Introduction to the NOAH Land Surface Model N

Introduction to the NOAH Land Surface Model N: O: A: H: National Centers for Environmental Prediction (NCEP) Oregon State University (Dept of Atmospheric Sciences) Air Force (both AFWA and AFRL - formerly AFGL, PL) Hydrologic Research Lab - NWS (now Office of Hydrologic Dev – OHD) Outline 1. 2. 3. What is an LSM? What is the NOAH LSM? NOAH @ UT Zong-Liang Yang Prepared for Surface Water Hydrology February 24, 2005

Outline 1. What is an LSM? (29 slides) 2. What is the NOAH LSM? (13 slides) 3. NOAH @ UT (14 slides)

Computer Code to Model Land Surface Processes • Exchanges with the atmosphere – – Momentum Energy Vapor Trace gases/dust/aerosols • Exchanges with the oceans – Runoff – Sediments / nitrogen • Land-memory processes – – Topography Snow cover Soil moisture Vegetation • Human activities – Land use (agriculture, deforestation, urbanization…) – Inland water pollution

Earth’s Radiation Budget Shortwave: Earth reflects 30% directly back to space, absorbs about 20% in the atmosphere, and absorbs about 50% at the surface. Longwave: Greenhouse effects keep the surface warm. Turbulent heat fluxes: sensible and latent heat keeps the surface cool.

Land-atmosphere coupling strength (JJA) Koster et al. , 2004 (Science) Hot spots: initial soil moisture impacts rainfall forecasts Locations for the routine monitoring of soil moisture

Interpretations Rainfall in a region depends on atmospheric moisture transport from upwind areas and local evapotranspiration. ü Moisture transport depends on large-scale circulations ü Evapotranspiration depends on vegetation, soil, and weather conditions L Qa In Qa Out Sa E Pm Pa Ql

Do Land Surface Processes Matter to Climate Prediction?

What Are Land Surface Processes • Land surface consists of – urban areas, soil, vegetation, snow, topography, inland water (lake, river) … • Land surface processes describe – exchanges of momentum, energy, water vapor, and other trace gases between land surface and the overlying atmosphere – states of land surface (e. g. , soil moisture, soil temperature, canopy temperature, snow water equivalent) – characteristics of land surface (e. g. , roughness, albedo, emissivity, soil texture, vegetation type, cover extent, leaf area index, and seasonality)

What Are Land Surface Processes • Land surface processes function as – lower boundary condition in Atmospheric Models • Atmospheric Boundary Layer Simulation • Numerical Weather Prediction Climate Simulation 4 -D Data Assimilation – upper boundary condition in Hydrological Models • Water Resources Estimation; Crop Water Use; Runoff Simulation – interface for coupled Atmospheric/Hydrological/Ecological Models

Land Surface Models (LSMs) • Computer code describing land surface processes (also called LSSs, LSPs, SVATs) – FORTRAN, C, C++, PASCAL, …. . . – Tens to thousands of lines • There a huge number of LSMs (100+ examples in literature) – many are just “research models’’, local-scale oriented, with specific process emphasis – up to ~100 canopy, ~100 soil, ~100 snow, even ~100 atmosphere layers! • LSMs in GCMs and Hydrological Models are less diverse – one dimensional, with 1 -2 canopy, 1 -10 soil, 1 -10 snow layers – three general classes • “Bucket” Models (no vegetation canopy) • “Micrometeorological” Models (detailed soil/snow/canopy processes) • “Intermediate” Models (some soil/snow/canopy features)

Land Surface Models (LSMs) • Four basic requirements – frequently-sampled (hourly or sub-hourly) weather “data” to “drive” LSMs • • • precipitation (rate; coverage, large-scale/convective in GCMs) radiation (shortwave, longwave) temperature wind components (u, v) specific humidity surface pressure – initialization of state variables • soil moisture (liquid, frozen) deep soil temperature – specification of surface characteristics • albedo roughness vegetation cover – validation of simulations of state variables and fluxes • soil moisture sensible/latent heat fluxes

Land Surface Models (LSMs) • Minimum Requirement is to describe: – bulk momentum exchange (roughness length, zero displacement height) – exchange of radiation for: • solar radiation (0 -3 micrometers): albedo • long-wave (3 -100 micrometers): surface emissivity – partition of radiant energy between soil heat, latent heat and sensible heat, along with its relationship to water availability • Micrometeorological Models typically – have 20 -50+ parameters – require to prescribe or predefine them before running the models • Research Issues – Obtaining and applying relevant “pure biome” data to test or calibrate LSMs – Dealing with spatial/temporal heterogeneity • defining area-average parameters; scaling up surface parameters from local to regional • defining space-time structure of atmospheric inputs – Making best use of remote sensing data for initialization, specification and validation – Improving key processes • Snow/Frozen soil • Runoff generation/routing • “Greening” of LSMs (carbon balance and vegetation dynamics)

Land Surface Models (LSMs) • Project for Intercomparison of Land-surface Parameterization Schemes (PILPS) – is helping define needs, understanding processes, from sensitivity tests and field data to coupled modeling – ongoing project (4 phases, starting from 1992, 20 -30 LSMs participating) • Phase 1: GCM-generated forcing data for three “biomes”: – models’ response time to initialization of soil moisture – sensitivity experiments with albedo, canopy/soil water storage, aerodynamic resistances • Phase 2 : observed data – 2(a) Cabauw, Netherlands (grass; 1 -year) – 2(b) HAPEX-MOBILHY, France (crop; 1 -year) – 2(c ) Red-Arkansas River Basin, USA (mixed land covers; 10 -year) – 2(d) Valdai, Russia (snow/frozen soil; 18 -year) – 2(e) Sweden Torne/Kalix Basins, 58000 sq. km, 218 grids, 1979 -98 • Phase 3: joint with Atmospheric Model Intercomparison Project -AMIP • Phase 4: selected LSMs coupled to one host model – Regional Model Global Model Coupling Strategy

Simple “Bucket” LSMs • “Bucket” Models” first in hydrological models, later in GCMs – – • In original “bare soil” form, e. g. , Manabe (1969) include no vegetation so specify (fixed) parameter values for – – – • “Bucket” is filled by precipitation, and emptied by evaporation Runoff occurs when “Bucket” is full (soil water level exceeds the field capacity) Evaporation is based on soil-water limited “Potential Evaporation” The limiting factor is a linear function of soil moisture albedo (Solar radiation) surface emissivity (Longwave radiation) roughness length (Momentum exchange) soil thermal properties (Soil heat storage) field capacity (15 cm) for all land points (Soil hydrology) snow areal coverage (covers 100% gridbox) (Snow energy/water budgets) Some variants of “Bucket” Models – allowing leakage by implicit transpiration or gravitational drainage – including surface resistance in evaporation formula – extending to 2 connected “Buckets”: Deardorff (1977) includes a thin surface layer and a deep bulk layer

Micrometeorological Model-Based LSMs • Micrometeorological Model-Based LSMs are – more complex than bucket models – less complex than “multi-layer” or “second order closure” canopy models • Best known examples: – “Biosphere-Atmosphere Transfer Scheme (BATS)” – “Simple Biosphere Model (Si. B)” – Si. B more complex (not at run time) some whole-canopy parameters from “preprocessing”, one calculation, prior to actual model run • Five components of radiation exchange considered in BATS and Si. B – – – Direct beam photosynthetically active radiation (PAR) (<0. 72 micrometers) Diffuse solar photosynthetically active radiation (PAR) (<0. 72 micrometers) Direct beam near infrared (NIR) (0. 72 -4. 0 micrometers) Diffuse solar near infrared (NIR) (0. 72 -4. 0 micrometers) Thermal infrared (>4. 0 micrometers)

Micrometeorological Model-Based LSMs Biosphere-Atmosphere Transfer Scheme BATS

Micrometeorological Model-Based LSMs • “Two-Stream Approximation” are used to calculate solar radiation exchange in Si. B for both PAR and NIR, diffuse and direct beam components from • Can obtain single-scattering albedo, diffuse fluxes within canopy (for stomatal resistances) and at soil surface, and total albedo of the vegetated surface Vegetation albedo depends on specifications of • • • – leaf area index scattering coefficients of leaves and soil – leaf angle distribution function angle of the direct-beam incident radiation – proportion of different components of the incident radiation Not suitable for application over heterogeneous or ‘clumpy’ canopies (coniferous forests); require three-dimensional radiation models BATS prescribes vegetation albedo; 4 -layer model for stomata

Micrometeorological Model-Based LSMs • Thermal Radiation from surface is computed from “Stefan-Boltzmann” law: • Whole-canopy turbulent transfer is important for exchange of momentum, heat and mass (water vapor, trace gases) – is a complex process – requires higher-order closure models, Markov-chain numerical simulations, or eddy simulation techniques – simpler treatments in LSMs because of • computational expense and storage requirements • aerodynamic resistance roughly an order of magnitude smaller than aerodynamic resistance – prescribed for BATS (and most other models); aerodynamic resistances from curve-fitting functions, iteration for nonneutral corrections – calculated in Si. B: pre-processed from canopy morphology using a “KTheory” equivalent of Shaw-Pereira (1982); aerodynamic resistances from canopy morphology using “K-Theory”, iteration for nonneutral corrections

Micrometeorological Model-Based LSMs • • • Canopy interactions described by a “big leaf” model in BATS and Si. B and others [Si. B allows an “understorey”; Si. B 2 allows only one layer] Dry canopy parameterized through surface resistance based on the model of Jarvis (1976). For a single leaf: Canopy surface resistance is obtained by integrating above equation over leaf angle and canopy depth. This integration and those dependencies are “coded” in different forms in BATS and Si. B and many other models.

Micrometeorological Model-Based LSMs • Different forms of coding for the integration: – In Si. B, an analytical integration is performed by using the Goudriaan formulation for some regular leaf angle distribution – In BATS, the integration is done by using a 4 -layer numerical scheme for canopy • Different forms of coding for the stress factors: – primarily determined from curves fits to data (e. g. Jarvis, 1976), but – in Si. B was a linear function of the leaf water potential, which, in turn, was determined by a catenary model of water flow from root zone to leaf – in BATS, was introduced only when the atmospheric demand exceeds the water supply from the soil through roots – in Si. B taken as a smooth curve varying from zero at some minimum temperature around freezing to unity at an optimum temperature between 25 degree C and 35 degree C, depending on species, and dropping sharply down to zero at 45 degree C to 55 degree C – in BATS parameterized as a quadratic function of canopy temperature, varying from 0 at freezing point to unity at 25 degree C, and remaining at unity when temperature is greater than 25 degree C

Micrometeorological Model-Based LSMs • • • Calibration of stomatal resistance largely speculative: guessed from single leaf data, but occasional “Calibration Data Sets” available e. g. forest, Shuttleworth (1989); Gash et al. (1996) Wet Canopy Description is a simplified “Rutter” Model: S, proportional to Leaf Area Index ~[(0. 1 -0. 25) x LAI] (mm) Both BATS and Si. B allow partially wet canopy

Micrometeorological Model-Based LSMs • • • “Plot-scale” calibration of interception models adequate, given tuning of S against LAI BATS/Si. B have simple “vertical hydrology”: soil water diffuses vertically through 2 -3 layers, but is ultimately inaccessible by draining from lowest layer Global Distributions of different land classes required, BATS has 15, Si. B has 12, with parameters defined for each class: Calibration of these Parameters Remains Speculative

“Intermediate” LSMs • • “Intermediate” LSMs trade process detail for simplicity, justified by still speculative need for more complete process models, and their poor calibration Often have distributed vegetation class, but simpler physics: e. g. – UKMO – French ISBA – Simple Si. B – “Super Bucket” • simple canopy model simple soil model Simplified empirical parameterization fitted to time consuming, Si. B-calculated functions basic bucket, with “reference crop” estimates instead of “potential” rate(+Several others) Permissible, practical alternative, pending evidence of need for more complex micrometeorological models

LSMs Calibration Issues • • Primary LSM short-coming is lack of reliable calibration: 3 needs – Obtaining and applying relevant “pure biome” data – Understanding how to define area-average parameters – Making best use of Remote Sensing Data Single Biome Calibration – Needs: • sets of continuous “driving variables” for whole years (frequentlysampled, near-surface, equivalent to GCM) • some substantial periods with measured exchanged fluxes (preferably for several seasons, all exchanges, time scales) – Some biomes done or in-hand • Rain forest (ARME, ABRACOS) • Prairie grass (FIFE) • Boreal forest (BOREAS) • Sonora Desert, Tucson, Arizona (Example of multi-objective calibration)

LSMs Calibration Issues: Area-average • Defining Area-average LSM parameters is active research which involves data collection, analysis and coupled models for mixed land covers, and for selected regions – mixed vegetation cover and/or soils uneven precipitation • sloping terrain Examples: – – – Hydrologic-Atmospheric Pilot Experiment (HAPEX) - S. W. France, 1986, (mixed temperate; agriculture, forest) First ISLSCP Field Experiment (FIFE) - Kansas, 1987, 9 (tall prairie grass) HAPEX-Sahel - Niger, 1992, (semi-arid Sahelian savannah-across rainfall gradient) BOREAS - Canada, 1993, 4, Boreal Forest AMAZONIA (LBA) - S. America, 1997, 8 - Amazonian land covers GEWEX

LSMs Calibration Issues: Atmospheric Data • • Calibration or model development depends on atmospheric input data Runoff estimation error is determined by the accuracy of the precipitation forcing as measured by the density of rain gauge

Micrometeorological Model-Based LSMs • New developments: The Greening of LSMs (improved canopy physiology) – enzyme kinetics-electron transport model (Farquhar et al. , 1980), relating leaf photosynthetic rate to PAR, leaf temperature, leaf internal carbon dioxide concentration, and leaf carboxylase concentration – photosynthesis modeled as the minimum of three potential capacities to fix carbon

Micrometeorological Model-Based LSMs • New developments: The Greening of LSMs (improved canopy physiology) – Ball-Berry model of stomatal conductance (Ball et al. , 1986): robust semi-empirical method, relating leaf stomatal conductance to photosynthetic rate and environmental parameters – Other forms: Leuning (1995), Foley et al. (1996), Dickinson et al. (1998): using vapor pressure deficit instead of relative humidity

Micrometeorological Model-Based LSMs • New developments: The Greening of LSMs (improved canopy physiology) – C 3 (all trees and many herbaceous) and C 4 (warm grasses) plants are similarly modeled as the minimum of three potential capacities to fix carbon, but – The three potential capacities are defined differently for C 3 and C 4 plants – The parameters are different for both types of plants – The Ball-Berry model of stomatal conductance replaces all the terms in the Jarvis model except for the soil water stress term

Micrometeorological Model-Based LSMs • New developments: The Greening of LSMs (improved vegetation phenology) – Traditionally, annual cycle in areal vegetation coverage and leaf area index prescribed as a quadratic function of deep soil temperature – Some recent attempts: relating vegetation to climate using empirical rules involving monthly or annual mean temperature and precipitation – New developments: • simple rule-based formulation to describe behavior of winter-deciduous and drought-deciduous plants (Foley et al. , 1996), using – daily average temperature, yearly carbon balance – competition between trees and grasses (through allocation/shading) • interactive canopy model (Dickinson et al. , 1998) – short time-scale leaf dynamics – consistent treatment of stomatal conductance and assimilation – assimilated carbon allocated into leaf, fine root and wood – soil carbon model » fast pool: leaf/root turnover (senescence, herbivory or mechanical), leaf loss (cold and drought stress), soil respiration » slow pool: inert for seasonal variations

Recent LSM Developments • Improved understanding in physical hydrology – Snow hydrology – Topographic controls on soil moisture/runoff • Enhanced linkage with biogeochemistry/ecology – Carbon and nitrogen – Vegetation growth/competition • Closed water cycle through river routing • Advanced techniques for using remotely-sensed land parameters • Extended field datasets and model calibration

Outline 1. What is an LSM? (29 slides) 2. What is the NOAH LSM? (13 slides) 3. NOAH @ UT (14 slides)

Land-Atmosphere Interaction

NOAH LSM Development Team NCEP/EMC: NCEP Environmental Modeling Center (EMC) (Mitchell, Chen, Ek, Lin, Marshall, Janjic, Manikin, Lohmann, Grunmann, Pan) OSU: Oregon State University (Mahrt, Pan, Ek, Kim, Rusher) HL: NWS Hydrology Lab - formerly Office of Hydrology (Schaake, Koren, Duan) AFWA: Air Force Weather Agency - formerly AFGWC (Moore, Mitchell, Gayno) AFRL: Air Force Research Lab - formerly AFGL and PL (Mitchell, Hahn, Chang, Yang)

NOAH LSM Evaluation Team E. H. Berbery and Rasmusson: U. Maryland (ARM/CART) C. Marshall and Crawford U. Oklahoma (OU Mesonet) I. Yucel and Shuttleworth: U. Arizona A. K. Betts: Atmospheric Res Inc (ARM/CART, AZNET) (ISLSCP/FIFE) C. D. Peters-Lidard, Wood Princeton U. L. Hinkelman and Ackerman: Penn State U. (TOPLATS extensions) (ARM/CART) T. H. Chen, W. Qu, Henderson-Sellers, et al. RMIT (PILPS-2 a) E. Wood, Lettenmaier, Liang, Lohmann: Princeton U. (PILPS-2 c) A. Schlosser, A. G. Slater, A. Robock, et al. U. Maryland (PILPS-2 d) R. Angevine NOAA/AL (Flatland Exp)

References 1. User’s Guide, Public Release Version 2. 2, Last Updated 25 June 2001, filename README_2. 2. doc at ftp: //ftp. ncep. noaa. gov/pub/gcp/ldas/noahlsm/ver_2. 2 2. Author / Point of Contact: Kenneth. Mitchell@noaa. gov (NCEP/EMC), phone 301 -763 -8000 3. PROGRAM HISTORY LOG 01 Mar 99: Ver_1. 0 08 Mar 99: Ver_1. 1 27 Jul 99: Ver_2. 0 23 Oct 00: Ver_2. 1 25 Mar 01: Ver_2. 2

Timeline of NOAH LSM Evolution

Atmospheric Forcing Variables Air temperature Air humidity Surface pressure Wind speed Surface downward longwave radiation Surface downward solar radiation Precipitation at height Z above ground

Initial Conditions 1. SMC: total volumetric soil moisture (liquid & frozen) in each soil layer 2. SH 2 O: liquid volumetric soil moisture in each soil layer 3. STC: temperature in each soil layer 4. T 1: skin temperature 5. CMC: canopy water content 6. SNOWH: snow depth 7. SNEQV: water-equivalent snow depth

Model Parameters There are five kinds of land-surface parameters, reviewed in order below. a) single universal values b) values dependent on the soil class index (default categories are 1– 9) c) values dependent on the vegetation class index (default categories are 1– 13) d) values dependent on the surface slope index (default categories are 1– 7) e) parameters specifying the numbers of vegetation, soil, and slope classes

Model Soil Parameters SMCMAX: maximum volumetric soil moisture (porosity) SMCREF: soil moisture threshold for onset of some transpiration stress SMCWLT: soil moisture wilting point at which transpiration ceases SMCDRY: top layer soil moisture threshold at which direct evaporation from soil ceases DKSAT: saturated soil hydraulic conductivity PSISAT: saturated soil matric potential B: the “b” parameter in hydraulic functions DWSAT: saturated soil water diffusivity QUARTZ: quartz content, used to compute soil thermal diffusivity FRZFACT: a parameter used with FRZK to compute the value of parameter FRZX

Model Vegetation Parameters Z 0 (m): RCMIN (s/m) : roughness length minimal stomatal resistance used in canopy resistance of routine CANRES RGL: radiation stress parameter used in F 1 term in canopy resistance of routine CANRES HS: coefficient used in vapor pressure deficit term F 2 in canopy resistance of routine CANRES LAI: presently set to universal value of 3. 0 across all vegetation classes Note: seasonality of vegetation greenness carried by fraction of green vegetation (SHDFAC) NROOT: number of soil layers from top down reached by roots SNUP: the water-equivalent snowdepth upper threshold at which 1) 100 percent snow cover is achieved for given veg class 2) maximum snow albedo is achieved for given veg class

Forecasts of Dew Point Temperature Old version New version

Forecasts of Dew Point Temperature

Outline 1. What is an LSM? (29 slides) 2. What is the NOAH LSM? (13 slides) 3. NOAH @ UT (14 slides)

NOAH @ UT • To put a dynamic vegetation component in the NOAH LSM – understand the feedback mechanisms between precipitation, soil moisture, and vegetation dynamics – To improve weather and climate forecasts on time scales from hourly to seasonal – To interpret and transfer weather and climate forecasts for water resources management applications • To put a biogenic VOC emission component in the NOAH LSM – understand the impacts of climate change and land use/land cover change on BVOC emissions and the formation of near ground ozone

The Modeling Domains: Multiple interactive nesting grids: Weather Research and Forecast (WRF) Model, version 2. 02 (http: //www. wrf-model. org/) The model is developed by NCAR, NCEP and Universities. 30 km grid; 31 Levels Temperature, wind, humidity, pressure predicted 3 -10 km grid; 31 Levels 90 km grid; 31 Levels Temperature, wind, humidity, pressure prescribed from large-scale analyses Temperature, wind, humidity, pressure predicted

Vegetation Distribution in the Model Other Land Data: Vegetation LAI Albedo Roughness length Stomatal parameters Soil Topography Validation: Weather stations Hydrologic stations Remote sensing Field data

Stomatal Conductance in NOAH • Transpiration parameterized through surface conductance based on the model of Jarvis (1976). For a single leaf:

NOAH with Modifications (1) The ability of roots to uptake soil water depends on the availability of soil water. In NOAH, this is a linear function, which may work if soil is near saturated. We make it a step-function, guided by observations. We call the experiment NOAH SWF (soil water factor).

Role of Land Surface Processes in Modulating NAMS Rainfall OBS WRF/SLAB WRF/NOAH SWF JUN JUL AUG WRF/SLAB fails to produce rainfall for June thru August. WRF/NOAH fails to capture rainfall in August. WRF/NOAH-SWF captures rainfall in August.

NOAH with Modifications (2) NOAH prescribes fractional green vegetation. In the real world, vegetation growth depends on precipitation, temperature, nutrients, and others. This can be modeled by 1) Relating stomatal conductance to photosynthesis & environmental conditions, and 2) Allocating assimilated carbon to leaves, stem, wood, and roots.

Leaf Anatomy Stomate (pl. stomata)

Carbon and Water • Plants eat CO 2 for a living • They open their stomata to let CO 2 in • Water gets out as an (unfortunate? ) consequence • For every CO 2 molecule fixed about 400 H 2 O molecules are lost

Photosynthesis and Conductance Stomatal conductance is linearly related to photosynthesis: (The “Ball-Berry-Collatz” parameterization) photosynthesis RH at leaf sfc stomatal conductance CO 2 at leaf sfc Photosynthesis is controlled by three limitations (The Farquahar-Berry model): Enzyme kinetics (“rubisco”) Light Starch

Photosynthesis and Carbon Allocation

Comparison of Observed & Simulated Rainfall OBS JUN JUL AUG WRF/NOAH SWF WRF/NOAH DV

Conclusions 1) Strong sensitivity to land surface processes 1) SLAB (without vegetation) fails to simulate the monsoon rainfall in all months 2) NOAH (with prescribed vegetation) is better, but underestimates the August rainfall 3) NOAH (with improved root water uptake and response) is even better 4) NOAH (with improved carbon and water coupling and leaf’s dynamic response to rainfall) gives the best overall simulation of the warm season rainfall 1) In the southern monsoon regions 2) In the southern Great Plains. 2) Dynamic vegetation has the largest impacts on the simulation of rainfall in the SGP, which is the area that shows the largest memory to soil moisture conditions.

PHYSICAL CLIMATOLOGY GEO 387 H TTh 9: 30 -11: 00 Fall 2005 www. geo. utexas. edu/courses/ 387 h/index. PHY. htm Taught by Dr. Liang Yang Department of Geological Sciences This course will cover the basics of a broad range of physical processes having an impact on precipitation and evapotranspiration at the earth's surface. Local processes include thermodynamics, cloud microphysics, net radiation at the surface, fluxes of heat and moisture into the soil and the atmospheric boundary layer, and the effects of a vegetation canopy. Large-scale interactions between atmosphere, ocean, soil moisture, snow, ice, biosphere, and lithosphere are reviewed in the context of potential local/large-scale transports. Methods for measuring and estimating precipitation and evapotranspiration are described, including the sophisticated models known as GCMs (General Circulation Models) and SVATS (Soil-Vegetation-Atmosphere Transfer Schemes). Some basic programming/modeling techniques will be covered. For both Geoscience and non-Geoscience graduate students.