GEWEX Global Land-Atmosphere System Study (GLASS):

Updates and WGNE Connections

Joseph A. Santanello, Jr. NASA-GSFC

Co-chair: Martin Best (UK Met Office)

WGNE 27th Meeting 20 October 2011

GLASS Status and Activities - 2011 •  The GLASS Mission Statement is formulated as follows: Support improved estimates

and representation of land states and fluxes in models, the interaction with the overlying atmosphere, and maximize the utilized fraction of inherent predictability.

•  For the next 3 years our contribution will be based on the inertia of developments that comprise the present structure (Model Data Fusion – Benchmarking – Coupling), introduced in 2009.

Ongoing Projects (POCs):

-GLACE2 (Bart vd Hurk)

-LoCo (Joe Santanello)

-PILDAS (Rolf Reichle)

-GSWP-3 (Hyungjun Kim)

-PALS (Gab Abramowitz)

-ALMIP2 (Aaron Boone)

-GLASS-GHP Links (Mike Ek)

•  LAC: Addressing land-atmosphere interaction/coupling/feedbacks in models

•  MDF: Incorporates data assimilation and parameter estimation/calibration studies

•  Benchmarking: Standardized way to evaluate models and their ‘goodness’

•  Metrics: Diagnostics and quantification at the heart of each component

GLASS Role in GEWEX Imperatives 3.  Processes:  Develop  diagnos.c  approaches  to  improve  process-­‐level  understanding  of  energy      and  water  cycles  in  support  of  improved  land  and  atmosphere  models.  

► Extend diagnostics of stand-alone modules or model components and establish metrics that quantify the strength of interactions and feedbacks in the coupled land-atmosphere system

GLASS Role/Actions (2013-18)

•  Identify feedbacks and interactions among different processes, and build confidence in their replication in models (GLACE2, LoCo).

•  Spin-up activities in advanced diagnostics through a joint pan-GEWEX effort/workshop (GRP, GLASS, GHP, and others).

•  Develop metrics to aid benchmarking activities for both un-coupled and coupled modeling activities.

•  With the current and expected increasing complexity of land models in terms of various hydrologic and vegetation treatments, model optimization (i.e., parameter estimation approaches) will continue to be relevant to GLASS efforts (through Model Data Fusion).

•  Investigate alternative representations of sub-grid processes in land surface schemes (heterogeneity).

•  Develop improved understanding of climate variability and change on land surface properties, including soils, vegetation and hydrological processes, and an associated modeling capability (GSWP3).

•  Investigate the scope for development of next generation land surface models with improved representation of subsurface hydrology, including groundwater processes; identify suitable areas for their evaluation.

GLASS Role in GEWEX Imperatives 4.    Modeling:  Improve  global  and  regional  simula.ons  and  predic.ons  of  precipita.on,  clouds,  and  land  hydrology,  and  thus  the  en.re  climate  system,  through  accelerated  development  of  models  of  the  land  and  atmosphere.   ► Target model development with a goal of improved weather and climate prediction on the global and regional scales, focusing on the core components of the land and its coupling to the atmosphere while acting as a facilitating group who supports model development and helps the community to progress.

GLASS Role/Actions (2013-18)

• Coordinate the construction of a global land reanalysis system, building on ongoing and preparatory activities in Landflux, GSWP3, GLDAS and operational weather centers.

• Develop a framework and infrastructure for evaluation of land-atmosphere feedbacks. This should include the development of more quantitative estimates of uncertainty in the land condition and how this uncertainty propagates through to the atmosphere (e.g., PBL, convection, water and energy). This objective will be advanced in conjunction with the Processes Imperative in developing diagnostics.

• Organize coordinated intercomparison experiments for a range of model components in state of the art land models, especially with regard to: groundwater hydrology; surface water treatment (snow, river routing, lakes, irrigation, and dynamic wetlands); vegetation phonology and links between carbon and water; and Land Data Assimilation systems (follow-up the PILDAS initiative).

• Evaluation of these land model components will also have to be considered in their interactive (coupled) context with the PBL, while taking into account and developing more quantitative measures of uncertainty in the land parameters and states will enable more robust evaluation of data assimilation systems.

Review of WGNE 2010

•  Data assimilation is being discussed a lot more now in WGNE, and also the idea to combine the land and atmospheric data assimilation into one suite is discussed.

–  This is at the heart of PILDAS, which should provide an assessment of offline LDAS in current system, and ultimately the impact of atmospheric feedback on DA approaches.

•  Previously, there have not been very tight links between GLASS and WGNE.

–  However - the land surface is an emerging topic in WGNE, and certainly via this data assimilation subject we should be able to have more mutual feedbacks as well.

Project  for  the  Intercomparison  of  Land  Data  Assimila.on  Systems  (PILDAS)  

PILDAS-­‐1  Experiment  Plan  

Rolf  Reichle*  (NASA/GSFC)    and    

Jean-­‐François  Mahfouf  (Météo-­‐France)  

     *Email:    Rolf.Reichle@nasa.gov            Phone:    +1-­‐301-­‐614-­‐5693  

•  Enable  beQer  communica.on  among  developers  of  land  data  assimilaSon  systems  (LDAS).  

•  Develop  and  test  a  framework  for  LDAS  comparison  and  evaluaSon.  

•  Compare  land  assimilaSon  methods.  

•  Conduct  sensiSvity  studies  of  assimila.on  input  parameters  (such  as  model  and  observaSon  errors).    

•  Provide  guidance  and  priori.es  for  future  land  assimilaSon  research  and  applicaSons.  

•  UlSmately,  produce  enhanced  global  data  sets  of  land  surface  fields.    

The  first  experiment  (PILDAS-­‐1)  will  focus  on    • systems  targeted  for  weather  and  seasonal  forecasSng  at  operaSonal  centers  and  research  insStuSons  

•   soil  moisture  assimilaSon  •   development  of  a  framework  for  LDAS  comparison.  

PILDAS-­‐1  will  use    •   various  assimilaSon  approaches  (EnKF,  EKF,  …)  •   “off-­‐line”  land  model  (not  coupled  to  atmosphere)  •   syntheSc  observaSons.    

PILDAS-­‐1  provides  a  first  assessment  of  the  capability  of  different  assimilaSon  systems  to  extract  the  useful  informaSon  from  current  and  future  soil  moisture  missions  (SMOS,  SMAP).  

“Core  group”:  

•  Disseminates  LDAS  input  data  (forcing,  syntheSc  obs).  

•  Collects  and  post-­‐processes  LDAS  output.  •  Coordinates  analysis  of  results  and  publicaSons.  

“Par.cipants”:  

•  Generate  syntheSc  truth  data  and  LDAS  output.  •  Contribute  to  analysis  of  results.  

TentaSve  experiment  setup  (details  TBD!):  

•  Domain:    Red-­‐Arkansas  river  basin  

•  Exchange  grid:  0.25  deg  lat/lon  •  DuraSon:    2002-­‐2008  

Forcing  data  will  be  provided  and  LDAS  output  is  expected  on  the  exchange  grid.      

ParScipaSng  systems  may  run  on  their  na.ve  grids.  ParScipaSng  systems  use  na.ve  model  parameters  (land  cover,  soil  texture,  …).  

“TRUTH”  FORCING  

LSM_i  

SFSM_i  RZSM_i  FLUX_i  (truth)  

Phase  A  

SFSM_i  (obs)  

NOISE  

Phase  B  

LDAS_k  

“LDAS”  FORCING  

SFSM_k,i  RZSM_k,i  FLUX_k,i  (assim)  

Phase  C  

compare  

Skill_k,i  Phase  D  

Phase  A:    Generate  truth  for  i=1:NT  land  models  (parScipants).  

Phase  B:    Generate  i=1:NT  sets  of  syntheSc  observaSons  (core).  

Phase  C:    Generate  NA    open  loop  and  NA  ·∙NT  assim.  runs  (parScipants).  Phase  D:    Analyze  results  (all).  

Phase  C:  

•  ParScipants  must  assimilate  all  NT  sets  of  syntheSc  observaSons  at  least  once  into  their  default  LDAS.    

•  ParScipants  may  addiSonally  use  LDAS  variants  (different  model,  different  assimilaSon  method,  different  assimilaSon  parameters,…).  

•  ParScipants  choose  assimilaSon  algorithm  and  assimilaSon  parameters  

•  LDAS  output  must  include  assimilaSon  diagnosScs  (O-­‐F,  increments,  error  parameters,  …)  

Phase  D:  

•  Core  group  computes  skill  metrics,  including    

•  “Normalized  InformaSon  ContribuSon”  (Kumar  et  al.  2009)  •  “VerScal  Coupling  Strength”  (Kumar  et  al.  2009)  

•  AssimilaSon  diagnosScs  (O-­‐F  mean,  O-­‐F  variance  etc.)  

Sep  2011:    Disseminate  experiment  plan  to  potenSal  parScipants.  

Oct  2011:  Refine  experiment  plan  (GLASS  panel  meeSng)  

Dec  2011:  Finalize  domain  and  exchange  grid.    Prepare  forcing  data.  Jan  2012:  Conduct  dry-­‐run  of  enSre  experiment  with  2  insStuSons.  

Mar  2012:    Phase  A  –  Truth  integra.ons  (par.cipants).  

Jun  2012:    Phase  B  –  Genera.on  of  synthe.c  observa.ons  (core).  

Aug  2012:    Phase  C  –  Data  assimila.on  experiments  (par.cipants).  

Oct  20112    Phase  D  –  Analysis  of  experiments  (all).  Dec  2012:  Drao  publicaSons.  

Which  of  these  groups  have  an  assimila?on  system  ready?    

Are  we  missing  anyone?    

Institution POC Email NASA/GMAO Rolf Reichle Rolf.Reichle@nasa.gov Meteo-France Jean-Francois Mahfouf jean-francois.mahfouf@meteo.fr ECMWF Patricia de Rosnay, Gianpaolo

Balsamo Gianpaolo.Balsamo@ecmwf.int, patricia.rosnay@ecmwf.int

Environment Canada Stephane Belair stephane.belair@ec.gc.ca UK Met Office Martin Best, Sam Pullen martin.best@metoffice.gov.uk KNMI Bart van den Hurk hurkvd@knmi.nl University of Colorado Andrew Slater aslater@cires.colorado.edu Princeton University Eric Wood efwood@princeton.edu Ghent University Valentijn Pauwels, Niko Verhoest vpauwels@taoren.ugent.be, Niko.Verhoest@ugent.be

Observatoire de Paris Carlos Jimenez carlos.jimenez@obspm.fr University of Tokyo Toshio Koike tkoike@hydra.t.u-tokyo.ac.jp Monash University Jeff Walker jeff.walker@monash.edu NASA/HSB Sujay Kumar, Christa Peters-Lidard sujay.v.kumar@nasa.gov,

christa.d.peters-lidard@nasa.gov Argentina Homero Lozza hlozza@ina.gov.ar MIT Dara Entekhabi, Dennis McLaughlin darae@mit.edu, dennism@mit.edu

UCLA Steve Margulis margulis@seas.ucla.edu NCEP Mike Ek Michael.Ek@noaa.gov NESDIS Xiwu Zhan Xiwu.Zhan@noaa.gov Colorado State University Andrew Jones Jones@cira.colostate.edu USDA Wade Crow Wade.Crow@ARS.USDA.GOV CESBIO/SMOS Yann Kerr yann.kerr@cesbio.cnes.fr US Air Force Weather Agency John Eylander

University of Tokyo Taikan Oki taikan@iis.u-tokyo.ac.jp NILU William Lahoz William.A.Lahoz@nilu.no SMHI Magnus Lindskog magnus.lindskog@smhi.se ??? ???

Follow-­‐on  experiments  (PILDAS-­‐x)  will  focus  on:  

• AssimilaSng  satellite  observaSons  (non-­‐syntheSc)  • Soil  Moisture  (AMSR,  SMOS,  SMAP)  • Land  Surface  Temperature  (MODIS)  • SNOW  (MODIS,  AMSR)  

•   Uncoupled  vs.  Coupled  assimilaSon    •   Current  systems  vary  

Global Soil Wetness Project – Phase 3 •  A follow-up project to the Global Soil Wetness Project 2 (1986-95) is

currently being planned. The new components being considered for this project are:

–  Provide a comprehensive set of land surface states for the period including entire 20th century and recent years (~1901 – recent) that can serve as a long-term land surface reanalysis suite.

–  Include carbon models, to explore/attribute a possible carbon-related effect or changes in Hydro-Energy-Eco functioning.

–  Explore uncertainties of input datasets and their propagation through different schemes (LSMs) and super-ensembles (multi-input and multi-model).

–  Build a robust evaluation framework through component-wise verification (e.g. routing scheme for a validation of discharge (flux); GRACE for a validation of terrestrial water storage variation (storage)).

–  Includes engagement of the carbon community and inclusion of a suite of LSMs in varying hydrological and carbon treatments.

GSWP3 Design

Experiment   Timespan   T.  Res.   S.  Res.   Base  Sim.   Reference  Data  

EXP  1   1901  –  2008    (1987  –  2008)  

3  h   0.5°   20th  Century  Reanalysis  

GPCC,  CRU,  SRB  

EXP  2   2001  –  2100    (2080  –  2100)  

3  h   0.5°   CMIP5  with  2  scenarios  

GPCC,  CRU,  SRB  

EXP  3   1998  –  present     3  h   0.5°   ERA  Interim  &  MERRA    

*MulTple  Prcp,  CRU,  SRB  

GSWP3 Design

Benchmarking

•  PALS –  The development of a beta version of the Protocol for the Analysis of Land Surface Models

(PALS, http://pals.unsw.edu. au) is underway. –  PALS is a web application for evaluating land surface models and the observed data sets used

to test them (e.g. FLUXNET, CEOP). –  The PALS website is designed to analyze in a standard way uploaded single site model

simulations with FLUXNET and other observations. –  A related activity is that of a joint GHP-GLASS project to demonstrate benchmarking

approaches using PALS, and establishing empirical benchmarks in PALS from which to evaluate a suite a models.

•  ALMIP2 (2nd AMMA Land MIP) –  Experiments on a much higher spatial resolution (5km) and covers a 4 year period. –  Focus on the hydrology and vegetation processes that dominate there, using high-res satellite –  The project will give recommendations on the parameterization of runoff scaling. –  As this project has regional hydrological aspects, it will likely serve as a collaborative project

between GLASS and GHP.

•  Coupled (L-A) Benchmarking –  Point of discussion for GLASS-WGNE?

Global Land-Atmosphere Coupling Experiment

- 2!

The 2nd phase of the Global Land-Atmosphere Coupling Experiment

Overall goal: Determine the degree to which realistic land surface (soil moisture) initialization contributes to forecast skill (rainfall, temperature) at 1-2 month leads

GLACE-2

•  10 GCMs

•  100 start dates (1 Apr 1986 … 15 Aug 1995)

•  10 members of 2 months runs

•  2 series –  series 1: initial soil from pseudo obs –  series 2: random initial soil

•  Main Diagnostics: –  difference in R2 relative to obs series 1 – series 2, averaged for

15-day lead periods

GRL-paper

Wet/dry quantiles

GLACE-2 Conclusions

•  Skill improvement in US better than in Europe (also potential predictability in US larger)

•  Skill in temperature and precipitation increases mainly in areas where: –  the precipitation forcing quality is high (high station density gives better

initial soil moisture data) –  soil moisture is relatively extreme –  where potential predictability is high.

•  Ongoing GLACE2 experiments from KNMI, ECMWF and ETH are being performed for 2000-2010 in order to check possible signals emerging from known strong droughts in this period.

•  Also yet to be carried out are possible studies involving hydrological forecast models fed by the GLACE2 GCMs (Eric Wood).

GLACE-2 Publications

•  Koster, R.D., S. P. P. Mahanama, T. J. Yamada, Gianpaolo Balsamo, A. A. Berg, M. Boisserie, P. A. Dirmeyer, F. J. Doblas-Reyes, G. Drewitt, C. T. Gordon, Z. Guo, J.-H. Jeong, W.-S. Lee, Z. Li, L. Luo, S. Malyshev, W. J. Merryfield, S. I. Seneviratne, T. Stanelle, B. J. J. M. van den Hurk, F. Vitart, and E. F. Wood (2011): The Second Phase of the Global Land-Atmosphere Coupling Experiment: Soil Moisture Contributions to Subseasonal Forecast Skill; J.Hydrometeorol., in press.

•  Hurk, B.J.J.M. van den, F. Doblas-Reyes, G. Balsamo, R.D. Koster, S.I. Seneviratne en H. Camargo Jr, Soil moisture effects on seasonal temperature and precipitation forecast scores in Europe; Clim. Dyn., 2010, doi:10.1007/s00382-010-0956-2.

•  Koster , R. D., S. Mahanama, T. Yamada, G. Balsamo, A.A. Berg, M. Boisserie, P. Dirmeyer, F. Doblas-Reyes, G. Drewitt, C.T. Gordon, Z. Guo, J.H. Jeong, D.M. Lawrence, W.-S. Lee, Z. Li, L. Luo, S. Maleyshev, W.J. Merryfield, S.I. Seneviratne, T. Stanelle, B.J.J.M. van den Hurk, F. Vitart and E.F. Wood (2010), Contribution of land surface initialization to subseasonal forecast skill: First results from a multi-model experiment, Geophys. Res. Lett., 37, L02402, doi:10.1029/2009GL041677.

Local Land-Atmosphere Coupling (LoCo)

Motivation:

•  Land-atmosphere interactions (L-A) play a critical role in supporting and modulating extreme dry and wet regimes, and must therefore be quantified and simulated correctly in coupled models.

Objectives:

•  Address deficiencies in NWP and climate models by developing diagnostics to quantify the strength and accuracy of the Local L-A Coupling (‘LoCo’) at the process-level.

•  Develop NASA’s Land Information System coupled to the WRF mesoscale model (LIS-WRF) as a testbed to diagnose the behavior and impact of land surface (LSM) and boundary layer (PBL) coupling during dry/wet extremes in the SGP.

Deliverables:

•  Diagnostics that can be applied to any model, scale, or observation (in-situ or satellite).

•  Assessment of coupled model components and their integration through the land-PBL ‘process-chain’ linking the soil to precipitation.

•  LIS-WRF system as a testbed for GEWEX-GLASS directed studies of LoCo and model intercomparisons.

GEWEX Global Land-Atmosphere System Study

GLACE

LoCo

Benchmarking

Land-­‐Atmosphere  Coupling  (LAC)

Model  Data  Fusion  (MDF)

Metrics

•  LAC: Combines the global (GLACE) and local (LoCo) studies •  MDF: Incorporates data assimilation and parameter estimation/calibration studies •  Benchmarking: Standardized way to evaluate models and their ‘goodness’ •  Metrics: Diagnostics and quantification at the heart of each component

New structure of GLASS (circa 2009)

Perspectives on Past Workshops

Sept. 2005: GLASS/GABLS workshop on local L-A coupling

Overarching Goals of LoCo

•  Are the results of PILPS, GSWP, or data assimilation experiments affected by the lack of L-A coupling?

•  Can we explain the physical mechanisms leading to the coupling strength differences found in GLACE or other coupled NWP/climate experiments?

•  Is there an observable diagnostic that quantifies the role of local land-atmosphere coupling?

Perspectives on Past Workshops

June 2008: GLASS-WATCH LoCo Workshop

Challenges of LoCo

•  Land-atmosphere coupling takes place at many different spatial and temporal scale and involves many physical processes simultaneously.

•  The multi-scale and multi-process phenomenon makes a proper definition of “local” land-atmosphere coupling not easy.

•  Defining the ‘Realm of LoCo’ is certainly useful for: –  identifying where L-A interaction has a significant impact on the local climate –  defining proper diagnostics expressing the strength of coupling –  designing model intercomparison experiments in order to evaluate the coupling

(GEWEX Newsletter; van den Hurk and Blyth, 2008)

Complexity of L-A Interactions

Ek, M. B., and A. A. M. Holtslag, 2004: Influence of Soil Moisture on Boundary Layer Cloud Development. J Hydrometeorol., 5, 86-99.

Perspectives on Past Workshops

•  Direct moistening/drying and heating/cooling of the PBL, and the feedback exerted by this PBL change on the surface fluxes.

•  Impact of the change of the PBL depth or thermodynamic state on the formation/disappearance of PBL clouds (shallow cumulus) induced by land surface fluxes.

•  Triggering and fueling of shallow or deep convection.

•  The accumulation of hydrological anomalies in the soil water or snow reservoir, and the subsequent impacts of these surface states on the surface energy balance.

‘Realm of LoCo’

The temporal and spatial scale of all land-surface related processes that have a direct influence on the state of the PBL

LoCo Diagnostic Approach 2008-11 GLASS Collaborations

- Diagnose the components of GLACE at the diurnal process level:

- Our focus: Evaluate the ‘links in the chain’ and their sensitivities to land and PBL perturbations:

ΔSM → ΔEFsm → ΔPBL → ΔENT → ΔEFpbl ► ΔP/Clouds

(a) (b) (c) (d)

SM: Soil Moisture EF: Evaporative Fraction PBL: Mixed-layer quantities ENT: Entrainment fluxes at PBL top P/Cloud: Moist processes

‘LoCo Process-Chain’ defined by non-linear series of

interactions and feedbacks

LoCo Diagnostic Connections

•  GLASS-LoCo Working Group -  Craig Ferguson (Princeton/UTokyo)

-  L-A coupling in global reanalysis (MERRA) and satellite products (AIRS, MODIS, AMSR-E)

-  Obbe Tuinenburg and Cor Jacobs (Wageningen) -  Global and regional reanalysis (ERA, MERRA) and Indian monsoon

-  Paul Dirmeyer (COLA) -  Global climate models (IFS)

-  Kirsten Findell (GFDL) -  Regional reanalysis (NARR) and CTP-HIlow extension

-  Michael Ek (NCEP) -  EF/SM analytic development with in-situ data

-  Chiel van Heerwaarden (MPI) w/Bart vd Hurk (Wageningen) -  Single-column and LES studies

→ WCRP OSC poster cluster in Denver, CO (October), GEWEX-NEWS review (December), and U.S. SGP targeted/synthesis study (TBD)

L-A Coupling in NU-WRF

Coupled  LIS-­‐WRF    1-­‐km  resoluSon    NARR  init/bdy  condiSons    43  verScal  levels  (~42m  sfc)    3  PBL  +  3  LSM  schemes  

Land  Informa.on  System  (LIS)    Suite  of  LSMs  w/flexible  

resoluSon,  forcing,  parameters    Provides  spinup  capability  for  

improved  iniSalizaSon  of  land  surface  states  

  Plug-­‐in  design  supports  model  calibraSon  and    data  assimil.  

  LIS  VerificaSon  Toolkit  (LVT)  for  intra-­‐  and  inter-­‐model  evaluaSon  

PBLs  in  WRF  YSU  (Yonsei  University)    MYJ  (Mellor-­‐Yamada)  •  Counter-­‐gradient  fluxes          •  Level  2.5  closure  

•  Explicit  entrainment                      •  TKE  MRF  

•  Based  on  YSU  scheme  •  Implicit  verScal  diffusion  

LSMs  in  LIS  Noah  (NCEP)          CLM  (Community  Land  Model)  •  4  soil  layers                    •  10  soil  layers  (2  cm  upper)  

•  NCEP  operaSonal    •  Extensive  canopy/veg  H-­‐TESSEL  (ECMWF)  

•  4  soil  layers  •  Tiled  soil,  canopy,  snow  surfaces  

Impact of LIS Spinup on WRF Initialization

LIS Spinup – Noah LSM

Vol %

WRF Initial SWC

Vol %

Summary of IHOP-2002 Study

-  Soil moisture differences lead to significantly different signatures of heat and moisture evolution.

- The sensitivity of the L-A coupling is thus reflected in the balance between PBL and surface fluxes.

Dry Soils

Entrainment Fluxes

7am

Sfc Fluxes

7pm

7am

- - - Observations

7pm

Model Range

Wet Soils

Dry Soils

Wet Dry

Soil Moisture (m3/m3)

Fig. 2: Daytime evolution of specific humidity vs. potential temperature for the dry and wet soil moisture locations in Fig. 1

Fig. 1: Near-surface soil moisture map of the Southern Great Plains as simulated by LIS-WRF.

Vector length = Flux Vector slope = Bowen Ratio

Santanello, J. A., C. Peters-Lidard, and S. Kumar, C. Alonge, and W.-K. Tao, 2009: A modeling and observational framework for diagnosing local land-atmosphere coupling on diurnal time scales. J. Hydrometeor., 10, 577-599.

Dry  Soil  Site  IHOP02  –  Noah  +  3  PBLs

Theta-­‐e  (θe)  and  Rela.ve  Humidity  θe  –  Measure  of  Moist  StaSc  Energy  and  convecSve  potenSal  (LFC)  

-­‐  Entrainment  dominant  in  drying  out  the  PBL  

•   =  YSU  •   =  MYJ  •   =  MRF

Ah

βsfc

βent

Ale

320 K

330 K 340 K

350 K 360 K 370 K

25%

100% 75%

50%

7am

7pm

Wet  Soil  Site  IHOP02  –  Noah  +  3  PBLs

Theta-­‐e  and  Rela.ve  Humidity  -­‐  Surface  fluxes  and  limited  PBL  growth  support  buildup  of  MSE  and  near  saturated  RH  

•   =  YSU  •   =  MYJ  •   =  MRF

Ah βsfc

βent

Ale

320 K 330 K

340 K 350 K 360 K 370 K 25%

100% 75%

50%

7am

7pm

2006 & 2007 Soil Moisture Initialization

ARM-SGP Domain

%vol

 LIS-WRF w/ 9 LSM-PBL combinations were run for case studies over the SGP region:

a) 14-20 July 2006

b) 14-20 June 2007

 Each LSM was spun-up offline for 4 years prior to LIS-WRF initialization w/NLDAS forcing.

 Evaluation performed using LIS’ Land surface Verification Toolkit (LVT)

%vol

  Initial SM for LIS-WRF simulations

 4 year spinup of the Noah LSM

2006 Land Surface Energy Balance

Dry Regime (2006)

  All runs underestimate the Evaporative Fraction (EF)

  TESSEL performs best overall and is unbiased

  CLM performs worst for all fluxes

  Coupling tends to improve the energy balance in Noah and TESS

  Sensitivity to LSM >> Sensitivity to PBL scheme

2007 Land Surface Energy Balance

Wet Regime (2007)

  All runs overestimate EF due to very high net radiation vs. obs

  TESSEL performs worst overall, especially for Qle

  Coupling produces larger RMSE/Bias is all runs, and often reverses the sign

  Noah exhibits very low sensitivity to soil type or parameters (e.g. Czil)

2006 Mean Diurnal Cycles

TESSEL + YSU Domain Avg. Confirms RMSE/Bias stats

2006 Noah Spinup Domain Avg. Rnet low; Qle suffers

Noah + YSU Domain Avg. EF too low; LSM too dry

2007 Mean Diurnal Cycles

Noah + YSU Domain Avg. Rnet very high; all fluxes suffer

CLM + YSU Domain Avg. Rnet very high, but Qle is very good and Qg suffers

2007 Noah Spinup Domain Avg. Rnet low; highlights NLDAS vs. WRF difference

2006 & 2007 Sites of Interest

TESSEL + MYJ Site E4 Best Rnet, Qle, Qh, but Qg worst; best LoCo!

Noah + YSU Site E4 Bowen ratio high; see LoCo plots at this site

TESSEL + YSU Site E13 Rnet too high and manifested in Qle

Noah + YSU Site E13 Rnet high and spread out thru all 3 fluxes; best LoCo!

2006 Mixing Diagrams

Noah – Site E4 TESSEL – Site E4 CLM – Site E4 YSU MYJ MRF OBS

YSU MYJ MRF OBS

YSU MYJ MRF OBS

q*sat βsfc

βent

ΑH ΑLE

  7-­‐day  Composites  at  Site  E4:  

Noah+ YSU  

Noah+ MYJ  

Noah+ MRF  

TESS+ YSU  

TESS+ MYJ  

TESS+ MRF  

CLM+ YSU  

CLM+ MYJ  

CLM+ MRF  

RMSE T2  

7676.35   4010.32   7541.49   5374.09   2260.11   5095.88   3328.06   4118.45   4494.36  

Q2   4286.08   4955.41   3690.39   4033.14   2141.18   3467.70   4821.05   4238.01   4705.49  BIAS

T2  -­‐7573.25   -­‐3809.71   -­‐7386.57   -­‐4993.44   -­‐2137.12   -­‐4763.65   -­‐3239.87   -­‐4075.62   -­‐4432.84  

Q2   3679.64   4909.45   3108.82   3611.82   2076.45   3082.27   4777.64   3898.18   4628.91  Total  

Energy  -­‐1946.81   549.87   -­‐2138.88   -­‐690.81   -­‐30.33   -­‐840.69   768.88   -­‐88.72   98.04  

Statistics based on evolution of T2m and Q2m vs. observations, derived from the mixing diagrams above.

2007 Mixing Diagrams

Noah – Site E13 TESSEL – Site E13 CLM – Site E13 YSU MYJ MRF OBS

YSU MYJ MRF OBS

YSU MYJ MRF OBS

q*sat

βsfc

βent

ΑH ΑLE

  7-­‐day  Composites  at  Site  E13:  

Noah+ YSU  

Noah+ MYJ  

Noah+ MRF  

TESS+ YSU  

TESS+ MYJ  

TESS+ MRF  

CLM+ YSU  

CLM+ MYJ  

CLM+ MRF  

RMSE T2  

2002.19   1931.68   2191.51   3100.40   2479.13   3178.33   4700.82   2087.73   5499.77  

Q2   2174.83   3907.56   2197.26   1305.83   1565.65   1350.86   3384.12   3029.64   3246.01  BIAS

T2  -­‐1948.28   1311.79   -­‐2130.07   -­‐2797.02   -­‐1936.05   -­‐2825.40   3994.71   1312.28   4750.65  

Q2   1655.55   3314.64   1705.91   640.82   277.45   786.45   2667.00   1710.91   2455.45  Total  

Energy   -­‐146.37   2313.21   -­‐212.08   -­‐1078.10   -­‐829.30   -­‐1019.47   3330.85   1511.60   3603.05  

Statistics based on evolution of T2m and Q2m vs. observations, derived from the mixing diagrams above.

PBL Heat and Moisture Budgets

Heat and moisture budgets (SFC, ENT, and TOTAL) from the LIS-WRF simulations vs. observed, derived from the mixing diagrams above.

EF vs. PBL Height

E4 and E13 Composites

14-20 July 2006

14-20 June 2007

YSU MYJ MRF ● = Noah ○ = TESS □ = CLM X = Obs

----- Diurnal Std Deviation

Evaporative Fraction vs. PBL Height for each simulation vs. observed, along with the diurnal standard deviation through the 7-day period

LCL Deficit - Timeseries

LCL  Deficit  =  P(pbl)  -­‐  P(lcl)  +  =  LCL  not  reached  -­‐    =  LCL  reached  

  Measure  of  how  close  the  PBL  gets  to  Clouds/Precip  

  Larger  posiSve  (+)  indicates  drying  regime  

  Integrated  measure  of  the  diurnal  land-­‐PBL  coupling  

 ◘  

LCL Deficit time series from each LIS-WRF run at the E4 and E13 sites.

 ◘  

 ◘  =  Obs  

LCL Deficit - Spatial

LCL Deficit calculated spatially at 21Z on 19 June 2007 for the Noah-YSU and TESSEL-MRF simulations.

E13  

LCL  Deficit  (mb)  

E13  

LoCo: Key Findings

•  Significant errors exist in land surface energy balance simulations that depend on LSM and dry/wet regime.

•  Noah exhibits large insensitivity to soil and flux parameters during wet conditions.

•  LoCo Diagnostics can be used to evaluate LSM and PBL scheme behavior simultaneously in the context of their diurnal co-evolution.

• The sensitivity of L-A coupling is stronger towards the land during dry conditions.

• MYJ produces best total energy and PBL budgets in dry regime for all LSMs.

• CLM overestimates and TESSEL underestimates total energy in both regimes.

•  PBL height is largely insensitive to surface fluxes during wet regime.

LoCo Conclusions

•  The governing processes and feedbacks involved are explicitly integrated by the diurnal evolution of heat and moisture in the PBL, and therefore can be quantified using integrative diagnostic approaches.

•  Mixing diagram diagnostics offer a robust methodology that requires little input and can be applied equally to observations and any modeling system.

Santanello, J. A., C. Peters-Lidard, S. Kumar, C. Alonge, and W.-K. Tao, 2009: A modeling and observational framework for diagnosing local land-atmosphere coupling on diurnal time scales. J. Hydrometeor., 10, 577-599.

Santanello, J. A., C. Peters-Lidard, and S. Kumar, 2011: Diagnosing the Sensitivity of Local Land-Atmosphere Coupling via the Soil Moisture-Boundary Layer Interaction. J. Hydromet., in press.

Santanello, J. A., C. Peters-Lidard, and S. Kumar, C. Alonge, and W.-K. Tao, 2012: Diagnosing the Nature of Land-Atmosphere Coupling During the 2006-7 Dry/Wet Extremes in the U. S. Southern Great Plains. J. Hydrometeor., in prep.

•  This type of quantitative yet practical approach is a crucial first step towards a broader investigation of L-A coupling that will have implications for modeling and applications across a range of scales.

GLASS Relevance •  “Committee on Assessment of Intraseasonal to Interannual Climate Prediction and

Predictability“ (NAS BASC) reported last year:

"The realistic initialization of soil moisture in models can increase the accuracy of precipitation and temperature predictions at intraseasonal timescales…. To maximize the impact of land feedbacks on prediction quality, the mechanisms underlying the land-atmosphere coupling (e.g., evaporation, boundary layer dynamics, convection) need to be better understood and better represented in forecast systems."

Recommendations in that report include: • Systematic errors in dynamical models should be identified. • The representation of physical processes should be improved to reduce errors in dynamical models. • Many sources of predictability remain to be fully exploited by ISI forecast systems.

Two bullets from recent GLASS research:

•  GLACE-2 has shown that intraseasonal-seasonal prediction skill of temperature and precipitation over the U.S. can be significantly improved by realistic initialization of the land surface (soil moisture) in forecast models.

•  LoCo is significantly improving our understanding of the mechanisms by which soil moisture anomalies impact weather prediction via surface fluxes and atmospheric boundary layer processes, which can inform design of a real-time monitoring network to initialize those forecasts.

GLASS-WGNE: Past

•  “Land topics” don’t seem to be at the heart of WGNE

•  Added value of strong links not entirely clear

•  PILPS was originally co-sponsored by WGNE (?) –  What has been mutually gained –  Do we need another large intercomparison project?

GLASS-WGNE: Present

•  Offline projects are producing improved global estimates of land surface states (GSWP) and techniques (PILDAS).

•  Coupled projects show how the land impacts prediction (GLACE), and is dependent on coupling that can be diagnosed at the process-level (LoCo).

•  Benchmarking both offline and coupled systems is essential for translating understanding to model development.

•  GLASS panel meeting: Updates on all projects

•  WCRP OSC: Project leads and posters

GLASS-WGNE: Future?

•  Mutual Interests –  Involvement in projects like PILDAS, GSWP3, LoCo –  Coupled L-A benchmarking

•  Discussion…..