WRF-Solar® is the first numerical weather prediction model specifically designed to meet the growing demand for specialized numerical forecast products for solar energy applications (Jimenez et al. 2016). WRF-Solar is a specific configuration and augmentation of the Weather Research and Forecasting (WRF) model. The version 1 of the model was developed within the Sun4Cast® project funded by the U.S. Department of Energy that targeted to improve solar power forecasts at a wide range of temporal scales (Haupt et al. 2016). The original code is based on WRF Version 3.6 and it is available to download from this website. 

Sketch representing the physical processes that WRF-Solar® improves. The different components of the radiation are indicated.

The Community Version of WRF-Solar is in the public domain and can be downloaded from the official WRF Github repository. The WRF version 4.2 includes the enhancements of WRF-Solar Version 1 with upgrades in the physical parameterizations as well as other developments. Users are encouraged to use version 4.2 or upcoming versions.

This website provides a description of the model, the user’s guide, a reference configuration that should be used as a baseline for comparison by the WRF-Solar community, and ongoing developments.

References

Haupt, S. E., Kosovic, B., Jensen, T. L., Lee, J., Jimenez Munoz, P., Lazo, J. K., … Hinkleman, L. (2016). The Sun4Cast® Solar Power Forecasting System: The Result of the Public-Private-Academic Partnership to Advance Solar Power Forecasting (No. NCAR/TN-526+STR). doi:10.5065/D6N58JR2.

Jimenez, P.A., J. P. Hacker, J. Dudhia, S. E. Haupt, J. A. Ruiz-Arias, C. A. Gueymard, G. Thompson, T. Eidhammer and A. Deng, 2016a: WRF-Solar: Description and Clear-Sky Assessment of an Augmented NWP Model for Solar Power Prediction. Bull. Amer. Met. Soc.. 97, 1249-1264. doi:10.1175/BAMS-D-14-00279.1.

Description

Version 1

The development of WRF-Solar® Version 1 provided the first numerical weather prediction model specifically designed to meet the needs of irradiance forecasting (Jimenez et al. 2016a). The first augmentation improved the solar tracking algorithm to account for deviations associated with the eccentricity of the Earth’s orbit and the obliquity of the Earth. Second, WRF-Solar added the direct normal irradiance (DNI) and diffuse (DIF) components from the radiation parameterization to the model output. Third, efficient parameterizations were implemented to either interpolate the irradiance in between calls to the expensive radiative transfer parameterization, or to use a fast radiative transfer code that avoids computing three-dimensional heating rates but provides the surface irradiance (Xie et al. 2016). Fourth, a new parameterization was developed to improve the representation of absorption and scattering of radiation by aerosols (aerosol direct effect, Ruiz-Arias et al. 2015). A fifth advance is that the aerosols now interact with the cloud microphysics (Thompson and Eidhammer 2014), altering the cloud evolution and radiative properties (aerosol indirect effects), an effect that has been traditionally only implemented in atmospheric computationally costly chemistry models. A sixth development accounts for the feedbacks that sub-grid scale clouds produce in shortwave irradiance as implemented in a shallow cumulus parameterization (Deng et al. 2014).

Several works highlighted the benefits of the solar augmentations for solar irradiance forecasting. WRF-Solar largely reduced errors in the simulation of clear sky irradiances wherein is important to properly account for the impacts of atmospheric aerosols (Jimenez et al., 2016a). WRF-Solar have also been shown to reduce biases in the surface irradiance over the contiguous U.S. in all sky conditions (e.g. Jimenez et al. 2016b). In a formal comparison to the NAM baseline, WRF-Solar showed improvements in the Day-Ahead forecast of 22-42% (Haupt et al. 2016). Another work has pointed out the potential of WRF-Solar for nowcasting applications (Lee et al. 2016).  The study compared solar irradiance predictions using different nowcasting methodologies based on artificial intelligence or the utilization of satellite imagery to detect clouds. The comparison has shown that WRF-Solar was competitive, and in many times superior to these state-of-the-science methodologies of the short-term prediction (1-6 h).

The original code is based on WRF Version 3.6 and it is available to download from this website.

Community Version of WRF-Solar

The augmentations introduced in WRF-Solar Version 1 have been progressively incorporated in the official WRF release. All are available to the community since the WRF release version 4.2. The parameterizations introduced in Version 1 have been revisited, and enhancements and bug fixes have been introduced. In addition, new functionality has been incorporated. The model can output the clear sky irradiances and includes a solar diagnostic package. This new package adds to the standard output a number of two-dimensional diagnostic variables (e.g., cloud fraction, vertically integrated hydrometeor content, clearness index, etc). The solar diagnostic package can output these variables and the surface irradiances every model time step at selected locations.

On-going efforts continue developing the Community Version of WRF-Solar to further increase its value for solar energy applications.

References

Deng, A., B. Gaudet, J. Dudhia and K. Alapaty, 2014: Implementation and evaluation of a new shallow convection scheme in WRF. 94^th American Meteorological Society annual Metting, 26^th Conference on Weather Analysis and Forecasting/ 22^nd conference on Numerical Weather Prediction, 2-6 February, Atlanta, GA, 13 pp.

Haupt, S. E., Kosovic, B., Jensen, T. L., Lee, J., Jimenez Munoz, P., Lazo, J. K., … Hinkleman, L. (2016). The Sun4Cast® Solar Power Forecasting System: The Result of the Public-Private-Academic Partnership to Advance Solar Power Forecasting (No. NCAR/TN-526+STR). doi:10.5065/D6N58JR2.

Jimenez, P.A., J. P. Hacker, J. Dudhia, S. E. Haupt, J. A. Ruiz-Arias, C. A. Gueymard, G. Thompson, T. Eidhammer and A. Deng, 2016a: WRF-Solar: Description and Clear-Sky Assessment of an Augmented NWP Model for Solar Power Prediction. Bull. Amer. Met. Soc.. 97, 1249-1264. doi:10.1175/BAMS-D-14-00279.1.

Jimenez, P.A., S. Alessandrini, S. E. Haupt, A. Deng, B. Kosovic, J. A. Lee, L. Delle Monache, 2016b: The role of unresolved clouds on short-range global horizontal irradiance predictability. Mon. Wea. Rev., 144, 3099 - 3107. doi:10.1175/MWR-D-16-0104.1.

Lee, J. A., S. E. Haupt, P. A. Jimenez, M. A. Rogers, S. D. Miller and T. C. McCandless, 2016: Solar energy nowcasting case studies near Sacramento. J. Appl. Meterol. & Climatol. Doi: 10.1175/JAMC-D-16-0183.1, in press.

Ruiz-Arias, J.A., J. Dudhia and C. A. Gueymard, 2014: A simple parameterization of the short-wave aerosol optical properties for surface direct and diffuse irradiances assessment in a numerical weather model. Geosci., Model Devel., 7, 1159-1174.

Thompson, G., and T. Eidhammer, 2014: A study of aerosol impact on clouds and precipitation development in a large winter cyclone. J. Atmos. Sci., 71, 3636-3658.

Xie, Yu, M. Sengupta, J. Dudhia, 2016: A Fast-sky Radiation model for Solar applications (FARMS): Algorithm and performance evaluation. Solar Energy, 135, 435-445.

User’s guide

A description of the WRF-Solar® options is herein provided. The user is referred to the WRF User’s Guide and the README.namelist file in the run directory for further details on these options.

Certain WRF-Solar® enhancements are activated by default (e.g. improved solar tracking algorithm). The rest of the available options are: 

• WRF-Solar® calculates the direct and diffuse surface irradiance. The direct and diffuse irradiance are calculated directly by RRTMG (sw_physics = 4) and Goddard (sw_physics = 5) shortwave radiation parameterizations. For the rest of the shortwave parameterizations WRF-Solar parameterizes the contributions of the direct and diffuse components. The direct normal irradiance (DNI) is stored in the SWDDNI variable and the diffuse in the SWDDIF variable. RRTMG also outputs the clear sky GHI and DNI on variables SWDOWNC and SWDDNIC, respectively. By default, these variables are not in the standard output. To add them to the standard output the user needs to add a h in the IO column in the rows of the registry file containing these variables.

• Different options are available to account for the atmospheric aerosol direct and/or indirect effects.
  • By setting aer_opt = 1 WRF-Solar uses a monthly climatology to represent the aerosol direct effect. This option only works with RRTMG shortwave parameterization (sw_physics = 1).
  • An alternative parameterization to represent the aerosol direct effect is activated by setting aer_opt = 2. This option works for RRTMG (sw_physics = 4) and for Goddard (sw_physics = 5). In its simplest form, the user has to provide the aerosol optical depth (AOD) at 550 nm and select the predominant type of aerosol. Normally, using a rural type of aerosol is appropriate (aer_type = 1). Two options are available for setting the AOD at 550 nm: 1) constant value, and 2) variable read via auxiliary file(s) in WRF I/O API-conforming netCDF format. In addition, the user needs to indicate in the namelist how to calculate the Agnstrom exponent, single scattering albedo, and asymmetry parameter. Three options are available 1) constant values, 2) fields read from the auxiliary input, and 3) parameterize them based on the AOD at 550 nm and the predominant aerosol type. For more details on how to select these options the user is referred to the README.namelist file in the WRF run directory (the aer_opt and related variables).
  • To complement the previous options that impose the aerosol optical properties, WRF-Solar allows one to explicitly model the aerosol direct and indirect effects by setting the microphysics option, mp_physics, to 28. Documentation on how to run this parameterization should be consulted before running this option. The standard way to run the parameterization is using the climatological concentrations of the water- and ice-friendly aerosols as described in the previous link. In addition, the water- and ice-friendly aerosols can be initialized with concentrations with time stamps. The concentrations of the water- and ice-friendly aerosols (QNWFA and QNIFA variables) are read in WPS intermediate format by Metgrid. Documentation to write in intermediate format is also available. These aerosol data is read by WRF-Solar setting use_rap_aero_icbc = .true. By default, this option, mp_physics = 28, only accounts for the aerosol indirect effects. Setting mp_physics = 28 and aer_opt = 3 also accounts for the aerosol direct effect in order to fully couple the cloud-aerosol-radiation system.

• To activate the effects of unresolved clouds on shortwave radiation, set shcu_physics = 5. The Cu parameterization should be turned off since the WRF-Solar shallow cumulus scheme also accounts for deep convection. This option only works with two planetary boundary layer parameterizations, bl_pbl_physics = 2 or 5. The option 5 is recommended.

• A couple of options are available to have surface irradiances updated in between calls to the shortwave parameterization.

• Setting swint_opt = 1 interpolates the irradiance in between calls to the shortwave radiation parameterization. The parameterization keeps constant the concentration of hydrometeors and estimates the clear sky irradiance.
• By setting swint_opt = 2 the Fast All-sky Radiation Model for Solar applications (FARMS) scheme is activated. FARMS calculates the surface irradiance every model time step and stores the values in SWDOWN2, SWDDNI2 and SWDDIF2 variables. It also provides clear sky irradiances (SWDOWNC2 and SWDDNIC2). This option uses the current atmospheric state, including hydrometeors, to calculate the surface irradiance and it is the recommended one.

WRF-Solar includes a solar diagnostic package (solar_diagnostic = 1). This option adds to the standard output a number of two-dimensional variables (e.g., cloud fraction, vertically integrated hydrometeor content, clearness index, etc). A complete list of the variables can be found in the README.tslist file in the run directory. If the tslist option is activated, the solar diagnostic package outputs these variables and the surface irradiances every model time step at selected locations. This requires an ascii file with the latitude and longitude of the sites to output the time series (see README.tslist file in the run directory).

Reference configuration

The following configuration should be used as a reference by the WRF-Solar® community to help us better understand the model performance which would translate into faster development and more efficient use of solar technologies.

download namelist reference config

Users should use the Community Version of WRF version 4.2, or higher, to be able to use all these options:

&time_control
run_days = $RUN_DAYS,
run_hours = $RUN_HOURS,
run_minutes = 0,
run_seconds = 0,
start_year = $START_YEAR,
start_month = $START_MONTH,
start_day = $START_DAY,
start_hour = $START_HOUR,
start_minute = 00,
start_second = 00,
end_year = $END_YEAR,
end_month = $END_MONTH,
end_day = $END_DAY,
end_hour = $END_HOUR,
end_minute = 00,
end_second = 00,
interval_seconds = $INTERVAL_SECONDS,
input_from_file = .true.,
history_interval = 15,
frames_per_outfile = 1000,
restart = .false.,
restart_interval = 9999,
io_form_history = 2
io_form_restart = 2
io_form_input = 2
io_form_boundary = 2
io_form_auxinput1 = 2
io_form_auxhist2 = 2
debug_level = 0
/

&domains
time_step = 30,
time_step_fract_num = 0,
time_step_fract_den = 1,
max_dom = 1,
s_we = 1,
e_we = $E_WE,
s_sn = 1,
e_sn = $E_SN,
s_vert = 1,
e_vert = 45,
eta_levels = 1.00000, 0.99629, 0.99257, 0.98879,
0.98486, 0.98071, 0.97622, 0.97130,
0.96585, 0.95977, 0.95299, 0.94540,
0.93692, 0.92744, 0.91686, 0.90507,
0.89195, 0.87737, 0.86120, 0.84331,
0.82356, 0.80181, 0.77793, 0.75181,
0.72335, 0.69246, 0.65911, 0.62329,
0.58506, 0.54455, 0.50195, 0.45755,
0.41175, 0.36503, 0.31802, 0.27144,
0.22617, 0.18317, 0.14344, 0.10788,
0.07710, 0.05132, 0.03028, 0.01343,
0.00000,
p_top_requested = 5000,
num_metgrid_levels = $NUM_METGRID_LEVELS,
num_metgrid_soil_levels = $NUM_METGRID_SOIL_LEVELS,
dx = 9000,
dy = 9000,
grid_id = 1,
parent_id = 1,
i_parent_start = 1,
j_parent_start = 1,
parent_grid_ratio = 1,
parent_time_step_ratio = 1,
feedback = 0,
/

&physics
mp_physics = 8,
ra_lw_physics = 4,
ra_sw_physics = 4,
radt = 5,
sf_sfclay_physics = 1,
sf_surface_physics = 2,
bl_pbl_physics = 5,
bldt = 0,
bl_mynn_tkeadvect = .true.,
cu_physics = 0,
cu_rad_feedback = .false.,
cudt = 0,
bl_mynn_edmf = 0,
shcu_physics = 5,
isfflx = 1,
ifsnow = 1,
icloud = 1,
icloud_bl = 0,
num_soil_layers = 4,
sf_urban_physics = 0,
num_land_cat = 21,
aer_opt = 1,
swint_opt = 2,
usemonalb = .true.,
/

&dynamics
w_damping = 1,
diff_opt = 1,
km_opt = 4,
damp_opt = 3,
zdamp = 5000.,
dampcoef = 0.2,
non_hydrostatic = .true.,
/

&bdy_control
spec_bdy_width = 5,
spec_zone = 1,
relax_zone = 4,
specified = .true.,
nested = .false.,
/

&grib2
/

&namelist_quilt
nio_tasks_per_group = 0,
nio_groups = 1,
/

&diags
solar_diagnostic = 1,
/

Ongoing developments

• Enhancing WRF-Solar® to provide probabilistic forecasts. The National Renewable Energies Laboratory (NREL) is leading a project and collaborates with NCAR to incorporate a probabilistic framework specifically tailored for solar energy applications.
• Enhancing WRF-Solar physics for version 2. The Pacific Northwest National Laboratory (PNNL) is leading a project collaborating with NCAR to enhance the WRF-Solar physics and quantify uncertainties to model parameters.
• MAD-WRF project. NCAR is leading a project to couple WRF-Solar with a modified version of MADCast to create MAD-WRF in order to improve the cloud initialization for nowcasting applications.
• PV modelling: Arizona State University is leading a project collaborating with NCAR to incorporate an online parameterization of PV panels production.
• Enhancing microphysics and DNI modelling: Brookhaven National Laboratory (BNL) is leading a project to enhance the WRF-Solar microphysics as well as to improve the representation of the cloud interactions with the DNI.