improving solar radiation forecasting
DESCRIPTION
Improving Solar Radiation Forecasting. Betsy Weatherhead, U. Colorado Craig S. Long, NOAA/NWS/NCEP/CPC Kathy Lantz, U. Colorado Patrick Disterhoft, U. Colorado. AMS Summer Meeting, August 12, 2009. Needs of Renewable Energy Community. - PowerPoint PPT PresentationTRANSCRIPT
Improving Solar Radiation Improving Solar Radiation ForecastingForecasting
Betsy Weatherhead, U. ColoradoBetsy Weatherhead, U. Colorado
Craig S. Long, NOAA/NWS/NCEP/CPCCraig S. Long, NOAA/NWS/NCEP/CPC
Kathy Lantz, U. ColoradoKathy Lantz, U. Colorado
Patrick Disterhoft, U. ColoradoPatrick Disterhoft, U. Colorado
AMS Summer Meeting, August 12, 2009
Needs of Renewable Energy Needs of Renewable Energy CommunityCommunity
Solar radiation can reduce the Solar radiation can reduce the demand on the grid (personal demand on the grid (personal Photo-voltaic panels). Highly angle Photo-voltaic panels). Highly angle dependent. Total Solar Radiation.dependent. Total Solar Radiation.
PV plants depend on total solar PV plants depend on total solar radiation.radiation.• Fixed angleFixed angle• Single axisSingle axis• Full rotationFull rotation
Concentrated Solar Energy highly Concentrated Solar Energy highly dependent on direct beam radiationdependent on direct beam radiation• Various forms of energy storageVarious forms of energy storage
Time scales from seconds to Time scales from seconds to decades with a strong focus on decades with a strong focus on minutes to days.minutes to days.
Solar Radiation - BasicsSolar Radiation - Basics Solar radiation reaching the surface of the Earth Solar radiation reaching the surface of the Earth
is affected by many factors:is affected by many factors:• Sun angleSun angle• CloudsClouds• AerosolsAerosols• Surface reflectivitySurface reflectivity• Water vaporWater vapor• OzoneOzone
All of these factors can change dynamically on a All of these factors can change dynamically on a day-to-day basis.day-to-day basis.
Each part of the solar spectrum is affected Each part of the solar spectrum is affected differently by these factors.differently by these factors.
Technologies are Technologies are still developingstill developing
Wavelength Wavelength matters.matters.
Clouds, aerosols, Clouds, aerosols, ozone, water ozone, water vapor all matter.vapor all matter.
Multiple paths to making solar Multiple paths to making solar forecastsforecasts
ProprietaryProprietary Data extrapolationData extrapolation Statistical modelingStatistical modeling Physical modelingPhysical modeling Ensemble forecastsEnsemble forecasts Data assimilation, rapid refreshData assimilation, rapid refresh
Rapid Updatefor Aviation, Severe Weather
Air Quality
WRF NMMWRF ARW
WRF: NMM+ARW ETA, RSM GFS, Canadian Global Model
MostlySatellite+Radar
North AmericanMesoscaleWRF NMM
North American Ensemble Forecast System
Hurricane GFDLHWRF
GlobalForecastSystem
Dispersion
ARL’s HYSPLIT
Forecast
Severe Weather
ClimateCFS
~2B Obs/Day
Short-RangeEnsemble Forecast
NOAA’s NWS Model Production SuiteNOAA’s NWS Model Production Suite
MOM3
NOAH Land Surface Model
Coupled
Global DataAssimilation
OceansHYCOM
WaveWatch III
NAM+CMAQ
Courtesy Stan Benjamin
Forecasting Solar UV RadiationForecasting Solar UV Radiation
UV Forecasts have UV Forecasts have taken place for fifteen taken place for fifteen years in support of years in support of human health.human health.
NOAA, various weather NOAA, various weather services, Accu-Weatherservices, Accu-Weather
Private monitoring Private monitoring offers Nowcasts.offers Nowcasts.
NCEP 5-Day UV PredictionsFive Successive Days
UV Index Algorithm UV Index Algorithm Old (Old (virtually unchanged since 1994virtually unchanged since 1994))
• Weight UV relationship w/total ozone and SZA derived from Weight UV relationship w/total ozone and SZA derived from delta Eddingtondelta Eddington RT model. RT model.• Total ozone fcsts from GFS.Total ozone fcsts from GFS.• Surface UV albedo held constant at Surface UV albedo held constant at 5%,5%, (no allocation for snow or ice cover)(no allocation for snow or ice cover)..• Aerosols: AOD is Aerosols: AOD is constantconstant at 0.2 and SSA = 1.0. at 0.2 and SSA = 1.0.• Elevation increases UV flux by ~6% per km.Elevation increases UV flux by ~6% per km.• UVI clear sky field on 1x1 grid up to this point.UVI clear sky field on 1x1 grid up to this point.• Relationship between Relationship between NGM MOS cloud probNGM MOS cloud prob and cloud attenuation of UV flux. and cloud attenuation of UV flux.• Cloudy UV Index values at all (~800) MOS locations.Cloudy UV Index values at all (~800) MOS locations.• Bulletin, Eta grid 211 and 207 are created by interpolation from MOS cities.Bulletin, Eta grid 211 and 207 are created by interpolation from MOS cities.• Products are limited to next day solar noon over US.Products are limited to next day solar noon over US.
NewNew• Weight UV relationship with total ozone, and SZA derived from Weight UV relationship with total ozone, and SZA derived from NCAR-TUVNCAR-TUV RT model. RT model.• Total ozone fcsts from GFS.Total ozone fcsts from GFS.• Surface UV albedo held constant at Surface UV albedo held constant at 3%3% for non-snow conditions. for non-snow conditions.
GFS albedo is used when > 30% (where snow is possible).GFS albedo is used when > 30% (where snow is possible).• Aerosols: AOD and SSAAerosols: AOD and SSA vary vary, derived from , derived from seasonal global climatology grids.seasonal global climatology grids.• Elevation increases UV flux by ~ 6% per km.Elevation increases UV flux by ~ 6% per km.• Ratio of Ratio of GFS down-welling UVB flux w & w/o cloudsGFS down-welling UVB flux w & w/o clouds determines cloud attenuation. determines cloud attenuation.• Output remains on gaussian grid.Output remains on gaussian grid.• GFS inputs available at 3hr interval.GFS inputs available at 3hr interval.• UV Index products are outputted globally at 1 hr intervals out to 96 hoursUV Index products are outputted globally at 1 hr intervals out to 96 hours
UV Index : 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
- f19
Satellite IR Clouds obs at 11:45 UTC Oct 12, 2003
- f19
Bondville
25 mWatts/sq m = 1 UV Index unit
__ Diurnal trace of observed UVx 1 hour mean about 18 UTC| St Dev of obs +/- 0.5 hr about 18 UTCo 18 hr FCST from 00Z run valid at 18 UTC
40.04N, 88.36W
Ery
them
al D
ose
Rat
e (m
W/s
q m
)
Improvements in NWS’ solar UV Improvements in NWS’ solar UV forecastforecast
Improvements in cloud and ozone Improvements in cloud and ozone forecasts result in a much improved forecasts result in a much improved solar forecast.solar forecast.
Verification depends strongly on time Verification depends strongly on time period of averaging when single period of averaging when single locations are considered.locations are considered.
Significant improvements in solar Significant improvements in solar radiation forecastsradiation forecasts
Cloud modeling has improved Cloud modeling has improved significantly over the past twenty significantly over the past twenty years.years.
Inclusion of “cloud optical depth” and Inclusion of “cloud optical depth” and “cloud height” are of considerable “cloud height” are of considerable importance to assessing importance to assessing transmission.transmission.
Aerosol forecasts are currently under Aerosol forecasts are currently under consideration in NWS.consideration in NWS.
Erythemal Dose Rate - Bondville, IL - Oct 2003
0
25
50
75
100
125
274 277 280 283 286 289 292 295 298 301 304
Day of Year
Ery
thm
al D
ose
Rat
e (m
W/s
q m
)
SURFRAD New UVI-1 Current UVI25 mWatts/sq m = 1 UV Index unit
1 Day Forecast Verification
Erythemal Dose Rate - Bondville, IL - October 13 & 14, 200340.04N, 88.36W
0
25
50
75
100
125
286.0 286.5 287.0 287.5 288.0
Day of Year
Ery
them
al D
ose
Rat
e (m
Wat
ts/s
q m
)
OBS UVI Clear UVI w/clouds
25 mWatts/sq m = 1 UV Index unit278 DU 309 DU
Fine temporal structure is extremely difficult.
Renewable Energy NeedsRenewable Energy Needs
Improved forecasts are necessary for optimal use of solar energy Improved forecasts are necessary for optimal use of solar energy integration.integration.
Multiple factors must be considered to make accurate forecasts: Multiple factors must be considered to make accurate forecasts:
cloudsclouds, aerosols, water vapor., aerosols, water vapor. NWS currently makes solar radiation forecasts for 57 locations for UV NWS currently makes solar radiation forecasts for 57 locations for UV
index. Significant improvements have been achieved using state of the index. Significant improvements have been achieved using state of the art cloud forecasting.art cloud forecasting.
Fundamentally, improvements require accurate measurements of Fundamentally, improvements require accurate measurements of both direct and diffuse radiation.both direct and diffuse radiation.
Actual hourly variation of solar energy potential- accurate forecasts are essential
Long-term estimates of solar radiation will need to take into multi-year factors including el nino, solar cycles, NAO, changes in climate, including in changes in the jet stream.
Predictions of long-term changes will require research into past long-term datasets as well as best available climate models.
Long-term Predictions
Foundation for Model Foundation for Model Improvement--MeasurementsImprovement--Measurements
Surface Radiation monitoringSurface Radiation monitoring Direct beam separate from total solar radiation.Direct beam separate from total solar radiation. DOE’s Atmospheric Radiation Measurement, DOE’s Atmospheric Radiation Measurement,
National Renewable Energy LaboratoryNational Renewable Energy Laboratory NOAA’s SurfRad, Central Calibration FacilityNOAA’s SurfRad, Central Calibration Facility USDA’s UV Monitoring USDA’s UV Monitoring
NetworkNetwork Direct beam measurements Direct beam measurements
are difficult, but not are difficult, but not impossible.impossible.
NOAA’s SurfRad Network
Satellite IR Clouds obs at 11:45 UTC Oct 12, 2003
SummarySummary
Data are the foundation for Data are the foundation for improving forecasts.improving forecasts.
Current solar monitoring is Current solar monitoring is inadequate, but can be easily inadequate, but can be easily adapted and expanded.adapted and expanded.
Multiple modeling approaches will be Multiple modeling approaches will be needed to address the multiple needed to address the multiple needs of the renewable energy needs of the renewable energy communities.communities.