1. Sophia Antipolis, 17 octobre 2003 HELIOSAT-II

2. What is Heliosat? Raw image Irradiation (energy)

3. Why Heliosat-II? Heliosat 0 (Cano et al. 1986) Heliosat-I (customised versions) Need to Improve accuracy Improve reliability Ease implementation

4. Achievements of Heliosat-II No more parameter to tune (to be checked)  Easier to implement  Reliability in irradiation assessment More explicit physical modelling  Improvement possible and easy More accurate

5. Accuracy (relative RMSE) Type Irradiation (Wh m -2 ) Type Irradiance (W m -2 ) Hourly values100Hour100 Daily values600Day25 5-days sum of daily irradiation 25005-days20 10-days sum of daily irradiation 350010-days15 Monthly mean of hourly irradiation 50Month12 Monthly mean of daily irradiation 300 Single pixel. Hi-Res images. 30 and 60 stations WMO in Europe

6. How it works ? Same principle than Heliosat-0 and other methods (e.g., DWD)  t (i,j) -  t g (i,j) n t (i,j) = -------------------  t cloud -  t g (i,j) n is an attenuation: 1 – transmittance, of TOA irradiation => Irradiation at ground level

7. Kc = f(n) G = K c G c n  -0,2K c = 1,2 -0,2  n  0,8K c = 1-n 0,8  n  1,1K c = 2,0667 - 3,6667 n+1,6667 n² n  1,1K c = 0,05 How it works ? (2)

8. Prerequisite - Reflectance Given a radiance L, the reflectance is given by

9. Prerequisite – Linke turbidity factor TL : optical turbidity of the clear-sky (aerosols and WV) Transmittance  exp[ - k TL  Rayleigh ] TL = 1 => pure atmosphere TL = 5 => polluted atmosphere Europe, TL  3 – 3.5

10. Prerequisite – Basic modelling Clear sky – Broadband L sat = T atm L g + L atm L atm = path radiance  sat = T atm  g +  atm T atm = T down T up Downwelling T up L atm  Sensor Reflected L g T down

11.  t sat (i,j) =  t atm (  S,  v,  ) +  t g (i,j) T t (  S ) T t (  v ) L atm = (D c /  ) (I 0met / I 0 ) ( / cos  V ) 0,8  * t (i,j) =  t sat (i,j) -  atm (  S,  v,  )] / T(  S ) T(  v ) Atmospheric Reflectance (2)

12. L atm [W.m -2.sr -1 ] Modelling Modtran Modelling as a function of D c vis 50 km = Ground Albedo,  g

13. Cloud index  atm Linke TL Elevation z Reflectance  sat  g nGh AtmosphereIrradiation I0met Calib. coeff General scheme

14. calibration Lsat calcul_albedo Calibration and reflectance Calibration coefficients I0met  sat

15. Atmospheric reflectance Latm calcul_albedo I0met  atm calcul_Latm Elevation z Linke TL

16. Cloud index  atm calcul_rho_rhoc Elevation z Linke TL  sat cc **  g gg n n glitter calcul_n

17. Irradiation calcul_Kc Elevation z Linke TL KcKc KcGh Calcul_Gh Correction_Kc n

18.  G h  G c  n  Albedos of ground  g and clouds  c are important parameters n =  -  g  c -  g if gg gg = 10 %    c has a systematic influence upon n Accuracy on G h is linked to that of n    g has a influence small for overcast skies [2 - 5 %] large for clear skies [20 - 50 %] {  Definition of albedoes is very important cc cc = 10 % if  [10 - 20 %] nn n   Influence of Albedos Assessment

19.  * t (i,j) =  t sat (i,j) -  atm (  S,  v,  )] / T(  S ) T(  v ) Ground albedo (1)  atm and T(  S ) and T(  v ) are computed from the clear-sky model. Here, the ESRA model

20.  t eff (i,j) = 0.78 – 0.13  - exp  cos  S ) 5 ]  t cloud (i,j) =  t eff (i,j) -  t atm (  S,  v,  )] / T(  S ) T(  v )  t cloud (i,j) > 0.2, otherwise  t cloud (i,j) = 0.2 and  t cloud (i,j) < 2.24  eff (i,j), otherwise  t cloud (i,j) = 2.24  t eff (i,j) Cloud Albedo From Taylor and Stowe (1984, JGR), using maxima of time-series of  *

21. HelioClim Data since 1985 and on-going Covering the whole field-of-view of Meteosat, except limits Available through the SoDa service http://www.soda-is.com

22. VALIDATION Two years of ground measurements over Europe and Africa (90 sites) for 1994 and 1995 Cell size: 5' arc angle IRRADIANCE (W m-2) Daily average. 30 949 values. Correlation: 0.94 Mean value: 192 — Bias: -1 (0 %) — RMSE: 35 (18 %) Monthly average. 1 005 values. Correlation: 0.96 Mean value: 195 — Bias: -2 (1 %) — RMSE: 23 (12 %)

23. This Summer More validation. 324 000 values from 1985 to 1990 162 000 values from 1991 to 1993 126 000 values after 1994 (out previous ones) Then routine validation in co-operation with Meteo- France Preparation of the operational chain for Meteosat Second Generation, in cooperation with DLR, Eumetsat and other European partners

24. Long Wave Radiation Algorithms exist for computing downward and upward LW radiation from SW and other parameters. May be used in conjunction with HelioClim Typical errors (RMSE) are: L  Downward LW: 20 - 25 W/m2 L  Upward LW: 25 – 30 W/m2 (depends strongly on the wind) R Radiative Balance: 25 – 30 W/m2 R = L  - L  + I (1-  )