NCEP/EMC Land-Surface and PBL Parameterization Schemes By Curtis Marshall NCEP/EMC COMAP Symposium 00-1 Tuesday, 14 December 1999

Outline of Presentation  Land Surface Physics _ Observational examples and relevance to NWP _ Attributes of NCEP land-surface physics (NOAH model) _ Milestones of land-surface physics upgrades  PBL Physics _ Attributes of PBL physics  Recent Verification of Land-Surface / PBL schemes  Future Work

Land-Surface Physics

Is the Land Surface Important to NWP?  “The atmosphere and the upper layers of soil or sea form together a united system. This is evident since the first few meters of ground has a thermal capacity comparable with 1/10 that of the entire atmospheric column standing upon it, and since buried thermometers show that its changes for temperature are considerable. Similar considerations apply to the sea, and to the capacity of the soil for water. “ L.F. Richardson, 1922 Weather Prediction by Numerical Processes  “Much improved understanding of land-atmosphere interaction and far better measurements of land-surface properties, especially soil moisture, would constitute a major intellectual advancement and may hold the key to dramatic improvements in a number of forecasting problems, including the location and timing of deep convection over land, quantitative precipitation forecasting in general, and seasonal climate prediction.” National Research Council, 1996

Goals of Improved Land-Surface Physics  Better diurnal cycle of surface heating and evaporation (2 meter TAIR and TDEW)  Reproduce diurnal growth and decay of PBL  Improved convective index forecasts  Better QPF  Expand use of model outputs for hydrologic and agricultural applications (runoff, snowmelt, soil moisture and temperature)

Notable Examples Examples of the influence of land- surface processes on the atmosphere in both models and observations

Atmospheric signature over Oklahoma wheat fields (dark green area from north-central through southwest Oklahoma) during peak greenness.

 Relatively moist, cool PBL over wheat fields  Densely cultivated vegetation increases evapotranspiration  Sun’s energy used less for sensible heating  Result: surface layer more moist than surrounding areas by as much as 10 F  Result: surface layer cooler than surrounding areas by a few degrees

1998 Texas /Oklahoma Drought 10% Moisture Availability over Region by late July 1998

Parched, dry ground heats quickly under afternoon insolation. Note very warm Eta model soil temperatures over the Red River Valley

The hot, dry ground results in large sensible heat flux into the PBL, with very hot 2 meter temperatures (>40 C) over the area

So what does a land-surface scheme do? _ Provides albedo for calculating reflected shortwave radiation _ Calculates evapotranspiration (latent heat flux) from soil and vegetation canopy _ Provides ground surface (“skin”) temperature for determining surface sensible heat flux and upward longwave radiation _ Determine impact of snowpack on surface radiation and heat budgets  THE UPSHOT: PROVIDE MORE REALISTIC SURFACE FLUXES TO PBL SCHEME THAN OLDER, SIMPLE TREATMENTS (e.g, NGM)

Attributes of Eta Land-Surface Physics  4 soil layers (10, 30, 60, 100 cm thick) – predict soil moisture/temperature – Continuous 3-hour update in fully cycled EDAS  Explicit vegetation physics – 12 vegetation classes over Eta domain – annual cycle of vegetation greenness  Explicit snowpack physics – prognostic treatment of snowmelt  COMING SOON: – frozen ground (soil ice) treatment and patchy snow – explicit streamflow routing

 Key Assumption: Surface Energy Balance:

Prognostic Equations  Soil Moisture: – “ Richard’s Equation” for soil water movement – D, K functions (soil texture) – F  represents sources (rainfall) and sinks (evaporation) _ Soil Temperature – C, Kt functions (soil texture, soil moisture) – Soil temperature information used to compute ground heat flux

Operational Soil Texture Database

Evapotranspiration Treatment WHERE: E = total evapotranspiration from combined soil/vegetation E dir = direct evaporation from soil E t = transpiration through plant canopy E c = evaporation from canopy-intercepted rainfall

Evapotranspiration (continued) _ These terms represent a flux of moisture, that can be parameterized in terms of “resistances” to the “potential” flux. Borrowing from electrical physics (Ohm’s Law): FLUX = POTENTIAL/RESISTANCE _ Potential ET can roughly be thought of as the rate of ET from an open pan of water. In the soil/vegetation medium, what are some resistances to this? – Available amount of soil moisture – Canopy (stomatal) resistance: function of vegetation type and amount of green vegetation) – atmospheric stability, wind speed

Canopy Resistance Issues _ Canopy transpiration determined by: – Amount of photosynthetically active (green) vegetation. Green vegetation fraction (  f ) partitions direct (bare soil) evaporation from canopy transpiration: E t /E dir ≈ f(  f ) – Green vegetation in Eta based on 5 year NDVI climatology of monthly values – Not only the amount, but the TYPE of vegetation determines canopy resistance (R c ):

Canopy Resistance (continued)  Where: Rcmin: f(vegetation type) F1: drying power of the sun F2, F3: drying power of the air mass F4: soil moisture stress  Thus: hot air, dry soil, and strong insolation lead to stressed vegetation!  Eta model uses database of 12 separate vegetation classes

Operational Vegetation Type Database at NCEP

December Green Vegetation Fraction

June Green Vegetation Fraction

Annual Time Series of Green Fraction Over Oklahoma Wheat Country  Early Spring intense green up  Rapid senescence  Harvesting and return of land to near bare soil by early summer

Annual Time Series of Green Fraction over Iowa Corn Fields  Maturity of corn occurs less rapidly than for wheat  Corn harvested much later in the warm season than wheat

Annual Time Series of Green Fraction over Arizona Desert  Not much vegetation to speak of year around!  Any evaporation in model is from bare soil

Eta Model Albedo (snow free)

Snow Cover Treatment  Why so important? Marked effect on albedo and hence the surface fluxes  Snow cover / sea ice product from NESDIS analysis ingested daily at 0000 UTC into NCEP models  Prognostic snow depth during Eta integration, but not in NGM (snowfall computed using 5:1 density ratio from model QPF)  Available energy for snowmelt computed from surface energy balance assumptions

More Snow Information  cover: 23-km N. Hemisphere grid  produced daily by human analyst  multiple data sources: –GOES visible –SSMI snow cover –station reports –NIC ice cover –AVHRR visible Example NESDIS snow/ice cover cover: depth:

Milestones of Eta Land-Surface Physics  31Jan 1996 – Multi-layer soil/veg/snow model introduced – Initial soil moisture/temp 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  09 Jun 1998 – full self-cycling of EDAS/Eta soil moisture/temp – new NESDIS daily 23-km snow cover and sea ice

PBL Physics

Purpose of PBL Scheme  Two separate schemes for: – Surface layer (constant flux layer) – PBL turbulence above surface layer  Surface layer – Exchange of heat (water vapor) and momentum with the land surface  PBL turbulence – Vertical dispersion of heat (water vapor) and momentum throughout the PBL

Attributes of PBL Treatment  Surface layer – Monin-Obukhov similarity theory applied to determine exchange coefficient. Use of Paulson (1970) stability functions. Does not allow turbulence to diminish to zero near ground in nighttime hours. – Roughness length for heat differentiated from that for momentum(very important!)  PBL turbulence – Mellor-Yamada level 2.5 turbulence closure – local diffusion

Atmospheric Surface Layer  Sensible Heat Flux Calculation: - Traditional “bulk aerodynamic” approach - Ua = wind speed at first eta surface -  s = “skin temperature”, from land-surface scheme! -  a = Air temperature at first eta surface - Ch, Cd = Exchange coefficients for heat and momentum - diagnosed using “similarity theory”  Momentum Flux:

What the heck is “Similarity Theory”?  An empirical technique for drawing vertical profiles of wind and temperature in the surface layer  Rests on the assumption that all profiles have a “similar”shape that can be adjusted with “scaling parameters”  In practice, scaling parameters used to determine magnitide of the surface exchange coefficient

PBL Above the Surface Layer - Vertical Mixing of heat, moisture, and momentum based on prognostic “turbulent kinetic energy” (TKE): q = TKE ~ (u’) 2 + (v’) 2 + (w’) 2 - Turbulent eddys: - K M, K H (mixing coefficients) use info about TKE (q) - Vertical gradients computed using “local” as opposed to “non- local” information (a local mixing scheme is employed)

Recent Verification and New Initiatives

EMC Forecast Verification System (FVS) Regions

February 1999 Northern Plains (NPL) 2m TAIR In Eta snow physics, no such thing as “patchy” snow cover. Therefore, grid-cell albedo often too large over shallow snow. This plot illustrates that a cool bias results over shallow snow.

April 1999 NPL 2m TAIR and RELH Eta model exhibits marked cool, moist bias at 2 meters over moist, relatively bare soils…e.g., over the northern plains in spring.

August 1999 NPL 2m TAIR and RELH By August, vegetation is at peek greenness. Eta model appears to underestimate canopy transpiration, and cool/moist bias in April (when soil is relatively bare) is replaced by a warm/dry bias.

Use of OK Mesonet to examine land-surface physics

Oklahoma Vegetation Fraction During July State of Oklahoma varies widely from west to east in mean annual precipitation. Vegetation cover very different as a result of precip climatology. With 114 station OK Mesonet, the state provides a microcosm for studies of land sfc physics.

Humidity bias over OK Mesonet in July 1997 Clearly shows the moist bias over less vegetated areas, with a dry bias over the eastern green areas. Agrees well with FVS results for the northern plains at different times of year/

Improved soil moisture via continuous self-cycling Prior to June 1998, soil moisture was initialized from the Global Data Assimilation System, resulting in a severe positive bias !!!

Validation of Surface Fluxes Verification of model net radiation (RNET)at Norman, OK shows a positive bias. This positive bias in RNET bias appears to be the result of a high bias in downward shortwave radiation (SWRD).

Validation of Surface Fluxes (continued)  Too much RNET at the surface results in too much available energy for the other fluxes (ground, sensible and latent). Key question: how is this excess being partitioned among the three? Model ground heat flux (FXGH) appears to be underestimated in this case. Thus, excess RNET not being realized in FXGH.

Validation of Surface Fluxes (continued) Low ground heat flux results in overly warm skin temperature, which, coupled with high RNET, serves to exaggerate surface sensible heat flux.

Validation of Surface Fluxes (continued) Surface evapotranspiration (latent heat flux, FXLH) also appears to be slightly high owing to excess net radiation at the surface, among other factors. Remember: this is a single point validation example.

Areas Needing Improvement  Reduce remaining Eta surface insolation bias  Revise ground heat flux physics – too small (large) over dry (moist) soils  Add frozen soil and patchy snow physics – current 2 m cool bias over shallow snow (assumes complete coverage)  Higher resolution vegetation and soil classes  Refine infiltration and runoff formulations – prevent long-term drift of soil water in EDAS  Expand validation effort

Major Initiative: LDAS  A new Land Data Assimilation System (LDAS) for the Eta model  Goal: provide soil moisture/temperature initial conditions superior to current EDAS  Method: drive land-surface model (NOAH model) “off-line” with gage/radar precipitation and satellite-derived solar radiation  Additions: assimilate satellite-derived soil moisture and skin temperature. Use of much higher resolution soils and vegetation classes for improved surface fluxes

Improvement in soil moisture from using observed precipitation

LDAS High-Resolution Vegetation

Conclusions  New initiatives in improvement of physical parameterizations  An ongoing process  External comments and verification studies VERY helpful  Model biases: change with each upgrade to physics!!!