Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d] Bias score at selected thresholds Pan-Arctic WRF Background Introduction.

Slides:



Advertisements
Similar presentations
ISU Atmospheric Component Update – Part I Justin Glisan Iowa State University.
Advertisements

Justin Glisan Iowa State University Department of Geological and Atmospheric Sciences RACM Project Update: ISU Atmospheric Modeling Component: Part 1 6th.
Justin Glisan Iowa State University Department of Geological and Atmospheric Sciences RACM Project Update: ISU Atmospheric Modeling Component: Part 1 7th.
Coupled versus uncoupled radiation and microphysicsMicrophysics droplet concentration 5. Conclusions The sea ice state simulated in RASM is sensitive to.
John J. Cassano, Matthew Higgins, Alice DuVivier University of Colorado Wieslaw Maslowski, William Gutowski, Dennis Lettenmaier, Andrew Roberts.
Chukchi/Beaufort Seas Surface Wind Climatology, Variability, and Extremes from Reanalysis Data: Xiangdong Zhang, Jeremy Krieger, Paula Moreira,
Dynamical Downscaling of CCSM Using WRF Yang Gao 1, Joshua S. Fu 1, Yun-Fat Lam 1, John Drake 1, Kate Evans 2 1 University of Tennessee, USA 2 Oak Ridge.
Recent performance statistics for AMPS real-time forecasts Kevin W. Manning – National Center for Atmospheric Research NCAR Earth System Laboratory Mesoscale.
LHS are the linear terms. First 2 terms on the RHS are nonlinear terms in the bias. The group THF are transient heat advection bias. Q^ is the bias in.
Outline Further Reading: Detailed Notes Posted on Class Web Sites Natural Environments: The Atmosphere GG 101 – Spring 2005 Boston University Myneni L31:
THORPEX-Pacific Workshop Kauai, Hawaii Polar Meteorology Group, Byrd Polar Research Center, The Ohio State University, Columbus, Ohio David H. Bromwich.
John J. Cassano - University of Colorado Wieslaw Maslowski -Naval Postgraduate School William Gutowski - Iowa State University Dennis Lettenmaier – University.
Background Tropopause theta composites Summary Development of TPVs is greatest in the Baffin Island vicinity in Canada, with development possibly having.
The first 2 terms on the RHS are nonlinear terms in the bias. The group labeled THF are transient heat advection bias. Q^ is the bias in diabatic heating.
LHS are the linear terms. First 2 terms on the RHS are nonlinear terms in the bias. The group THF are transient heat advection bias. Q^ is the bias in.
Cyclone composites in the real world and ACCESS Pallavi Govekar, Christian Jakob, Michael Reeder and Jennifer Catto.
Regional Climate Modeling in the Source Region of Yellow River with complex topography using the RegCM3: Model validation Pinhong Hui, Jianping Tang School.
© Crown copyright /0653 Met Office and the Met Office logo are registered trademarks Met Office Hadley Centre, FitzRoy Road, Exeter, Devon, EX1.
WRF SLP Bias Issues Update from Santa Cruz meeting Matt Higgins John Cassano May 18,
HWRF ERROR ANALYSIS T N Krishnamurti A.Thomas A. Simon Florida State University.
NARCCAP Users Meeting April 2011 Results from NCEP-driven RCMs Overview Based on Mearns et al. (BAMS, 2011) Results from NCEP-driven RCMs Overview Based.
Russ Bullock 11 th Annual CMAS Conference October 17, 2012 Development of Methodology to Downscale Global Climate Fields to 12km Resolution.
Collaborative Research: Toward reanalysis of the Arctic Climate System—sea ice and ocean reconstruction with data assimilation Synthesis of Arctic System.
A Comparison of the Northern American Regional Reanalysis (NARR) to an Ensemble of Analyses Including CFSR Wesley Ebisuzaki 1, Fedor Mesinger 2, Li Zhang.
Comparison of Different Approaches NCAR Earth System Laboratory National Center for Atmospheric Research NCAR is Sponsored by NSF and this work is partially.
Dataset Development within the Surface Processes Group David I. Berry and Elizabeth C. Kent.
Collaborative Research: Towards Advanced Understanding and Predictive Capability of Climate Change in the Arctic using a High- Resolution Regional Arctic.
1.Introduction Prediction of sea-ice is not only important for shipping but also for weather as it can have a significant climatic impact. Sea-ice predictions.
The climate and climate variability of the wind power resource in the Great Lakes region of the United States Sharon Zhong 1 *, Xiuping Li 1, Xindi Bian.
Justin Glisan Iowa State University Department of Geological and Atmospheric Sciences RACM Project Update: ISU Atmospheric Modeling Component: Part 1 3rd.
WRF Problems: Some Solutions, Some Mysteries Cliff Mass and David Ovens University of Washington.
Towards development of a Regional Arctic Climate System Model --- Coupling WRF with the Variable Infiltration Capacity land model via a flux coupler Chunmei.
The European Heat Wave of 2003: A Modeling Study Using the NSIPP-1 AGCM. Global Modeling and Assimilation Office, NASA/GSFC Philip Pegion (1), Siegfried.
Simulated and Observed Atmospheric Circulation Patterns Associated with Extreme Temperature Days over North America Paul C. Loikith California Institute.
An example of vertical profiles of temperature, salinity and density.
Southern California February 9, 2002 MISR mesoscale climate dynamics in Southern California Sebastien Conil Alex Hall IRI, April 4, 2006.
Mixed Layer Ocean Model: Model Physics and Climate
Evapotranspiration Estimates over Canada based on Observed, GR2 and NARR forcings Korolevich, V., Fernandes, R., Wang, S., Simic, A., Gong, F. Natural.
International Workshop on Antarctic Clouds Columbus, OH Polar Meteorology Group, Byrd Polar Research Center, The Ohio State University, Columbus, Ohio.
PAPER REVIEW R Kirsten Feng. Impact of global warming on the East Asian winter monsoon revealed by nine coupled atmosphere-ocean GCMs Masatake.
NARCCAP Meeting September 2009 Results from NCEP-driven RCMs ~ Overview ~ Results from NCEP-driven RCMs ~ Overview ~ William J. Gutowski, Jr. & Raymond.
Fig Decadal averages of the seasonal and annual mean anomalies for (a) temperature at Faraday/Vernadsky, (b) temperature at Marambio, and (c) SAM.
An Examination Of Interesting Properties Regarding A Physics Ensemble 2012 WRF Users’ Workshop Nick P. Bassill June 28 th, 2012.
The first 2 terms on the RHS are nonlinear terms in the bias. The group labeled THF are transient heat advection bias. Q^ is the bias in diabatic heating.
Institut für Küstenforschung I f K Numerical experimentation with regional atmospheric models Hans von Storch and Ralf Weisse 8IMSC, Lüneburg, 15. March.
Effect of the Gulf Stream on Winter Extratropical Cyclones Jill Nelson* and Ruoying He Marine, Earth, and Atmospheric Sciences, North Carolina State University,
One-year re-forecast ensembles with CCSM3.0 using initial states for 1 January and 1 July in Model: CCSM3 is a coupled climate model with state-of-the-art.
Analysis of Typhoon Tropical Cyclogenesis in an Atmospheric General Circulation Model Suzana J. Camargo and Adam H. Sobel.
Arctic climate simulations by coupled models - an overview - Annette Rinke and Klaus Dethloff Alfred Wegener Institute for Polar and Marine Research, Research.
Atmospheric Circulation Response to Future Arctic Sea Ice Loss Clara Deser, Michael Alexander and Robert Tomas.
2. WRF model configuration and initial conditions  Three sets of initial and lateral boundary conditions for Katrina are used, including the output from.
ESSL Holland, CCSM Workshop 0606 Predicting the Earth System Across Scales: Both Ways Summary:Rationale Approach and Current Focus Improved Simulation.
Development of the Regional Arctic Climate System Model (RACM) --- Department of Civil and Environmental Engineering University of Washington May, 2010.
IASC Workshop Potsdamr, Germany Polar Meteorology Group, Byrd Polar Research Center, The Ohio State University, Columbus, Ohio, USA The Arctic System Reanalysis.
IC2_I Scenarios of future changes in the occurrence of extreme storm surges Nilima Natoo A. Paul, M. Schulz (University of Bremen) M.
Towards development of a Regional Arctic Climate System Model ---
Changes in the Melt Season and the Declining Arctic Sea Ice
Upper Rio Grande R Basin
Pogoreltsev A., Ugrjumov A..
Mid Term II Review.
Climate and Global Dynamics Laboratory, NCAR
Coupled atmosphere-ocean simulation on hurricane forecast
Overview of 2016/17 Winter Climate over South Korea
The El Niño/ Southern Oscillation (ENSO) Cycle Lab
Performance of the VIC land surface model in coupled simulations
ATMS790: Graduate Seminar, Yuta Tomii
On HRM3 (a.k.a. HadRM3P, a.k.a. PRECIS) North American simulations
Integration of NCAR DART-EnKF to NCAR-ATEC multi-model, multi-physics and mutli- perturbantion ensemble-RTFDDA (E-4DWX) forecasting system Linlin Pan 1,
1 GFDL-NOAA, 2 Princeton University, 3 BSC, 4 Cerfacs, 5 UCAR
Department of Geological and Atmospheric Sciences
Presentation transcript:

Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d] Bias score at selected thresholds Pan-Arctic WRF Background Introduction Efforts to develop a coupled atmosphere – ocean – sea ice – land Arctic regional model, known as the Regional Arctic Climate Model (RACM), are on-going. The atmospheric model in RACM is a version of the National Center for Atmospheric Research (NCAR) Weather Research and Forecasting (WRF) model that has been optimized for the polar regions. In the fully coupled RACM, the NCAR CCSM flux coupler CPL7 links the WRF atmosphere to the POP/CICE ocean and VIC land models. Here we present tests of stand-alone WRF done for two purposes: To develop a baseline version suitable for the pan-Arctic domain To assess couplings between the surface and atmosphere in initial one-way simulations in which ocean and ice properties are specified. For these simulations, we use WRF 3.1 with WRF’s built in NOAH land model. 1. RACM The simulation domain of RACM covers the entire pan-Arctic region and includes all sea ice-covered regions in the Northern Hemisphere and all terrestrial drainage basins that flow into the Arctic Ocean. The ocean and sea ice models use a horizontal grid spacing of less than 10 km, while the atmosphere and land models use a horizontal grid spacing of 50 km or less. Domains Figure: Model domains used in evaluating WRF for RACM pan-Arctic simulations. The domain indicated with the shaded topography is the model domain that will be used for fully coupled regional Arctic climate system model simulations. Colored outlines represent boundaries for the other model domains used when evaluating WRF. 2. CORDEX We are also contributing to the WCRP Coordinated Regional Downscaling Experiment (CORDEX). The baseline CORDEX simulation uses the pan-Arctic domain below, with a 50-km grid. Figure: The CORDEX pan-Arctic domain. The region interior to the “sponge zone” is outlined by the yellow box.

In order to determine if the large errors illustrated in the figure above were specific to the model domain being used, several simulations were run using smaller or larger domains than the RACM domain. In these simulations the large errors evident above persisted with slight changes in location and/or intensity. Only simulations run on much smaller pan-Arctic domains (e.g., CORDEX domain) had small errors. Given the large errors that occur in long-duration climate-type simulations with WRF on a large pan-Arctic model domain additional simulations using a variety of data assimilation techniques were evaluated. The use of data assimilation to constrain the large-scale atmospheric features to be consistent with lateral boundary conditions is common for many regional climate models and was viewed as critical for conducting fully coupled simulations with RACM in the future. Results from a WRF simulation that used the same model physics options as was used for the simulations shown above, but that also used spectral nudging of wind and temperature for wave numbers 1 and 2 showed drastically reduced errors across the entire domain (below). For this simulation spectral nudging was applied in the top 20 (out of 40 total) vertical levels in WRF, with the strength of nudging linearly increased from zero at level 20 to full value at level 30. Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d] Bias score at selected thresholds Pan-Arctic WRF RACM Domain Evaluation An extensive suite of simulations with WRF were conducted to determine biases in the WRF simulated atmospheric state. 3-member ensembles were run for the months of January, April, July, and October 2007 using a variety of model physics options and pan-Arctic domains. The basline domain is the RACM domain appearing on the previous panel. All simulations were compared to NCEP reanalysis fields of pressure, temperature, humidity, and geopotential height at several levels in the atmosphere. After completing 24 different sets of simulations using a range of radiation, cloud microphysics, boundary layer, and land parameterizations it became obvious that WRF, when run on a large pan-Arctic domain, develops significant circulation biases in the North Pacific, and occasionally in the North Atlantic, storm track regions. The figure below shows results from the WRF simulation that has the smallest errors of all parameterization combinations evaluated. Figure: Overlay of monthly mean sea level pressure (SLP) contours from the NCEP reanalysis (red lines) and 3-member WRF ensemble (blue lines) (left panel) and difference between the monthly mean SLP (WRF ensemble minus NCEP reanalysis) (right panel) for January Large positive biases in SLP are present in the North Pacific and to a lesser extent in the North Atlantic. Figure: Same as the figure above except for WRF simulation using spectral nudging of wave numbers 1 and 2. Bias Problem Problem Solution

Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d] Bias score at selected thresholds Pan-Arctic WRF CORDEX Arctic Domain January – August 2007 Simulations using ERA-Interim Reanalysis and NSIDC Fractional Sea Ice right top: Plotted 500-hPa geopotential height root mean square difference (RMSD) between Pan-Arctic WRF and ERA-Interim Reanalysis (blue line) For comparison, observation RMSD (red line) is plotted right bottom: Plotted MSLP root mean square difference (RMSD) between Pan-Arctic WRF and ERA- Interim Reanalysis (blue line) For comparison, observation RMSD (red line) is plotted. Reanalysis RMSD is computed by taking the RMS difference between the ERA-Interim daily MSLP and corresponding monthly mean values of the reanalysis. Reanalysis RMSD is useful in determining the amount of internal variability in the atmosphere. Magnitudes of both RMSD curves correspond to a high degree. Thus, the model versus reanalysis RMSD may have a considerable dependence on unforced, quasi- random internal variability. above: Monthly plots of mean 500-hPa height biases. The weakest bias couplet is found in January while the strongest exists in April. Each of the first three plot groups contain: top left: Pan-Arctic WRF (PAW) monthly MSLP contours (blue) plotted against Era-Interim monthly MSLP (red). top right: PAW Monthly MSLP. bottom left: PAW - Era-Interim Monthly MSLP bias. Large MSLP anomalies over regions of high topography are as a result of the different algorithms (WRF vs Era-Interim used to convert station pressure to SLP. bottom right: Pan-Arctic WRF Monthly MSLP. Mean Monthly 500-hPa Geopotential Height Bias January 2007 April 2007 July 2007 time step Geopotential meters hPa July 2007 January 2007April 2007

Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d]Bias score at selected thresholds Two-meter, 850-hPa, and 500-hPa Temperature July April January Each temperature bias plot set contains: 2-m (left), 850-hPa (center), and 500-hPa (right). top: Monthly mean temperature bias for January Surface anomalies are largest over regions of sea ice and snow cover. Aloft, there is good agreement between PAW and Era-Interim, thus biases are minimal. center : Monthly mean temperature bias for April The surface plot shows the largest negative bias over the East Siberian Sea into the Barents Sea. Large cold biases may be a result of the treatment of sea ice and snow in PAW. bottom: Monthly mean temperature bias plots July All three plots show good agreement between the model output and reanalysis. July January April Each of the first three plot groups contain: left: Monthly mean RMSD sea-level pressure right: Monthly mean RMSD 500-hPa geopotential heights. The largest RMS differences appear in April. However, when compared with RMSD time series on the previous panel, a considerable amount of variability can be attributed to model/reanalysis; both RMSD curves are comparable in magnitude to their respective reanalysis RMSD curves. Pan-Arctic WRF CORDEX Arctic Domain January – August 2007

Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d] Bias score at selected thresholds Figure: WRF-binary ensemble minus ERA- Interim MSLP (hPa) for A. September and B. October. Differences are relatively small vs. RMSD above, except for land areas, where differences occur between WRF and ERA MSLP algorithms. WRF-fractional minus WRF-binary ensemble MSLP differences (hPa) for C. September and D. October. Differences are largest in October, a month with relatively warm sea- surface temperature, cooling atmospheric temperatures, and relatively large fractions of open ocean.. Figure: RMS differences (hPa) between daily mean sea-level pressure (MSLP) and the monthly average MSLP from the same source (ERA-Interim, WRF- binary,WRF-fractional) A. September, B. October. The simulations reproduce well the observed behavior. The RMSDs collectively give a measure of unpredictable internal variability or “noise”. D.D. B. A.A. C. F. A.B. In this study, we examine effects of ice treatment in PAW, comparing simulations allowing fractional sea ice in each 50-km grid box versus those using the original WRF treatment allowing only 0% or 100% ice cover in a grid box (binary sea ice). WRF used ERA-Interim to simulate June-Dec 2007, encompassing a period when there was substantial fractional sea ice in the Chukchi/Beaufort Sea region. Two four-member ensembles were simulated: one using binary sea ice everywhere and the other using fractional sea-ice in the Chukchi/Beaufort Sea region and binary elsewhere. Start times for each member were staggered by one day, from late May Pan-Arctic WRF CORDEX Arctic Domain Sea-Ice Treatment Figure: Same as the panels above, but for 500 hPa heights (m). Differences from the reanalysis are a little larger than the internal variability (see previous panel), though less so in October. Differences due to ice treatment are deeper in the atmosphere in October than September. C. D.D.

Bias score versus precipitation threshold [mm/d] Model - observed precipitation [mm/d] Bias score at selected thresholds K.K. L. Pan-Arctic WRF CORDEX Arctic Domain Sea-Ice Treatment & Summary These figures are also from the fractional versus binary sea-ice study described in the previous panel. A.B. C. D.D. Figure: WRF-binary ensemble minus ERA- Interim 2-m temperature (˚C) for A. September and B. October. Differences partly reflect WRF difficulties in simulating snow-covered sea ice. WRF-fractional minus WRF-binary ensemble 2-m temperature differences (˚C) for C. September and D. October. Differences do not align with surface sensible heat flux differences (below) due to advection of heat by the anomaly circulation (previous panel). A.B. Figure: Same as the panels above, but for surface sensible heat flux (W/m 2 ), positive downward. Region of largest differences occurs where there are differences in sea-ice treatment between the two ensembles. 1.Pan-Arctic WRF simulations on the large RACM domain perform well when using spectral nudging in the upper half of the model atmosphere. Without nudging, substantial bias appears in monthly sea-level pressure. 2.Pan-Arctic WRF simulations on the smaller CORDEX domain perform well without need for nudging. 3.Substantial differences appear between an ensemble of simulations that allow fractional sea ice and an ensemble that does not. The largest differences occur in October when the ocean is relatively warm, the atmosphere is cooling, and fractional ice cover covers a relatively large area. Summary C. D.D.