A satellite stand-alone procedure for deriving net radiation by using SEVIRI and MODIS products

https://doi.org/10.1016/j.jag.2018.08.018Get rights and content

Highlights

  • Independent ground measurements procedure to assess net radiation at regional level.

  • MODIS and MSG-SEVIRI time series products for the estimation of net radiation.

  • Estimated net radiation and its components are consistent with ground measurements.

  • Ability to derive spatial and temporal variability of net surface radiation in Italian territory.

Abstract

In this study, a new stand-alone satellite approach for the estimation of net surface radiation (Rn) has been implemented and validated for the Italian territory. The method uses the MODIS and MSG-SEVIRI time series products and it is independent of the use of ancillary data (i.e. ground measurements). A database of daily measurements of Rn, provided by 9 stations of the FLUXNET network, was used to validate the method in different ecological scenarios in the period 2010-12.

The Rn modelled by the proposed approach and the corresponding FLUXNET measurements were in good agreement, with RMSE and R2 of 19.8 Wm−2 and 0.87, respectively, at 8-days scale, and 23.3 Wm−2 and 0.92, respectively, at daily scale.

Therefore, the proposed approach can be considered effective for the estimation of spatial and temporal variability of Rn, which is a key variable related to the management of water resources, agriculture, ecology and climate change.

Introduction

Net surface radiation (Rn) is the main driver for energy balance involved Earth processes, encompassing the total energy available for most of physical and biological activities (Carmona et al., 2015; Pan et al., 2015; Soltani et al., 2017). In numerical terms, Rn consists of the sum of 4 components: the downward/incoming short (Rss↓) and long (Rsl↓) wave radiation and the upward/outgoing long (Rsl↑) and short (Rss↑) wave radiation. An accurate determination of Rn is critical for the assessment of Rn-dependent processes (such as evapotranspiration, ET), since any error in its estimation will derive in miscalculations. Rn calculation is not an easy task, since multiple time and space changing atmospheric and land surface parameters are involved in it, including surface albedo (α), surface emissivity (ε0), and land surface temperature (Ts).

Traditionally, Rn has been measured using field instruments (such as, pyrradiometer, pyranometer, pyrgeometer and net radiometer). Several ground-based networks for radiation measurements are spread around the world, as Global Energy Balance Archive (GEBA, Ohmura et al., 1989; Gilgen and Ohmura, 1999); Word Radiation Monitoring Center - Baseline Surface Radiation Network (WRMC-BSRN, Ohmura et al., 1998); Surface Radiation Budget Network (SURFRAD, Augustine et al., 2000, 2005); and FLUXNET (Baldocchi et al., 2001). In particular, GEBA is a database for surface energy fluxes measurements, maintained by the Institute for Climate and Atmospheric Sciences at ETH Zurich. It contains more than 2500 stations worldwide distributed, providing monthly averages of the surface energy balance components. BSRN became operational in 1992 and it consists on 48 stations placed all around the world in different climatic zones that provide observations of the short- and long-wave surface radiation fluxes with a high sampling rate. Established in 1993, SURFRAD is composed by 7 operational stations that provide long-term measurements of the surface radiation budget over the United States. FLUXNET global network consists of more than 500 micro-meteorological tower sites that measure the exchanges of carbon dioxide, water vapour, and surface energy fluxes between terrestrial ecosystems and the atmosphere.

Nevertheless, the data provided by ground measurements refers only to point-scale and their quality is greatly affected by instrumental error and/or maintenance faults (uncorrected sensor configurations or calibration), and land surface and atmosphere heterogeneities (Llasat and Snyder, 1998; Wilson et al., 2002; Foken et al., 2006; Foken, 2008; Jacobs et al., 2008). In addition, it is difficult to extend the measurements made at specific locations up to regional scales (Blad et al., 1998), due to the strong influence of surface heterogeneity. Despite these limitations, the quality assurance and the data quality control performed at these networks allow their use for validating and evaluating satellite-based estimates of surface radiative fluxes.

Unlike ground-based observations, remotely sensed observations have the advantage of global coverage. In the last decades, a wide number of radiation products derived from satellite has been developed, highlighting: Earth Radiation Budget Experiment (ERBE, Smith et al., 1977; Jacobowitz et al., 1979, 1984; Barkstrom and Smith, 1986); Clouds and the Earth’s Radiant Energy System (CERES, Wielicki et al., 1998); International Satellite Cloud Climatology Project (ISCCP, Zhang et al., 2004); Geostationary Earth Radiation Budget Project (GERB, Harries et al., 2005); and Surface Radiation Budget project (SRB, Cox et al., 2006). In addition, several surface radiation products have been derived from multispectral sensors such as Spinning Enhanced Visible and Infrared Imager (SEVIRI) aboard METEOSAT Second Generation (MSG) satellites; Visible Infrared Spin Scan Radiometer (VISSR) and Advanced Baseline Imager (ABI) on board Geostationary Operational Environmental Satellites (GOES); and Moderate Resolution Imaging Spectroradiometer (MODIS) aboard TERRA and AQUA satellites. However, one of the main limitation of approaches estimate Rn by using satellite data is that they usually requires ancillary data (i.e. ground measurements, such as maximum and minimum air temperature).

Aiming to develop a new approach for actual ET assessment, Minacapilli et al. (2016) proposed and tested, in an Italian region (Sicily), a new method based on a combined and stand-alone use of MODIS and MSG-SEVIRI products able to estimate Rn and ET without the use of ancillary data (i.e. ground measurements).

The objective of this work was to evaluate the satellite stand-alone approach for deriving Rn, proposed by Minacapilli et al., (2016), over different climate/land-use sites and at different temporal resolution (daily and 8-day scale). The obtained Rn product, as well as its singular components, was validated using the radiation fluxes provided by 9 stations of FLUXNET network spread along Italy. A step-forward consists of testing the sensitivity of the algorithm in function of site-dependent parameters (ε0, Ts, EVI, elevation) and their relations.

Section snippets

Rn model description

The estimation of Rn (W m−2) is obtained by summing up the net surface shortwave (Rns) and longwave (Rnl) radiation, as follows:Rn=Rns+Rnl=Rss-Rss+Rsl-Rsl-(1-ε0)Rslwhere, Rss↓ is the incoming shortwave radiation emitted by the sun and transmitted through atmosphere (W m−2); Rss↑ is the outgoing shortwave radiation (W m−2) depending on Rss↓ and surface albedo (α); Rsl↓ is the downward longwave radiation (W m−2), Rsl↑ is the upward longwave radiation (W m−2), and εo is the surface thermal

Results

This section is divided into three subsections. The first one is focused on the evaluation of the Rn estimates by the proposed model (and its components) with respect to the radiation fluxes measured at the selected Italian FLUXNET demo-sites both at daily and 8-day temporal scale. Second, the results of the sensitivity analysis of the main parameters involved in Rn estimation are reported. Finally, in the third subsection, some examples of final outputs retrieved by the model are showed.

Discussion

The proposed model has provided accurate estimates of Rn both at daily and 8-day time scale, with a RMSE and a MAE of 23.3 and 17.6 Wm−2 for daily scale; whereas these values were of 25.4 and 19.8 Wm−2 for a 8-day scale. Similar accuracies have been obtained by other authors under clear sky conditions (Bisht et al., 2005; Geiger et al., 2008; Journee and Bertrand, 2010; Hou et al., 2014; Gómez et al., 2016). In view of the results, no large differences among the different temporal scales in

Conclusion

The major key-points of the proposed approach can be summarized as in the follows:

  • -

    The procedure requires only remote sensing products, and therefore it is completely independent from ancillary ground-based measurements and benefits of the recent MODIS and MSG-SEVIRI time-series products;

  • -

    The simulated values of Rn and its components resulted consistent and comparable with the local ground measurements, therefore the procedure might be used in regions where radiation measurements are missing or

Acknowledgments

This workused eddy covariance data acquired and shared by the FLUXNET community, including these networks: CarboEuropeIP, CarboItaly, CarboMont. The ERA-Interim reanalysis data are provided by ECMWF and processed by LSCE. The FLUXNET eddy covariance data processing and harmonization was carried out by the European Fluxes Database Cluster and Fluxdata project of FLUXNET (GHG-Europe EU-FP7, EuroFlux EU-FP4, CarboEuroFluxEU-FP5).

The authors thank the EU and the Italian Ministry of Education,

References (49)

  • B. Marcolla et al.

    Climatic controls and ecosystem responses drive the inter-annual variability of the net ecosystem exchange of an alpine meadow

    Agric. For. Meteorol.

    (2011)
  • M. Minacapilli et al.

    A time domain triangle method approach to estimate actual evapotranspiration: application in a Mediterranean region using MODIS and MSG-SEVIRI products

    Remote Sens. Environ.

    (2016)
  • K.B. Wilson et al.

    Energy balance closure at FLUXNET sites

    Agric For. Meteorol.

    (2002)
  • J.A. Augustine et al.

    SURFRAD – a national surface radiation budget network for atmospheric research

    Bull. Am. Meteorol. Soc.

    (2000)
  • J.A. Augustine et al.

    An update on SURFRAD – the GCOS surface radiation budget network for the continental United States

    J. Atmos. Ocean. Technol.

    (2005)
  • D. Baldocchi et al.

    FLUXNET: a new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities

    Bull. Am. Meteorol. Soc.

    (2001)
  • B.R. Barkstrom et al.

    The earth radiation budget experiment: science and implementation

    Rev. Geophys.

    (1986)
  • B.L. Blad et al.

    Estimating net radiation with remotely sensed data: results from KUREX‐91 and FIFE studies

    Remote. Sens. Rev.

    (1998)
  • N.A. Brunsell et al.

    Validating remotely sensed land surface fluxes in heterogeneous terrain with large aperture scintillometry

    Int. J. Remote Sens.

    (2011)
  • A. Cescatti et al.

    Drag coefficient and turbulence intensity in conifer canopies

    Agric. For. Meteorol.

    (2014)
  • S.J. Cox et al.

    The NASA/GEWEX surface radiation budget project: overview and analysis

    Proceedings of the 12th Conference on Atmospheric Radiation

    (2006)
  • M. Derrien et al.

    MSG/SEVIRI cloud mask and type from SAFNWC

    Int. J. Remote Sens.

    (2005)
  • T. Foken

    The energy balance closure problem – an overview

    Ecol. Appl.

    (2008)
  • T. Foken et al.

    Some aspects of the energy balance closure problem

    Atmos. Chem. Phys. Discuss.

    (2006)
  • Cited by (13)

    • Daily grass reference evapotranspiration with Meteosat Second Generation shortwave radiation and reference ET products

      2021, Agricultural Water Management
      Citation Excerpt :

      Furthermore, the increasing number of satellite products that are freely available make these products a practical and cost-effective means of gathering information over land surfaces for a wide range of applications, including the nearly continuous assessment of vegetation state and water stress (Trigo et al., 2011). The accuracy of remote sensing solar radiation products has been widely evaluated (Roerink et al., 2012; Bojanowski et al., 2014; Mokhtari et al., 2018; Ramírez-Cuesta et al., 2018). However, some constraints have been reported, namely those referring to rapidly changing cloud sky conditions (Geiger et al., 2008; Macke et al., 2010; Cristóbal and Anderson, 2013; Carrer et al., 2019).

    • Energy balance and irrigation performance assessments in lemon orchards by applying the SAFER algorithm to Landsat 8 images

      2021, Agricultural Water Management
      Citation Excerpt :

      Besides variations of these balances with the weather, they will also depend on the root-zone moisture levels, which in turn are related to irrigation water management. Due to limitations of field measurements at the irrigation scheme scale, up-scaling tools based on remote sensing from satellite images are important for crop water management decisions (Kamble et al., 2013; Zheng et al., 2016; Wagle et al., 2016; Bhattarai et al., 2017b, Ramírez-Cuesta et al., 2018; Tazekrit et al., 2018; Holtzman et al. 2018; Nyoley et al., 2019; Teixeira et al., 2019; Santos et al., 2020; Mhawej et al., 2020a, 2020b), which, from the knowledge of the authors of the current paper, have not yet been performed on farm scales for lemon orchards covering different irrigation systems. A generalized use of these tools requires a deep knowledge of biophysical variables and their relations with remote sensing parameters (Cancela et al., 2019).

    • Estimation of all-sky all-wave daily net radiation at high latitudes from MODIS data

      2020, Remote Sensing of Environment
      Citation Excerpt :

      Satellite remote sensing can continuously and synoptically retrieve Rn with better spatial and temporal coverage as well as reasonable accuracy (Bisht and Bras, 2010; Bisht et al., 2005; Brown and Caldeira, 2017; Huang et al., 2016; Jiang et al., 2018; Jiang et al., 2016; Jung et al., 2019; Wang et al., 2015a). Algorithm development applied to the estimation of Rn at regional and global scales has attracted considerable attention in recent years and achieved significant advances (Amatya et al., 2015; Bisht and Bras, 2010; Jiang et al., 2015; Liang et al., 2019; Mira et al., 2016; Moukomla and Blanken, 2016; Ramírez-Cuesta et al., 2018; Wang et al., 2015a; Zhao et al., 2019; Zhong et al., 2019). The algorithms widely used and validated can be divided into three classes: 1) parameterization algorithms, 2) hybrid algorithms, and 3) empirical algorithms.

    • A satellite system for monitoring the radiation budget at the Baltic Sea surface

      2020, Remote Sensing of Environment
      Citation Excerpt :

      Numerous comparative analyses performed by various authors for different regions of the world have shown that, in some cases, global algorithm-based models utilising satellite data substantially over- or underestimate values of radiation budget components (Posselt et al., 2012; Pan et al., 2015; Zhang et al., 2015), depending on season, measurement location and the kind of input data used. The greatest accuracy is achieved using local solutions, based on satellite algorithms and validated for particular regions: this applies to both long- and shortwave components (Krężel, 1985, Zapadka et al., 2007, Flerchinger et al., 2009, Behar et al., 2015, Quej et al., 2016, Ramirez-Cuesta et al., 2018, Kim and Lee, 2018). Hence, the adaptation of hitherto used models to local conditions is an amply justified approach to improving their accuracy.

    View all citing articles on Scopus
    View full text