Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies - PowerPoint PPT Presentation

raja
slide1 n.
Skip this Video
Loading SlideShow in 5 Seconds..
Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies PowerPoint Presentation
Download Presentation
Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies

play fullscreen
1 / 25
Download Presentation
Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies
123 Views
Download Presentation

Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies J. A. Ruiz-Arias MMM/NCAR, University of Jaén Spain J. Dudhia MMM/NCAR C.A. Gueymard Solar ConsultingServices Colebrook, NH D. Pozo-Vázquez University of Jaén Spain 13th Annual WRF User’sWorkshop, 25-29 June 2012, Boulder, CO

  2. MOTIVATIONS – TheScope • Solar energyapplications are demandinganincreasedreliabilityin thecurrentmethodsforGHI and DNI... • ... assessment(bankability) from, at least, 10 yearsbut, ideally, up to 20 or 30 years. • lendersneedtoassessrisksdueto “bad” years and long-termvariability • ...and short-termforecasting(minutes tofewdaysahead) • forimproved solar powerplantsoperation, grid stability and higherpenetration. Veryimportantfor CSP plants (thermalstoragemanagement) GHI: Global Horizontal Irradiance (Fuel for PV – H&C Building) DNI: Direct Normal Irradiance (Fuel for CSP – CPV) SURFACE SOLAR RADIATION Jose A. Ruiz-Arias

  3. MOTIVATIONS – WhatIsThisAbout? Surfaceshortwavesolar radiationcomponents(DNI1 and DIF2) are notamongthe regular outputs of WRF. Butsome SW schemescalculatetheminternally Forinstance, theGoddardSpace Flight Center (GSFC) shortwaveradiationscheme Biascausedbylack of aerosols Majorimpactison DNI and DIF irradiances Here, wepresent a new aerosol parameterization and earlytestsusingthe GSFC SW scheme 1: Direct Normal Irradiance 2: DiffuseIrradiance Jose A. Ruiz-Arias

  4. MOTIVATIONS – Current Gaps • Currently, satellite-basedmethodsdominate in the solar industry. Buttheysuffer of twomainshortcomings: • satellite records for solar assessment are heterogeneous and limitedboth in time and space. Series longerthan 10-15 years are desirable!! • Theperformance of thesatellite-basedforecastsdrasticallydisminishesbeyond 4-6 hours!! PersistenceObs Satellite NWP Satellite – NO forecast Perez et al. (2009) NDFD: National Digital ForecastDatabase Jose A. Ruiz-Arias NDFD: http://www.nws.noaa.gov/ndfd/technical.htm

  5. OBJECTIVES – Solar IndustryRequirements ToprovideGHI and DNI Toassesssurface solar resourcefrom, at least 10 years. Ideally, 20 oreven 30 years Toforecasts2-3 daysaheadat sub-hourlyscale Jose A. Ruiz-Arias

  6. OBJECTIVES – Solar IndustryRequirements ToprovideGHI and DNI Toassesssurface solar resourcefrom, at least 10 years. Ideally, 20 oreven 30 years Toforecasts2-3 daysaheadat sub-hourlyscale What do weneedtoimprove in WRF? Jose A. Ruiz-Arias

  7. OBJECTIVES – Solar IndustryRequirements • ToprovideGHI and DNI • Toassesssurface solar resourcefrom, at least 10 years. Ideally, 20 oreven 30 years • Toforecasts2-3 daysaheadat sub-hourlyscale • What do weneedtoimprove in WRF? • Toinclude DNI in the output dataset Jose A. Ruiz-Arias

  8. 1.) DNI CALCULATION GSFC SW schemealreadycalculatesinternallythedirect and diffusecomponents at eachspectral band However, aerosol opticalproperties are turneddown!! ATMOSPHERE GSFC SW SCHEME (AOD=0; SSA=0; ASY=0) SW RADIATION Jose A. Ruiz-Arias

  9. OBJECTIVES – Solar IndustryRequirements • ToprovideGHI and DNI • Toassesssurface solar resourcefrom, at least 10 years. Ideally, 20 oreven 30 years • Toforecasts2-3 daysaheadat sub-hourlyscale • What do weneedtoimprove in WRF? • Toinclude DNI in the output dataset Jose A. Ruiz-Arias

  10. OBJECTIVES – Solar IndustryRequirements • ToprovideGHI and DNI • Toassesssurface solar resourcefrom, at least 10 years. Ideally, 20 oreven 30 years • Toforecasts2-3 daysaheadat sub-hourlyscale • What do weneedtoimprove in WRF? • Toinclude DNI in the output dataset • Toincludethedirecteffect of aerosols in the SW radiative transfer • An aerosol parameterizationisrequired Jose A. Ruiz-Arias

  11. 2.) AEROSOL PARAMETERIZATION Theinclusion of aerosols in themodelshouldbe... ...as simple as possibletomakeiteasytothepeople of thesolar industry, ...versatiletoallowrapidupdating of aerosolsfrommultiple(heterogeneous) sources ...and sufficientlyprecise and accuratefor, specially, DNI assessment and short-termforecast. In turn, currentatmosphericchemistrymodels, stillunderstrongdevelopment,... ...are stillcomputationallyexpensive–notdesirableforoperationalforecasting. ...require a complexand somehowrigidinitializationof aerosols – notsuitableforvery short-termapplications ...a thoroughdescriptionof aerosols as in thesemodelsmightnotberequiredforsurfacesolar radiationassessment Jose A. Ruiz-Arias

  12. 2.) AEROSOL PARAMETERIZATION • Toaccountforaerosols, weneedtoparameterize... • ... aerosol opticaldepth (AOD) • ... aerosol single-scattering albedo (SSA), and • ... aerosol asymmetry factor (ASY) • at eachspectral band (11) and every grid-cell of thedomain, includingeachmodel vertical layer (weassumeanexponentialprofile) Theproposedparameterizationonlyrequires... ... thetotal aerosol opticaldepth at 550 nm, ... thepredominanttype of aerosol, and ... therelativehumidity. Jose A. Ruiz-Arias

  13. 2.) AEROSOL PARAMETERIZATION (Shettle and Fenn, 1979) Jose A. Ruiz-Arias

  14. Jose A. Ruiz-Arias

  15. CASE STUDY – ExperimentDesign • CONUS, June – August 2009, ERA-Interim, 27 km, data every 10 minutes • Dailygridded AOD at 550 nmfrom Level-3 MODIS dataset (1ºx1º) • 3 runs: no-aerosols, rural aerosol, urbanaerosol • ThreeradiometricstationsfromNOAA’s SURFRAD network: • Bondville (IL), Boulder (CO) and Desert Rock (NV) • 1-minute measurementsof GHI, DNI and DIF • Concurrentmeasurements of spectral AOD • Validationundercloudlessconditionsonlyboth in WRF and observations (cloud-screeningalgorithm) Jose A. Ruiz-Arias

  16. CASE STUDY – Gridded AOD Source 1.) Initial L3 MODIS AOD 2.) InterpolatedwithKriging 3.) CressmanAnalysis Thisprocedurewasrepeatedeveryday of thestudyperiod Jose A. Ruiz-Arias

  17. RESULTS – Bondville (IL) Non-screenedObs: 2.6% Mean AOD: 0.17 Mean Precip. Water: 2.74 cm 34 6.2% 133 17.2% -39 -44.2% 20 3.7% 51 6.6% -14 -15.8% 4 0.8% 38 4.9% -24 -26.7% -6 -1.1% -14 -1.8% -8 -8.8% Jose A. Ruiz-Arias

  18. RESULTS – Boulder (CO) Non-screenedObs: 9.2% Mean AOD: 0.13 Mean Precip. Water: 1.13 cm 47 7.3% 160 19.0% -39 -45.9% 31 4.9% 71 8.4% -4 -4.6% 4 0.6% 49 5.8% -19 -23.0% -8 -1.1% -7 -0.8% 4 4.9% Jose A. Ruiz-Arias

  19. RESULTS – Desert Rock (NV) Non-screenedObs: 21.3% Mean AOD: 0.09 Mean Precip. Water: 1.54 cm 24 3.4% 53 6.1% -4 -5.1% 33 4.6% 108 12.3% -30 -35.4% 4 0.5% 40 4.5% -17 -19.4% -7 -1.0% -5 -0.6% -2 2.4% Jose A. Ruiz-Arias

  20. RESULTS – Boulder (DIF) Jose A. Ruiz-Arias

  21. RESULTS – Boulder (DNI) Jose A. Ruiz-Arias

  22. RESULTS – Boulder (GHI) Jose A. Ruiz-Arias

  23. NEXT STEPS Wewanttomakealsoavailabletheparameterizationtoother SW schemes, as the RRTMG. New implementation: The extended validationtoother SW schemeswillincreaseourknowledgeontheparameterizationand may lead tofurtherimprovements TheAngstrom exponentcouldbeusedtoinferthetype of aerosol and spectral AOD. But so farthereis no a reliable data source. Theparameterizationcouldingest AOD fromtheIFS/ECMWF Are more aerosol typesrequired? Gridded AOD Aerosol Type WRF Aerosol Module GSFC RRTMG Jose A. Ruiz-Arias

  24. CONCLUSIONS Thebiascaused in DNI and DIF bythelack of aerosolsismuchlargerthan in GHI WhentherightAOD isprovided,the aerosol parameterizationcorrectsthebias in GHI and most of it in DNI and DIF irradiances Thevertical profileof aerosolsseemstobea secondordereffect Finally, theparameterizationwouldbenefitfrom athoroughvalidationof thecurrentlyavailable AOD satellitedatasetsand development of biasreductionmethodsforthem. Jose A. Ruiz-Arias

  25. Improvement of the WRF Model for Solar Resource Assessment and Forecast Under Clear Skies J. A. Ruiz-Arias MMM/NCAR, University of Jaén Spain Thankyouverymuch! J. Dudhia MMM/NCAR C.A. Gueymard Solar ConsultingServices Colebrook, NH D. Pozo-Vázquez University of Jaén Spain 13th Annual WRF User’sWorkshop, 25-29 June 2012, Boulder, CO