1 / 46

The Configuration, Physics, and Optimal Initialization of the

The Configuration, Physics, and Optimal Initialization of the Unified Noah-OSU Land Surface Model for WRF. Ken Mitchell Mike Ek, Fei Chen, George Gayno, Dag Lohmann, John Smart, Jerry Wegiel Brent Shaw, Qingyun Duan, Jinwon Kim, Tom Black, Ying Lin, Eric Rogers.

alaula
Télécharger la présentation

The Configuration, Physics, and Optimal Initialization of the

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. The Configuration, Physics, and Optimal Initialization of the Unified Noah-OSU Land Surface Model for WRF Ken Mitchell Mike Ek, Fei Chen, George Gayno, Dag Lohmann, John Smart, Jerry Wegiel Brent Shaw, Qingyun Duan, Jinwon Kim, Tom Black, Ying Lin, Eric Rogers NCEP, NCAR, AFWA, FSL, OHD, OSU, UCLA The WRF Land Surface Modeling Working Group WRF Workshop at NCAR 25-26 June 2002 NCEP: Where America's Climate and Weather Services Begin

  2. Motivation for Land Surface Modeling Attention in WRF • Land surface forcing is a primary forcing source for mesoscale and microscale circulations • Horizontal gradients of surface evaporation and sensible heat flux (soil moisture, vegetation, snow) • Driver of PBL development • Strong control on convective instability • The primary target resolution for WRF: 1-10 km • Observation density is woefully inadequate to initialize mesoscale and microscale circulations • We depend on cycled 4DDA to spin-up realistic mesoscale and microscale circulations • Cycled 4DDA is also the only practical source of realistic initial land surface states and land surface heat fluxes

  3. Logical choices for first LSM options in WRF: Noah, RUC, CLM • Those LSMs routinely used as the background land model in continuous 4DDA (realtime, ops, retrospective) • NCEP: Noah LSM(North and Central America) • Realtime: EDAS and N-LDAS • Retrospective: Regional Reanalysis (EDAS, 25-yr) and N-LDAS (5-yr) • AFWA: Noah LSM(Global) • Realtime: AGRMET • Retrospective: AGRMET (3-yr) • FSL: RUC LSM • Realtime: RUC • NCAR: Noah LSM (soon) • Realtime: HRLDAS • NASA/GSFC: (Noah, CLM, VIC, MOSAIC, CATCHMENT) • Realtime:

  4. Coupled Models Executing the Noah LSM • Operational at NCEP • Eta Model (12-km, 60-layer) and Eta Data Assimilation System (EDAS) • 25-year Regional Reanalysis System (32-km, 60-layer) • Testing at NCEP • MRF • GFDL hurricane model • WRF model (Janjic non-hydrostatic meso dynamic core) • Testing at AFWA • MM5 and WRF • Testing at NCAR • WRF and MM5 • Testing at Navy Research Lab • NOGAPS global model • Testing at University of Alaska • MM5 model • CAPS • ARPS model

  5. The Lineage of the Noah LSM • OSU LSM (1990) • Implemented in NCEP MRF and Global Reanalysis (1993) • Implemented in AFWA AGRMET (1990) • NCEP “Modified OSU” LSM (1996) • Canopy resistance, infiltration, bare soil evaporation • Chen, Mitchell, et al. (JGR, 1996) • Implemented in NCEP ops Eta model (Jan 1996) • Implemented in NCEP ops EDAS (Jun 1998) • Implemented in NCAR Community MM5 (1997) • Implemented in NCAR Community WRF (2002) • NCEP “Noah” LSM (1999) • Frozen soil physics, snowpack upgrades, ground heat flux upgrade • Koren et al (1999, JGR), Mitchell et al (2000, 2002) • Implemented in Eta and EDAS (Jul 2001) • Implemented in AFWA AGRMET (Jul 2001) • NCEP/NCAR/AFWA/OSU “Unified Noah” LSM (2002) • For WRF, Eta, MM5 (WRF “beta” release target Fall 2002) • For EDAS, AGRMET, NLDAS, HRLDAS

  6. ETA/NOAH LAND-SURFACE MODEL UPGRADES: 24 Jul 01 - assimilation of hourly precipitation -- hourly 4-km radar/gage analysis (Stage IV) - cold season processes(Koren et al 1999) -- patchy snow cover -- frozen soil (new state variable) -- snow density (new state variable) - bare soil evaporation refinements -- parameterize upper sfc crust cap on evap - soil heat flux -- new soil thermal conductivity (Peters-Lidard et al 1998) -- under snowpack (Lunardini, 1981) -- vegetation reduction of thermal cond. (Peters-Lidard et al 1997) - surface characterization-- maximum snow albedo database (Robinson & Kukla 1985) -- dynamic thermal roughness length refinements - vegetation -- deeper rooting depth in forests -- canopy resistance refinements NOAH LSM tested in various land-model intercomparison projects, e.g., GSWP, PILPS 2a, 2c, 2d, 2e, Rhone, and (near-future) DMIP.

  7. REDUCING SURFACE COOL BIAS OVER MELTING SNOW FEB 2001 ETA MODEL RETROSPECTIVE RUNS snow melt North Platt, Neb. North Platt, Neb. =0 C >0 C skin temp obs>0 C 2-m air temp 0 C 0 C obs,model>0 C model0 C 18Z 18Z 2-m air temp, new formulation 2-m air temp, current formulation warm advection/melting snowpack case: 00Z 02 FEB 2001, 60-hr model run old model formulation (upper left) => bulk of incoming energy melts/sublimates snow => skin temp held at freezing => 2-m air temp held near freezing new model formulation(upper right) => patchy snow cover for snow depth less than threshold depth (veg-type dependent) => reduces surface albedo => more available energy at sfc => skin temp can exceed 0 C => 2-m air temp rises further above freezing.

  8. Mean diurnal cycle of 2-m air temperature of observations and Eta model 48-hr forecast from 12Z, averaged over 30-day WINTER period of 01 Feb – 01 Mar 2001 at all surface stations over East U.S. Station OBS: solidOPS Eta/NOAH: short dashTEST Eta/NOAH: long dash) +5 Temperature (C) -3 0 48 Forecast Hour

  9. REDUCING SURFACE MOIST-COOL BIAS OVER WET-BARE GROUND 12Z, 27 APR 2001, 60-hr model run OLD 2-meter T=> Td=> NEW 2-meter T=> Td=> Champaign, Illinois 36-hr old model formulation - cool, moist bias in 2-m T, Td new model formulation – reduced cool, moist bias better ABL representation as well

  10. 00 12 24 36 48 REDUCING NEAR-SURFACE MOIST-COOL BIAS OVER WET-BARE GROUND IN SPRING 84 old NOAH LSM USA northern mid-west new NOAH LSM 75 2-m relative humidity (%) 66 57 obs Eta forecast hour Improved 2-m RH in 48-hour diurnal forecast cycle

  11. LSM Development Priorities • Many land surface modeling groups overly emphasize improving LSM physics • Insufficient attention given to: • 1) Optimizing LSM parameters • 2)Initializing LSM Land States

  12. LAND DATA ASSIMILATION SYSTEMS: •Three Broad Approaches to Land Data Assimilation • 1) Coupled Land/Atmosphere 4DDA: GDAS, Global Reanal-1 • precipitation forcing at land surface is from parent atmospheric model • surface insolation at land surface is from parent atmospheric model • precipitation/insolation may have large bias: >large soil moisture bias • 2)Uncoupled Land 4DDA (land model only):NLDAS,AGRMET • observed precipitation/insolation used directly in land surface forcing • 3)Hybrid Land 4DDA: EDAS, Global Reanal-2 • Coupled land/atmosphere, but observed precipitation replaces model precipitation for driving the land surface

  13. AFWA Global LDAS Example (Top 5 cm soil moisture, from late Oct 2000) Forced by gage/satellite precip analysis and satellite-based surface solar insolation, executing NCEP NOAH LSM -- Blue is wilting point. Red is maximum value.

  14. ETA MODEL LAND-SURFACE MODELING MILESTONES Since 1996, a series of GCIP/GAPP-sponsored land-surface model related advances have been made to the NCEP mesoscale Eta model and its Eta-based 4-D data assimilation system (EDAS). 31 Jan 1996 multi-layer soil/vegetation/snow model introduced adjusted initial soil imoisture/temperature from GDAS 18 Feb 1997 new vegetation greenness database from NESDIS refined adjustment of initial GDAS soil moisture refined evaporation over snow and bare soil 09 Feb 1998 increase from 2 to 4 soil layers (10, 30, 60, 100 cm layers) 03 Jun 1998 full self-cycling of EDAS/Eta soil moisture and temperature new NESDIS daily 23-km snow cover and sea ice 15 Mar 1999 "NOAH" name designated for Eta land-surface model 01 Apr 1999 GOES vs Eta skin temperature verification 24 Mar 2000 Eta near-surface regional Forecast Verification System 15 Mar 2001 retrospective NOAH LSM Eta/EDAS testing initiated 24 Apr 2001 realtime NOAH LSM Eta/EDAS testing initiated 02 July 2001 pre-implementation NOAH LSM testing in parallel Eta/EDAS 24 July 2001 operational implementation of 1) NOAH LSM and 2) theassimilation of precipitation in the EDAS

  15. 15 JUL 98 OPS EDAS: 15-DAY OBS PRECIP (1-15 JUL 98) 1-15 JUL OPS EDAS: 15-DAY PRECIP SOIL MOISTURE (e) (c) (a) 15 JUL 98 TEST EDAS: TEST EDAS: 1-HR STAGE IV PRECIP 1-15 JUL 15-DAY PRECIP SOIL MOISTURE (b) (d) (f) IMPACT OF HOURLY PRECIPITATION ASSIMILATION IN ETA MODEL Figure 8. (a) 1-15 July 1998 gage-observed total precip (mm), (b) 'snapshot' of hourly Stage IV radar/ Gage precip (06Z, 15 July 1998); EDAS total precip of 1-15 July 1998 for (c) control run without precip assim, and (d) test run with hourly Stage IV precip assim; EDAS soil moisture availability (% saturation) of top 1-m soil column valid at 12Z 15 July 1998 (e) without precip assim, and (f) with precip assim.

  16. NCEP Eta model forecast during July 1998: Texas/Oklahoma drought, 24-hour forecast valid 00Z 27 July 1998 In late July 1998, after nearly two months of self-cycling the land states in the EDAS, the Eta model successfully captured the extremely dry soil moisture (upper left) and warm soil temps (upper right) over the Texas/Oklahoma region, yielding forecasts of high 2-m air temps (lower left) and deep, dry, hot boundary layers that verified well against raobs (e.g., at Norman, OK – lower right). soil moisture availability (1-m) soil temperature (5-cm) air temperature (2-meter) Norman, OK sonde (obs=solid, model=dashed)

  17. Eta model Oct 2001 end-of-month soil moisture states (left) and monthly observed total precipitation (right).

  18. Mean diurnal cycle of 2-m air temp: obs (solid) and Eta model 0-48 hour forecast (dashed) from all 12Z runs, over full month of OCT 01 and all reporting surface stations over a) southern Great Plains (top frame) and southern Midwest (bottom frame) 24 Temperature (C) 12 23 Temperature (C) 13 0 Forecast Hour 48

  19. Monthly 90-day precipitation departures from Normal December October February April

  20. End of Month Eta/EDAS 40-100 cm Soil Moisture (% Vol) Oct Dec Apr Feb

  21. Interannual variability of North American Monsoon - interior Southwest moist dry 30 33 C 29 C 32 32 23 24 24 16 16 16 00 00 12 12 24 24 36 36 48 48 00 12 24 36 48 July 2001 July 1999 July 2000 semi-dry semi- dry moist dry 32 C obs obs Eta Eta Eta forecast hour

  22. N. American LDAS: N-LDAS(NCEP uncoupled LDAS demonstration ) • 1. Force models with Eta model 4DDA analysis (EDAS) meteorology, except use actual observed precipitation (gage-only daily precip analysis disaggregated to hourly by radar product) and hourly downward solar insolation (derived from GOES satellites). • 2. Use 4 different land surface models: • MOSAIC (NASA/GSFC) • NOAH (NOAA/NWS/NCEP) • VIC (Princeton University/University of Washington) • Sacramento (NOAA/OHD) • 3. Evaluate results with all available observations, including soil moisture, soil temperature, surface fluxes, satellite skin temperature, snow cover and runoff.

  23. NCEP/EMC NWS/OHD NESDIS/ORA Dan Tarpley Ken Mitchell Dag Lohmann Andy Bailey NASA/GSFC Princeton Univ. Paul Houser Brian Cosgrove Eric Wood Justin Sheffield Univ. Washington Univ. Oklahoma Rutgers Univ. NOAA/ARL Dennis Lettenmaier Ken Crawford Jeff Basara John Schaake Qingyun Duan Alan Robock Lifeng Luo NCEP/CPC Tilden Meyers John Augustine Univ. Maryland Wayne Higgins Huug Van den Dool Rachel Pinker N-LDAS Collaborators NOAA NASA Universities http://ldas.gsfc.nasa.gov

  24. LDAS Run Modes:1) Realtime, 2) Retrospective • REALTIME: 15 Apr 1999 to 15 Dec 2001 • -- NCEP realtime forcing • RETROSPECTIVE: 01 Oct 1996 to 30 Sep 99 • -- Mandated largely by spin-up issues • -- NASA-assembled retrospective forcing • --- Higgins NCEP/CPC reprocessed precipitation forcing: • ---- more gages obs, more QC • --- Pinker U.Md reprocessed solar insolation forcing • ---- better cloud screening, more QC • Rutgers University compared the soil moisture, soil temperature, surface flux results from the retrospective LDAS runs to observations over Oklahoma/Kansas for last retro year.

  25. LDAS Soil Wetness Comparison LDAS retrospective output example (similar spread as in PILPS-2c)

  26. LDAS-NOAH Skin Temperature October 2001 Validation cont. Region 5 Region 2 15 Z 21 Z

  27. Snowpack Simulation Comparison Snow depth from USAF, cover: global 1/8 bedient, unit [in], daily Snow cover product from NESDIS daily, cover: 1/16 bedient N.Hemisphere grid, flag = estimated = future

  28. LDAS Models Streamflow 02192000 = Broad River, GA, 1430 sq. miles 01631000 = Shenandoah River, VA, 1642 sq. miles 01503000 = Susquehanna River, NY, 2232 sq. miles

  29. Questions • How similar are soil moisture fields from different LDAS LSMs ? • Disturbingly dissimilar! • Can the soil moisture field “spun-up” from one LSM be used to initialize another LSM? • NO!! • How long should the LSM spin-up period be ? • 1-2 years!! • Forced with observed precipitation!!

  30. Location of LSASWStudy Area

  31. Soil Moisture Anomaly Validation

  32. LSASW Average Soil Water Content

  33. VIC Simulation with Unmatched Local Soil Type(at sand site MANG)(Note: observed soil moisture somewhat suspect at all sand sites)

  34. VIC Simulation with Soil Type Matching Local Type(at clay-loam site ALTU)

  35. Model Simulated Sensible Heat Flux at the Norman Site (13 July 2000) • The Maximum Value of Sensible Heat Flux is ~480 W m-2 for the Simulation Using the Clay Loam Parameterization and Minimum Values of Soil Water Content Determined from field Samples. • Most Simulations Using the Silt Loam Parameterization Peak near 65 W m-2.

  36. Simulated Depth of the PBL at the Norman Site (13 July 2000) • PBL Depth after 12 hours varied over 900 meters between the Simulations Using the Silt Loam Parameterization and Field Sample Maximum Values of Soil Water Content and the Clay Loam Parameterization Using Field Sample Minimum Values of Soil Water Content. Observed PBL Depth

  37. RECOMMENDATION / ADMONITION / PLEA • We must broaden our perception of “acquiring initial land states” for coupled land/atmosphere models (Eta, MM5, WRF) • Not just naïve extraction of soil moisture and soil temperature from the most readily available database: e.g. Global Reanalysis • Acquiring initial land states must also include acquiring/considering soil types/parameters, vegetation types/parameters/greenness, background LSM, terrain heights, frozen soil

  38. WARNING!! Far too many groups executing and testing coupled land/atmosphere regional models are initializing their land states naively and dangerously from the land states of the NCEP/NCAR Global Reanalysis.

  39. Optimum Initialization of Land Statesin Coupled Land/Atmosphere Models • Execute exactly the same land physics model in your coupled forecast model as that in the LDAS from which you are extracting land state initial conditions • On exactly the same horizontal grid • Assign exactly the same land surface characteristics and exactly the same parameter values • Soil class, Vegetation class, porosity, wilting point, etc • With exactly the same terrain height field • Avoid the “hot mountain problem” • Assign your soil and veg classes and parameters from the same database from which you extract initial land states • soil and vegetation classification and maps • porosity, wilting point, rooting depth, greenness, LAI, etc • e.g. from EDAS, AGRMET, Regional Reanalysis • Retrieved soil temperature and frozen soil moisture profiles must be adjusted for model vs LDAS terrain height difference • Drive your hybrid or uncoupled LDAS with observed precipitation

  40. WRF LSM Milestones • Present Release of WRF • LSM choices • “5-layer/slab”, fixed summer/winter sfc wetness • “Modified OSU” • WRF SI land-state inputs • GDAS/MRF/AVN postprocessed output • EDAS/Eta postprocessed output • Nov 2003 Target WRF Release • Additional LSM choices • “Unified Noah” • RUC • Additional WRF SI land state inputs • EDAS native grid files (from cycled “Unified Noah”) • AGRMET native grid (virtually) files (from cycled “Unified Noah”) • RUC native grid files (from cycled RUC LSM)

  41. LDAS Scientific Questions • 1. Can land surface models forced with observed meteorology and radiation reproduce point-wise soil moisture/temperature states and surface fluxes? • If not, what are the relative contributions to the differences between models and observations owing to a) errors in the soil-state/surface-flux observations or b) differences in the following between model and observed: • a. Forcing? no • b. Soil properties? yes • c. Vegetation characteristics? yes • d. Scales of representativeness? some • e. Vertical representation? yes • f. Other (e.g. tiling, variable infiltration assumptions)

More Related