Newly Developed Fraction of Diffuse and Improved Total Downwelling Surface Shortwave Flux

Results of the validation over the period of two years


The information on incoming solar radiation is important not only in climatological and meteorological models, but also in the energy sector and agriculture. Since 2005 the downwelling surface short-wave radiation flux (DSSF) is operationally available from two products based on MSG/SEVIRI: the MSG Downward Shortwave Surface flux (MDSSF, LSA-201) provides estimates every 30 minutes, while the Daily Downward Surface Shortwave Flux (DIDSSF, LSA-203)  provides daily integrals of the DSSF.

In the last years a new product was developed, the MSG Total and Diffuse Downward Surface Shortwave Flux (MDSSFTD, LSA-207) (Carrer, Ceamanos, et al. 2019; Carrer, Moparthy, et al. 2019). In addition to the DSSF, this new product also provides estimates of all sky fraction of diffuse irradiance (FD). Several auxiliary quantities are also provided: aerosol optical depth (AOD), opacity index (OI), and clearness index (Kt). Comparisons of the MDSSFTD product to ground station measurements is presented further.


The validation process

The diurnal total and diffuse down-welling surface flux components from the MDSSFTD product are compared against the same flux components derived from measurements provided by the Baseline Surface Radiation Network (BSRN) (Driemel, 2018) and the Slovenian NMS ground network. Data are separated into cloudy sky, clear sky and all sky (cloudy and clear sky combined) based on the MDSSFTD quality flaq. The Mean bias error (MBE) and relative mean bias error (rMBE) are then calculated. More details and results for 14 stations, with in depth discussion can be found in the respective Validation Report.

An example of such comparison for selected clear and partly cloudy days is displayed in Fig. 1 for the Bilje (Slovenia) site and in Fig. 2 for the Cener (Spain) site. There is an overall good agreement with in-situ for both DSSF and FD, not only for clear sky but also for cloudy sky.

For the period 2017 - 2018 scatter plots are presented for clear, cloudy and all sky conditions, in Fig. 3 for Bilje (Slovenia) and in Fig. 4 for Cener (Spain).

The DSSF and FD show a better agreement with in-situ for clear sky than for cloudy sky conditions. For clear sky, the DSSF is well correlated with the station measurements for all flux intensities. For cloudy sky, there is a higher spread for both DSSF and FD, but the majority of the estimates (warm colors) are well aligned with the ground measurements. Larger/smaller values of FD for clear/cloudy sky could indicate misclassified cloud conditions , suggesting that some improvement may be needed in the future.

Performance was also assessed through the mean bias error and relative mean bias error, as seen in Table 1. For selected thresholds of DSSF and FD of 200 W/m2 and 0.5, respectively, results for all sky conditions suggest errors with respect to ground stations below 10 %, for both MBE and rMBE.


Table 1: Statistical scores obtained from the comparison between MDSSFTD derived and in-situ measurements for total DSSF and respective diffuse fraction for the selected sites under all-sky conditions. Presented statistics are mean bias error (MBE) and relative mean bias error (rMBE). Number of samples for each statistic is noted with #.


Fig. 1: Diurnal cycle of the total DSSF (left panels) and respective diffuse fraction (FD; right panels) of the MDSSFTD product, for a cloud free (top panels) and cloudy day (bottom panels) in comparison with in-situ measurements for the Bilje site in Slovenia. Dots represent satellite data, with yellow/blue color for clear/cloudy sky conditions. Red crosses represent in-situ measurements. The exact dates are indicated in labels.


Fig. 2: The same as Fig. 1 but for the Cener site in Spain.





Fig. 3: Scatterplots of MDSSFTD data (x-axis) compared against in-situ data (y-axis) for the period 2017 – 2018, for the Bilje site in Slovenia. Top panels show the total DSSF and bottom panels show the respective diffuse fraction. Additionally, from left to right are sky conditions (all sky, clear sky, cloudy sky). Colors represent the density of points, with warmer colors for higher density.


Fig 4: Same as Figure 3, but for the Cener site in Spain.



We have presented results of the comparison between total and diffuse downwelling surface short-wave radiation fluxes against two stations in different ground networks. Good performance backed by a statistical validation enables MDSSFTD to become operationally available. Product LSA-207 will supersede LSA-201 in 2021 and will continue to be upgraded in the future. Values that are comparable to ground stations can be of great value at remote areas. An important additional part of the MDSSFTD product is the fraction of diffuse flux (FD) which is newly available.