Assimilating satellite cloud information with an Ensemble Kalman Filter at the convective scale

1. Annika Schomburg, Christoph Schraff, Hendrik Reich, Roland Potthast Assimilating satellite cloud information with an Ensemble Kalman Filter at the convective scale EnKF workshop 18-22 May 2014, Buffalo

2. Motivation: Weather situation 23 October 2012 12 UTC synoptic situation: stable high pressure system over central Europe

3. Motivation: Weather situation 23 October 2012 12 UTC synoptic situation: low stratus clouds over Germany Satellite cloud type classification

4. Motivation: Verification for 23 October 2012 12 hour forecast from 0:00 UTC Low cloud cover: COSMO-DE versus satellite Total cloud cover: COSMO-DE versus synop T2m: COSMO-DE minus synop Green: hits;black: misses red: false alarms, blue: no obs Courtesy of K. Stephan

5. Problematic weather situation for photovoltaic power production: low stratus clouds Low stratus clouds not predicted Low stratus clouds observed in reality Power from PV modules Hochrechnung Projection Day-Ahead Intra-Day time Error Day-Ahead: 4800 MW courtesy by TENNET

6. Problematic weather situations for photovoltaic power prediction • Cloud cover after cold front pass • Convectivesituations • Low stratus / fogweathersituations • Snow coverageofphotovoltaicmodules

7. Motivation • Photovoltaic power productionforecasts: Germany planstoincreasethepercentageofrenewableenergyto 35% in 2020  Increasingdemandsforaccurate power predictionsfor a safeandcost-effective power system • Project EWeLiNE: Objective: improveweatherand power forecastsforwind andphotovoltaic power • Main motivation: improvecloudcoversimulationoflowstratusclouds in stable wintertime high-pressuresystems • Should also proveusefulfor frontal systemorconvectivesituations

8. The COSMO model • COSMO-DE : • Limited-area short-rangenumerical modelweatherpredictionmodel • x  2.8 km / 50 vertical layers • Explicit deep convection • New data assimilation system : Implementation of the Ensemble Kalman Filter: LETKF after Hunt et al. (2007) • See also posters on • Observation impact in a convective-scale LETKF by Martin Weissmann • Usage of convective-scale LETKF to provide initial conditions for ensemble forecasts by Florian Harnisch

9. Observation systems Geostationary satellite data: Meteosat-SEVIRI (Δx ~ 5km over central Europe, Δt=15 min) Source: EUMETSAT NWCSAF satellite product: cloud top height Cloud top height Cloud top height Relative humidity at cloud top height Cloud cover 1 2 3 4 5 6 7 8 9 10 11 12 13 Height [km]

10. relative humidity height of model level k = 1 Z [km] Determine the model equivalent cloud top model profile k1 k2 Cloud top CTHobs k3 k4 k5 • (make sure to choose the top of the detected cloud) • use y=CTHobsH(x)=hk • and y=RHobs=1H(x)=RHk(relative humidity over water/ice depending on temperature) • as 2 separate variables assimilated by LETKF RH [%] • Avoid strong penalizing of members which are dry at CTHobs but have a cloud or even only high humidity close to CTHobs  search in a vertical range hmaxaround CTHobs for a ‘best fitting’ model level k, i.e. with minimum ‘distance’ d:

11. Example: 17 Nov 2011, 6:00 UTCObservations and model equivalents „Cloud top height“ Observation Model RH model level k

12. Determinemodelequivalent: cloudfreepixels Z [km] What information can we assimilate for pixels which are observed to be cloudfree? 12 „no high cloud“ • assimilate cloud fraction CLC = 0 separately • for high, medium, low clouds • model equivalent: • maximum CLC within vertical range 9 „no mid-level cloud“ 6 3 „no low cloud“ CLC

13. Example: 17 Nov 2011, 6:00 UTC • COSMO cloud cover where observations “cloudfree” Low clouds (oktas) Mid-level clouds (oktas) High clouds (oktas)

14. “Single observation“ experiment • Analysis for 17 November 2011, 6:00 UTC (no cycling) • Each column is affected by only one satellite observation • Objective: • Understand in detail what the filter does with such special observation types • Does it work at all? • Detailed evaluation of effect on atmospheric profiles • Sensitivity to settings

15. Single-observation experiments: missed cloud event • 1 analysis step, 17 Nov. 2011, 6 UTC (wintertime low stratus) vertical profiles relative humidity cloud cover cloud water cloud ice observed cloud top 3 lines in one colour indicate ensemble mean and mean +/- spread

16. Missed cloud case: Effect on temperature profile temperature profile [K] (mean +/- spread) first guess analysis observed cloud top • LETKF introduces inversion due to RH  T cross correlations • in first guess ensemble perturbations

17. Comparisoncyclingexperiment: onlyconventional vs conventional + clouddata • 1-hourly cycling over 20 hours with 40 members • 13 Nov., 21UTC – 14 Nov. 2011, 18UTC • Wintertime low stratus • Thinning: 14 km • Results from additional “deterministic“ simulation based on LETKF Kalman gain matrix:

18. Comparison “only conventional“ versus “conventional + cloud obs" Time series of first guess errors, averaged over cloudy obs locations RH (relative humidity) at observed cloud top assimilation of conventional obs only assimilation of conventional + cloud obs RMSE Bias (OBS-FG) • Cloud assimilation reduces RH (1-hour forecast) errors

19. Comparison of cycled experiments Total cloud cover of first guess fields after 20 hours of cycling conventional + cloud conventional only satellite obs Satellite cloud top height 12 Nov 2011 17:00 UTC

20. Cycled assimilation of dense observations Time series of first guess errors, averaged over cloud-free obs locations (errors are due to false alarm clouds) mean square error of cloud fraction [octa] Solid: conv only Dashed: conv + clouds low clouds High clouds Mid-level clouds • False alarm clouds reduced through cloud data assimilation

21. Comparison “only conventional“ versus “conventional + cloud obs" ‘false alarm’ cloud cover (after 20 hrs cycling) high clouds mid-level clouds low clouds conventional obs only [octa] conventional + cloud

22. Comparisonforecastexperiment: onlyconventional vs conventional + clouddata • 24h deterministic forecast based on analysis of two experiments (after 12 hours of cycling) • 14 Nov., 9UTC – 15 Nov. 2011, 9UTC • Wintertime low stratus

23. Comparison of free forecast: time series of errors Cloudfree pixels Cloudy pixels RH (relative humidity) at observed cloud top averaged over all cloudy observations Mean squared error averaged over all cloud-free observations • The forecastofcloudcharacteristicscanbeimprovedthroughtheassimilationofthecloudinformation Solid: conv only Dashed: conv + clouds RMSE Bias (Obs-Model) Low clouds Mid-level clouds High clouds Conventional + cloud data Only conventional data

24. Verification: fit against independent measurements Conventional + cloud data Only conventional data Fit to SEVIRI infrared brightness temperatures (model values computed with RTTOV) RMSE Bias (Obs-Model)  RMSE issmallerforfirst 16 hoursofforecastforcloudexperiment, biasvaries

25. Conclusion / Outlook • Use of (SEVIRI-based) cloud observations in LETKF: • Increaseshumidity / cloud where it should and reduces ‘false-alarm’ clouds • Long-lasting free forecast impact for a stable wintertime high pressure system • Current work: Evaluate impact on other variables (temperature, wind) and other weather situations • Also work on cloudy infrared SEVIRI radiance assimilation (see poster by Africa Perianez) • Application in renewable energy project EWeLiNE to improve photovoltaic power predictions • Also planned to assimilate the PV power itself... Thankyouforyourattention!

26. Single-observation experiments: missed cloud event Cross section of analysis increments for ensemble mean relative humidity [%] specific water content [g/kg] observed cloud top observation location • Moistening of the layer where cloud is observed.