, the correlation scores of daily H and LE heat fluxes in 2006 increase from 0.66 to 0.71, and from 0.58 to 0.62, respectively (Table 5).

The DSLF product has no significant impact on the scores. Table 5 summarizes these statistics for 2006 at the six stations. The energy partitioning between H and LE depends on the fraction of vegetation and on soil moisture. If the latter are not correctly represented in the model, then large discrepancies may occur through a comparison with ground-measured fluxes. This is potentially the reason why the impact on heat fluxes (H or LE) is moderate in comparison with the impact on RN. Another potential reason is that the eddy-covariance technique does not always produce perfect energy closure (Twine et al. 2000). In the presented work, no closure correction is done. The inability of the eddy-covariance technique to achieve energy budget closure at several FLUXNET sites was reported by Wilson et al. (2002). Kucharik et al. (2006) argued that observations could be adjusted to force energy. Based on the observed energy partitioning, Kucharik and Twine (2007) attributed approximately 64%–72% of the imbalance to LE. To investigate this issue, the same closure correction was applied by partitioning the energy imbalance into sensible and latent heat fluxes according to the observed ratio of H and LE (Kucharik and Twine 2007). A poor impact was noticed (see Table 6).

Table 6.

As in Table 5, but a closure correction is applied.

Table 6.

4. Conclusions

The paper presents a comparison between SAFRAN and LSA SAF shortwave (DSSF) and longwave (DSLF) incoming radiations. Mean statistics over France and for the year 2006 show that SAFRAN and LSA SAF are quite consistent. The mean average bias in the 2006 year is 3.75 W m−2 for DSSF and 2.61 W m−2 for DSLF, which represent 2.5% and 0.8% of their average values, respectively. For 2006, the average daily SDD for DSSF and DSLF values are 34 and 17 W m−2, respectively. Significant discrepancies could be noticed locally. The comparison indicates an important day-to-day evolution of these statistics. Also, SAFRAN underestimates the incoming solar radiation at midday (around 50 W m−2) in comparison with LSA SAF. Statistics of differences of incoming solar radiation also increase with altitude, especially from 0 to 200 m. Finally, an analysis of the daily DSLF cycle questions the representation of the diurnal cycle in SAFRAN: the maximum of incoming infrared radiation is observed at 1500 UTC with MSG–SEVIRI, whereas SAFRAN indicates a minimum at the same time. In summer, convective clouds usually appear around 1500 UTC, which tends to cast confidence in the LSA SAF product. Nevertheless, it should be stressed that this behavior also appears at wintertime, when it cannot be caused by convective clouds. Absolute value of bias (SDD) of SAFRAN daily incoming solar radiation over validation stations are around 9.22 (35) W m−2 during 2006. The LSA SAF DSSF product is more accurate on a daily basis, with bias and SDD values of 5.31 and 18 W m−2, respectively. The lack of measurement of downward longwave radiation did not allow a proper assessment of this variable.

The impact on LSM simulation of radiation discrepancies between SAFRAN and LSA SAF was examined. The performance of SAFRAN analysis related to the presented study compares well with results obtained by Quintana-Seguí et al. (2008), Szczypta et al. (2011), and Vidal et al. (2009). This 8-km-resolution atmospheric analysis over France is accurate enough to obtain an LSM-derived net radiation with a mean monthly bias generally lower than 30 W m−2. Nevertheless, the use of LSA SAF radiative forcing improves the simulation of the surface energy fluxes. When satellite data are considered, SDD of net radiation (RN) simulated with SURFEX model can decrease by 20 W m−2 for monthly averaged SDD. The bias is slightly reduced and the correlation (r) is strongly improved on a monthly basis (up to 0.4). The use of satellite DSSF or DSFL forcings have also a positive impact on the simulated sensible heat flux and a moderate impact on the simulated latent heat flux. Discrepancies between SAFRAN and the LSA SAF incoming solar radiation could explain, in some places, 25 W m−2 of the daily RN bias. The SDD of the sensible heat flux (H) over validation stations could decrease by 15 W m−2 and the correlation score slightly increase. This study shows the importance of continuous in situ measurement, in particular the existence of the FLUXNET network, for the verification of gridded atmospheric surface variables.

The use of the satellite radiative forcing has a significant impact on simulated surface and soil temperatures (up to 1 K on the yearly mean over western France). The impact on the simulated surface soil water content is not negligible: up to 5%. On average for 2006, over the France domain, a mean difference of 2.5% on DSSF and 0.8% on DSLF forcing affected the averaged values of TG1/TG2/WG1 by 3.5%/3.3%/0.5%, repectively. In conclusion, the DLSF and DSSF products appear to be of better quality than the radiative forcing derived from the SAFRAN analysis, with a similar spatial and temporal resolution. The SAFRAN analysis exists only over France, whereas the LSA SAF provides a uniform coverage over Europe, Africa, and South America. A lot of areas lack a high-resolution meteorological forcing. The LSA SAF products provide new and valuable information.

Acknowledgments

This study was performed in the framework of the EUMETSAT LSA SAF activities and of the French GICC CarboFrance Project “Impact des extrêmes climatiques sur les flux de carbone.” S. Lafont was supported by the GEOLAND2 project, which is cofunded by the European Commission within the GMES initiative in FP7. The authors thank E. Ceschia, A. Granier, A. Michelot, and S. Rambal for providing the FLUXNET data. We also feel particularly indebted to the anonymous reviewers for their helpful suggestions and comments, which improved the readability of the manuscript.

REFERENCES

  • Albergel, C., , Calvet J.-C. , , Gibelin A.-L. , , Lafont S. , , Roujean J.-L. , , Berne C. , , Traullé O. , , and Fritz N. , 2010: Observed and modeled gross primary production and ecosystem respiration of a grassland in southwestern France. Biogeosciences, 7, 16571668.

    • Search Google Scholar
    • Export Citation
  • Baldocchi, D. D., and Coauthors, 2001: FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Amer. Meteor. Soc., 82, 24152434.

    • Search Google Scholar
    • Export Citation
  • Béziat, P., , Ceschia E. , , and Dedieu G. , 2009: Carbon balance of a three crop succession over two cropland sites in South West France. Agric. For. Meteor., 149, 16281645.

    • Search Google Scholar
    • Export Citation
  • Calvet, J.-C., and Coauthors, 1999: MUREX: A land-surface field experiment to study the annual cycle of the energy and water budgets. Ann. Geophys., 17, 838854.

    • Search Google Scholar
    • Export Citation
  • Chen, T. H., and Coauthors, 1997: Cabauw experimental results from the project for intercomparison of land-surface parameterization schemes. J. Climate, 10, 11941215.

    • Search Google Scholar
    • Export Citation
  • Courtier, P., , Freydier C. , , Geleyn J.-F. , , Rabier F. , , and Rochas M. , 1991: The Arpège project at Météo France. Proc. ECMWF Seminar on Numerical Methods in Atmospheric Models, Vol. 2, Reading, United Kingdom, ECMWF, 193–232.

  • Diak, G. R., , Bland W. L. , , Mecikalski J. R. , , and Anderson M. C. , 2000: Satellite-based estimates of longwave radiation for agricultural applications. Agric. For. Meteor., 103, 349355.

    • Search Google Scholar
    • Export Citation
  • Douville, H., , Royer J. F. , , and Mahfouf J. F. , 1995: A new snow parameterization for the Météo-France climate model, Part I: Validation in stand-alone experiments. Climate Dyn., 12, 2135.

    • Search Google Scholar
    • Export Citation
  • Durand, Y., , Brun E. , , Mérindol L. , , Guyomarc’h G. , , Lesaffre B. , , and Martin E. , 1993: A meteorological estimation of relevant parameters for snow models. Ann. Glaciol., 18, 6571.

    • Search Google Scholar
    • Export Citation
  • Durand, Y., , Giraud G. , , Brun E. , , Merindol L. , , and Martin E. , 1999: A computer-based system simulating snowpack structures as a tool for regional avalanche forecasting. J. Glaciol., 45, 469484.

    • Search Google Scholar
    • Export Citation
  • Frouin, R., , Lingner D. W. , , Gautier C. , , Baker K. S. , , and Smith R. C. , 1989: A simple analytical formula to compute clear sky total and photosynthetically available solar irradiance at the ocean surface. J. Geophys. Res., 94, 97319742.

    • Search Google Scholar
    • Export Citation
  • Gandin, L. S., 1965: Objective Analysis of Meteorological Fields (in Russian). Isreal Program for Scientific Translations, 242 pp.

  • Geiger, B., , Meurey C. , , Lajas D. , , Franchistéguy L. , , Carrer D. , , and Roujean J.-L. , 2008: Near real-time provision of downwelling shortwave radiation estimates derived from satellite observations. Meteor. Appl., 15, 411420.

    • Search Google Scholar
    • Export Citation
  • Gibelin, A.-L., , Calvet J.-C. , , Roujean J.-L. , , Jarlan L. , , and Los S. O. , 2006: Ability of the land surface model ISBA-A-gs to simulate leaf area index at the global scale: Comparison with satellites products. J. Geophys. Res., 111, D18102, doi:10.1029/2005JD006691.

    • Search Google Scholar
    • Export Citation
  • Granier, A., , Bréda N. , , Longdoz B. , , Gross P. , , and Ngao J. , 2008: Ten years of fluxes and stand growth in a young beech forest at Hesse, North-eastern France. Ann. For. Sci., 65, 704, doi:10.1051/forest:2008052.

    • Search Google Scholar
    • Export Citation
  • Henderson-Sellers, A., , Yang Z.-L. , , and Dickinson R. E. , 1993: The project for intercomparison of land–surface parameterization schemes. Bull. Amer. Meteor. Soc., 74, 13351349.

    • Search Google Scholar
    • Export Citation
  • Ineichen, P., , Barroso C. , , Geiger B. , , Hollmann R. , , Marsouin A. , , and Mueller R. , 2009: Satellite Application Facilities irradiance products: Hourly time step comparison and validation over Europe. Int. J. Remote Sens., 30, 55495571.

    • Search Google Scholar
    • Export Citation
  • Josey, S. A., , Pascal R. W. , , Taylor P. K. , , and Yelland M. J. , 2003: A new formula for determining the atmospheric longwave flux at ocean surface at mid-high latitudes. J. Geophys. Res., 108, 3108, doi:10.1029/2002JC001418.

    • Search Google Scholar
    • Export Citation
  • Kucharik, C. J., , and Twine T. E. , 2007: Residue, respiration, and residuals: Evaluation of a dynamic agroecosystem model using eddy flux measurements and biometric data. Agric. For. Meteor., 146 (3–4), 134158.

    • Search Google Scholar
    • Export Citation
  • Kucharik, C. J., , Barford C. , , El Maayar M. , , Wofsy S. C. , , Monson R. K. , , and Baldocchi D. D. , 2006: A multiyear evaluation of a Dynamic Global Vegetation Model at three AmeriFlux forest sites: Vegetation structure, phenology, soil temperature, and CO2 and H2O vapor exchange. Ecol. Modell., 196, 131.

    • Search Google Scholar
    • Export Citation
  • Le Moigne, P., 2002: Description de l’analyse des champs de surface sur la France par le système SAFRAN (Description of the analysis of near-surface atmospheric fields over France with Safran system). CNRM/GAME Tech. Note 77, 30 pp.

  • Le Moigne, P., Ed., 2009: SURFEX scientific documentation. Centre du Groupe de Météorologie à Moyenne Echelle Tech. Note 87, 211 pp. [Available online at http://www.cnrm.meteo.fr/surfex/.]

  • Martin, E., and Coauthors, 2007: Le code de surface externalisé SURFEX de Météo-France. Extended Abstracts, Ateliers de Mod´elisation de l’Atmosphère, Toulouse, France, 4 pp.

  • Masson, V., , Champeaux J. L. , , Chauvin F. , , Meriguet C. , , and Lacaze R. , 2003: A global database of land surface parameters at 1-km resolution in meteorological and climate models. J. Climate, 16, 12611282.

    • Search Google Scholar
    • Export Citation
  • Michelot, A., , Eglin T. , , Dufrêne E. , , Lelarge-Trouverie C. , , and Damesin C. , 2010: Comparison of seasonal variations in water-use efficiency calculated from the carbon isotope composition of tree rings and flux data in a temperate forest. Plant, Cell Environ., 34, 230244, doi:10.1111/j.1365-3040.2010.02238.x.

    • Search Google Scholar
    • Export Citation
  • Noilhan, J., , and Planton S. , 1989: A simple parameterization of land surface processes for meteorological models. Mon. Wea. Rev., 117, 536549.

    • Search Google Scholar
    • Export Citation
  • Noilhan, J., , and Mahfouf J. F. , 1996: The ISBA land surface parametrisation scheme. Global Planet. Change, 13, 145159, doi:10.1016/0921-8181(95)00043-7.

    • Search Google Scholar
    • Export Citation
  • Otkin, J., , Anderson M. C. , , Mecikalski J. R. , , and Diak G. R. , 2005: Validation of GOES-based insolation estimates using data from the U.S. Climate Reference Network. J. Hydrometeor., 6, 460475.

    • Search Google Scholar
    • Export Citation
  • Pinker, R. T., , Zhang B. , , and Dutton E. G. , 2005: Do satellites detect trends in surface solar radiation? Science, 308, 850854.

  • Prata, A. J., 1996: A new long-wave formula for estimating downward clear-sky radiation at the surface. Quart. J. Roy. Meteor. Soc., 122, 11211151.

    • Search Google Scholar
    • Export Citation
  • Quintana-Seguí, P., and Coauthors, 2008: Analysis of near surface atmospheric variables: Validation of the SAFRAN analysis over France. J. Appl. Meteor. Climatol., 47, 92107.

    • Search Google Scholar
    • Export Citation
  • Rambal, S., , Joffre R. , , Ourcival J. M. , , Cavender-Bares J. , , and Rocheteau A. , 2004: The growth respiration component in eddy CO2 flux from a Quercus ilex Mediterranean forest. Global Change Biol., 10, 14601469.

    • Search Google Scholar
    • Export Citation
  • Ritter, B., , and Geleyn J. F. , 1992: A comprehensive radiation scheme for numerical weather prediction models with potential applications in climate simulations. Mon. Wea. Rev., 120, 303325.

    • Search Google Scholar
    • Export Citation
  • Salgado, R., , and Le Moigne P. , 2010: Coupling of the FLake model to the Surfex externalized surface model. Boreal Environ. Res., 15, 231244.

    • Search Google Scholar
    • Export Citation
  • Sarrat, C., and Coauthors, 2009: Mesoscale modeling of the CO2 interactions between the surface and the atmosphere applied to the April 2007 CERES field experiment. Biogeosciences, 6, 633646.

    • Search Google Scholar
    • Export Citation
  • Simmons, A., , Uppala S. , , Dee D. , , and Kobayashi S. , 2006: ERA-I: New ECMWF reanalysis products from 1989 onwards. ECMWF Newsletter, No. 110, ECMWF, Reading, United Kingdom, 25–35.

  • Szczypta, C., , Calvet J. , , Albergel C. , , Balsamo G. , , Boussetta S. , , Carrer D. , , Lafont S. , , and Meurey C. , 2011: Verification of the new ECMWF ERA-Interim reanalysis over France. Hydrol. Earth Syst. Sci., 15, 647666, doi:10.5194/hess-15-647-2011.

    • Search Google Scholar
    • Export Citation
  • Trigo, I. F., , Barroso C. , , Viterbo P. , , Freitas S. C. , , and Monteiro I. T. , 2010: Estimation of downward longwave radiation at the surface combining remotely sensed data and NWP data. J. Geophys. Res., 115, D24118, doi:10.1029/2010JD013888.

    • Search Google Scholar
    • Export Citation
  • Trigo, I. F., and Coauthors, 2011: The Satellite Application Facility for Land Surface Analysis. Int. J. Remote Sens., 32, 27252744.

  • Twine, T. E., and Coauthors, 2000: Correcting eddy-covariance flux underestimates over a grassland. Agric. For. Meteor., 103, 279300.

    • Search Google Scholar
    • Export Citation
  • Vidal, J.-P., , Martin E. , , Franchistéguy L. , , Baillon M. , , and Soubeyroux J.-M. , 2009: A 50-year high-resolution atmospheric reanalysis over France with the Safran system. Int. J. Climatol., 30, 16271644, doi:10.1002/joc.2003.

    • Search Google Scholar
    • Export Citation
  • Wielicki, B. A., , Barkstrom B. R. , , Harrison E. F. , , Lee III R. B. , , Smith G. L. , , and Cooper J. E. , 1996: Clouds and the Earth’s Radiant Energy System (CERES): An Earth Observing System experiment. Bull. Amer. Meteor. Soc., 77, 853868.

    • Search Google Scholar
    • Export Citation
  • Wilson, K. B., and Coauthors, 2002: Energy partitioning between latent and sensible heat flux during the warm season at FLUXNET sites. Water Resour. Res., 38, 1294, doi:10.1029/2001WR000989.

    • Search Google Scholar
    • Export Citation
  • Zhao, W., , Kuhn W. R. , , and Drayson S. R. , 1994: The significance of detailed structure in the boundary layer to thermal radiation at the surface in climate models. Geophys. Res. Lett., 21, 16311634.

    • Search Google Scholar
    • Export Citation
Save