evaluation of wrf radiation and microphysics ... · evaluation of wrf radiation and microphysics...

33
Evaluation of WRF Radiation and Microphysics Parameterizations for use in the Polar Regions Mark W. Seefeldt Michael Tice Department of Engineering – Physics – Systems Providence College John J. Cassano and Matthew D. Shupe John J. Cassano and Matthew D. Shupe Cooperative Institute for Research in Environmental Sciences (CIRES) University of Colorado at Boulder Atmospheric Model Parameterizations in the Polar Regions Workshop Boulder, CO – July 12, 2012

Upload: others

Post on 13-Mar-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Evaluation of WRF Radiation and Microphysics

Parameterizations for use in the Polar Regions

Mark W. Seefeldt

Michael TiceDepartment of Engineering – Physics – Systems

Providence College

John J. Cassano and Matthew D. ShupeJohn J. Cassano and Matthew D. ShupeCooperative Institute for Research in Environmental Sciences (CIRES)

University of Colorado at Boulder

Atmospheric Model Parameterizations in the Polar Regions Workshop

Boulder, CO – July 12, 2012

Outline

• Objective / Motivation

• WRF General Configuration

• Radiation Parameterizations• Radiation Parameterizations

• Microphysics Parameterizations

• Observations for Evaluation

• Statistical Comparisons

• WRF 3.0 Physics Evaluation (2009)

Objective

• The preferred physics parameterizations in the

polar regions is still under debate and evaluation

• The purpose of this study is to evaluate a suite of • The purpose of this study is to evaluate a suite of radiation and microphysics parameterizations for

the use of the WRF model in the polar regions

• The anticipated outcome is the identification of a best-use combination of microphysics and radiation

parameterizations for use in the polar regionsparameterizations for use in the polar regions

• A secondary goal is the elimination of several of the

physics parameterizations as viable options

WRF in the Polar Regions• Atmospheric models have been found to have difficulty in

reproducing representative conditions in the polar regions

(Wyser et al. 2008)

• The problems are generally related to the atmospheric physics• The problems are generally related to the atmospheric physics

cloud processes

PBL and surface fluxes

radiation

processes

PBL and surface fluxes

sea ice / ice

covered land

WRF-ARW Version 3 Modeling System

User’s Guide

WRF Physics Parameterizations

• There are seven general WRF physics

parameterizations:

– radiation – shortwave (ra_lw_physics)– radiation – shortwave (ra_lw_physics)

– radiation – longwave (ra_sw_physics)

– microphysics (mp_physics)

– cumulus (cu_physics)

– boundary layer (bl_pbl_physics)– boundary layer (bl_pbl_physics)

– surface layer (sf_sfclay_physics)

– land surface model (sf_surface_physics)

WRF Physics Parameterizations• The performance of a given physics parameterization is often

related to the associated parameterizations

Image from

• The most complete way to identify the preferred physics

parameterizations is to evaluate different physics combinations

Image from WRF Tutorial

WRF Physics Evaluation – Summary

• This study will focus on the performance of the

radiation and microphysics parameterizations

• The study will compare WRF simulations and • The study will compare WRF simulations and observations for two locations:

Barrow, Alaska Summit, Greenland

• The month-long climate simulations will be evaluated against observations of meteorology (surface and

upper-level), clouds, and radiationupper-level), clouds, and radiation

• A statistical evaluation will be completed with the results providing a ranking of the preferred

parameterizations

WRF Physics Evaluation – General Configuration

• WRF-ARW v3.4 (released April, 2012)

• month-long climate simulations:

one-month + one-day, discard the first 24 hoursone-month + one-day, discard the first 24 hours

• simulations for four different months covering different seasonal

and radiation forcing conditions:

October 2011, January 2012, April 2012, July 2012

• Initial and lateral boundary conditions: ERA-Interim

• Fractional sea ice: NSIDC Near Real-Time DMSP SSMI

• Domains: outer – 50 km, inner – 10 km (one-way nesting)

• Vertical: 40 levels with a 50 mb top

• Timestep : 50 km – 300 s, 10 km – 60 s

• Post-processing: CF NetCDF files using wrfout_to_cf.ncl

http://foehn.colorado.edu/wrfout_to_cf/

WRF Physics Evaluation – Domains

• Two domains: Barrow – Alaska, Summit Camp – Greenland

WRF Physics Evaluation – Non-Varying Physics

General Physics:

•Land surface model (sf_surface_physics): Noah LSM (2)

•Surface Layer (sf_sfclay_physics): Eta (2) •Surface Layer (sf_sfclay_physics): Eta (2)

•Boundary layer (bl_pbl_physics): MYJ (2)

•Cumulus (cu_physics): Grell-Devenyi (3)

Other:Other:

•Fractional Sea Ice (fractional_seaice = 1)

•SST Updates (sst_update = 1 – uses wrflowinp_d0n)

WRF Physics – Radiation Parameterizations

• Limit the shortwave (ra_sw_physics) and longwave

(ra_lw_physics) radiation parameterizations to paired selections

• Four selected radiation combinations (LW / SW):

– RRTM / Goddard (lw_1-sw_2)– RRTM / Goddard (lw_1-sw_2)

– CAM / CAM (lw_3-sw_3)

– RRTMG / RRTMG (lw_4-sw_4)

– New Goddard / New Goddard (lw_5-sw_5)

• RRTM / Goddard was selected because of its use in past WRF

studies covering the polar regionsstudies covering the polar regions

WRF Physics – Radiation Parameterizations

To be evaluated:

• RRTM / Goddard (lw_1-sw_2)

• CAM / CAM (lw_3-sw_3)

• RRTMG / RRTMG (lw_4-sw_4)• RRTMG / RRTMG (lw_4-sw_4)

• New Goddard / New Goddard (lw_5-sw_5)

Eliminated:

• Dudhia (1) shortwave scheme was not selected based on poor

performance in a previous studyperformance in a previous study

• Fu-Liou Gu (7/7) was not selected as it is a new scheme with no

particular known strengths for the polar regions

• GFDL (99/99) was not chosen as it is old and being phased out

WRF Physics – Microphysics Parameterizations

• Eliminate the warm-rain and 3-phase parameterizations:

– Kessler (1)

– WRF Single-Moment 3-class (3)

• Eliminate the parameterizations with a focus on operational • Eliminate the parameterizations with a focus on operational

NCEP models:

– Eta (5) / (95)

• Eliminate the WRF double-moment parameterizations:

– WRF Double-Moment 5-class (14)

– WRF Double-Moment 6-class (16)

• Eliminate severe storm focused parameterizations:• Eliminate severe storm focused parameterizations:

– NSSL 2-moment (17, 18)

• Eliminate the 7-class paratmerization:

– Milbrandt-Yau Double-Moment 7-class (9)

WRF Physics – Microphysics Parameterizations

To be evaluated – seven microphysics parameterizations:

•Lin et al. (2)

•WRF Single-Moment 5-class (4)

•WRF Single-Moment 6-class (6)•WRF Single-Moment 6-class (6)

•Goddard (7)

•New Thompson (8)

•Morrison Double-Moment (10)

•Stony Brook University (Lin) (13)

WRF Physics Evaluation – Simulations Summary

Summary of simulations:

• Two domains (Barrow, AK and Summit Camp)

– evaluating the 50 km and the 10 km domains separately

• Four months (October 2011, January, April, July 2012)• Four months (October 2011, January, April, July 2012)

• Four radiation parameterization combinations

• Seven microphysics parameterizations

Total number of simulations:

2 x 4 x 4 x 7 = 224 simulations2 x 4 x 4 x 7 = 224 simulations

WRF Physics Evaluation – Observations

• Reviewed the available the data from the International Arctic

Systems for Observing the Atmosphere (IASOA) observatories

http://iasoa.org/

WRF Physics Evaluation – Observations

Barrow, Alaska – 71.323 N, 156.609 W, 11 m asl:

http://www.esrl.noaa.gov/gmd/obop/brw/

Summit, Greenland – 72.580 N, 38.48 W, 3238 m asl:

http://www.geosummit.org/

WRF Physics Evaluation – Observations

Evaluating WRF using observations from Barrow and Summit:

• surface meteorology (temperature, pressure, dew point /

mixing ratio, wind speed)

• upper-air meteorology (temperature, height, pressure, • upper-air meteorology (temperature, height, pressure,

moisture)

• surface radiation (downwelling longwave, downwelling

shortwave)

• radar and microwave (IWP, LWP)

WRF Physics – Statistical Comparisons

• Monthly time series plots comparing WRF simulations to

observations will be created

• Statistical measures of bias, RMSE, and correlation will be

calculated for each simulation and domaincalculated for each simulation and domain

WRF Physics – Statistical Comparisons

• The statistical performance for each radiation and microphysics

combination will be evaluated for each month and each domaincombination will be evaluated for each month and each domain

• The rankings of the statistical comparisons will be used to

determine preferred radiation and microphysics parameterizations

• Evaluation of the time series plots can be conducted to dig

further into the statistical comparisons

WRF 3.0 Physics Evaluation (ca. 2009)• Goal: identify preferred radiation and microphysics parameterizations

– radiation – 5 combinations (lw-sw): RRTM-Dudhia,

RRTM-Goddard, RRTM-CAM, CAM-Goddard, CAM-CAM

– microphysics – 6 schemes:– microphysics – 6 schemes:

Lin, WSM5, WSM6, Goddard, Thompson, Morrison

• Observations:

– Barrow – Baseline Surface Radiation Network

– SHEBA – Surface-Met Tower, Cloud Radiation

• Evaluate: temperature, pressure, shortwave down, longwave down, • Evaluate: temperature, pressure, shortwave down, longwave down,

liquid water path (SHEBA), ice water path (SHEBA)

• Evaluate: over different months: January, March, May, June

• Evaluate: 10 km domain versus 50 km domain

WRF 3.0 Physics Evaluation (ca. 2009)• The WRF results are compared against observations from the SHEBA met tower and Barrow – Baseline Surface Radiation Network

observations:-longwave downward-longwave downward

-shortwave downward-2 m temperature-surface pressure

SHEBA only:-LWP-IWP

Statistics are calculated to objectively evaluate

the performance of the model in comparison to the observations.

Correlation, Bias (WRF – obs), Root Mean Square Error (RMSE)

WRF 3.0 Physics Evaluation (ca. 2009)• The statistics of each sensor for each physics configuration, location, month, and model domain are aggregated together in a list

Note: The values for all 30 physics configurations and the different sensors have been

removed for display.

Each sensor statistic for each physics configuration is ranked by performance (1-30)

The average rank and standard deviation is calculated for each sensor

WRF 3.0 Physics Evaluation (ca. 2009)• The average of the three primary sensors (2 m temperature, shortwave downward, and longwave downward) is calculated for each domain and location

Note: The values for Barrow have been removed to simplify the display.

WRF 3.0 Physics Evaluation (ca. 2009)• The average rank for each physics configuration is calculated across the different domains, months, and locations.

• The physics parameterizations are sorted by the average rank, from best to worst

WRF 3.0 – 3 Sensor (T, SW, LW) Rankings• The CAM-CAM (3-3) radiation combination shows consistently the best performance (4 of top 8)

• The Goddard (7) microphysics is a strong showing with the top 3 performancestop 3 performances

• Overall, the CAM-Goddard (3-2) and RRTM-CAM (1-3) radiation schemes perform well

• The RRTM-Dudhia (1-1) and RRTM-Goddard radiation

combinations do not do wellcombinations do not do well

• The Lin (2) and Morrison (10) microphysics schemes do

very poorly, no matter the radiation combination

WRF 3.0 – 5 Sensor(T, SW, LW, IWP, LWP)• The CAM-CAM (3-3) radiation continues to perform well (5 of top 11)

• The Goddard (7) and Lin microphysics does well followed by the WSM5 (4) and WSM6 (6)by the WSM5 (4) and WSM6 (6)

• The RRTM-CAM (1-3) radiation performs fair, but all other than CAM-CAM have a mixture of results

• The Morrison (10) microphysics scheme continues to do

poorlypoorly

RRTM – Goddard (1-2), Morrison (10) – May 1998

CAM – CAM (3-3), Goddard (7) – May 1998

Shortwave and Longwave Radiation Rankings• The CAM SW (3) consistently does very well with the SW rankings (top 8)

• The Goddard SW (2) does moderate with SW rankings

• The Dudhia SW (1) performs very poorly • The Dudhia SW (1) performs very poorly

with the SW sensor

• The CAM LW (3) does well, but not

spectacular with LW rankings

• The RRTM LW (1) does well when matched with Dudhia SW (1) but not Goddard SW or

CAM SW (3)CAM SW (3)

• The CAM-CAM (3-3) radiation combination provides the best results

Liquid Water Path and Ice Water Path Rankings• The Goddard (7) and Thompson (8) show consistently good performance with LWP (6 of top 7)

• The Lin (2) and Morrison (10) do poorly with LWPwith LWP

• The Lin (2) and Morrison (10) do very well with IWP

• The Goddard does poorly with IWP

• Overall, using LWP and IWP is suspect as the model results showed limited success

(i.e. LWP in January)

WRF 3.0 Physics Evaluation (ca. 2009) - Summary

• The CAM-CAM (3-3) radiation combination consistently

performs with the best results in nearly every ranking and

categorization

• The RRTM-CAM (1-3) and CAM-Goddard (3-2) performs • The RRTM-CAM (1-3) and CAM-Goddard (3-2) performs

reasonable, but not as good as CAM-CAM

• The results for the microphysics is not as clear and will require

further analysis

• The Goddard (7) microphysics scheme could be classified as

strongest, but with some questionsstrongest, but with some questions

Mark Seefeldt [email protected]