Second Global Soil Wetness Project (GSWP-2) couples land surface models with a mircrowave emissions model toprovide a global 1-degree map of the synthetic L-band H-polarized brightness temperature corresponding to theincidence angle and equator crossing time of the HYDROS Satellite for 1 June 1992. See article on page 10.

What's New• CALL FOR PAPERS: 5th International Scientific

Conference on the Global Energy and Water Cycle(www.gewex.org/5thconf.htm)

NEWS NEWSVol. 14, No. 3 August 2004Global Energy and Water Cycle Experiment

World Climate Research Programme(A Programme of WMO, ICSU and IOC)

The new Land Information Sys-tem (LIS) Project is now pro-viding 1-km surface fluxes.Sample latent heat flux snap-shot from LIS/Noah simulationfor 11 June 2001. (a) 5-km simu-lation (contours) overlaid by a1-km simulation (shaded). (b) ¼-degree simulation. (c) 5-km simu-lation. (d) Full 1-km simulation.See article on page 8.

GSWP-2 FACILITATES USE OF SATELLITE DATA IN MODEL VALIDATIONBY PRODUCING BRIGHTNESS TEMPERATURE MAPS

NEW 1-KM RESOLUTION MAPPING OF SURFACE FLUXES HELPS TODRIVE CLOUD RESOLVING MODELS

• Workshop on Trends in Global Water Cycle Vari-ables organized to support IPCC Assessment(http://www.gewex.org/trendswkshp.htm)

August 20042

Contents PAGE

Commentary: Earth Observations: A 2Renewed Opportunity Area for GEWEX?

Recent News of Relevance to GEWEX 3

Ensemble Streamflow Forecasting in 4Snowmelt Dominated River Basins

ISLSCP Initiative II Update 6

Land Surface Modeling in BALTEX 7

Ultra-High Resolution Observation-Driven Land 8Modeling Needed to Enable the Developmentof Global Cloud Resolving Earth System Models

The Role of Fine-Scale Landscape and Soil 9Moisture Variability in the Initiation of DeepConvection

Update on the Second Global Soil Wetness 10Project (GSWP-2)

Soil Moisture—Muddy Prospects for a Clear 11Definition

GEWEX Relevant Publications of Interest 12

Workshop/Meeting Summaries:– 4th Study Conference on BALTEX 14

Meetings Calendar 15

COMMENTARY

EARTH OBSERVATIONS: A RENEWEDOPPORTUNITY AREA FOR GEWEX?

Rick Lawford, DirectorInternational GEWEX Project Office

Given the recent reorganization in a number ofspace agencies, the above title may seem poorlytimed. However, that is not necessarily the case.Space agencies are maintaining strong commitmentsto missions such as Cloudsat, the National Polar-Orbiting Operational Environmental Satellite System(NPOESS), the Soil Moisture and Ocean SalinityMission (SMOS), the Hydrosphere State Mission(HYDROS) and the Global Precipitation Mission(GPM) that will open new scientific frontiers forGEWEX researchers. In addition, activities relatedto the high level satellite committee of the WorldMeteorological Organization (WMO), the IntegratedObserving Strategy (IGOS) and the Global EarthObserving System of Systems (GEOSS) have re-sulted in new opportunities to influence the directionof future Earth observing activities. Although somemay question the benefits of these planning exer-cises, it is important for GEWEX and the WorldClimate Research Programme (WCRP) to continueto position themselves through these activities. Pos-sible benefits to GEWEX and the WCRP couldinclude strengthened support for our research pro-grams, greater stability for data streams upon whichwe rely, and a greater recognition for water andenergy cycle issues in the priorities of the agenciesthat fund our projects.

Within the IGOS Partnership (IGOS-P), spaceagencies and international programs consider thefull range of satellite and in situ observing systemsthat provide environmental data. Through the de-velopment of themes, IGOS-P is creating newopportunities for dialogue between the researchand applications communities. For example, withinthe IGOS Global Water Cycle Observations(IGWCO) theme, recommendations about support-ive missions can be brought to senior managers inspace agencies through the Committee on EarthObserving Satellites-Strategic Implementation Team(CEOS-SIT). These activities should supplementthe high level WMO satellite committee, which alsoprovides statements on WCRP satellite data needs.Granted the processes are new and still have grow-ing pains, but evidence of the effectiveness of thisexchange should soon be evident to the community.

The GEOSS initiative is a new multi-nationactivity that was initiated at the first Earth Observ-ing Summit in Washington in the summer of 2003.Since completing its framework document, theGEOSS activities are now focused on developinga 10-year implementation plan for observationalsystems. Prof. Toshio Koike, the chief scientistfor the Coordinated Enhanced Observing Period(CEOP), is participating on the four-member taskteam that is preparing the implementation plan. Hisrole is providing the GEWEX and WCRP commu-nities with an opportunitiy to make significant inputsto these planning documents.

In the past, GEWEX has had a leading role indeveloping data products. Although GEWEX re-mains a leader in developing satellite products,there is now a much larger community using sat-ellite data. With the exception of some regionalprojects with a strong prediction focus, most GEWEXsupport comes through space agencies. To main-tain their support we need to demonstrate the valueof satellite products in understanding the climatesystem, in improving climate predictions, and inapplications such as water resource management.In the future, linkages to GEOSS, IGOS and WMOshould give GEWEX a more effective role in con-solidating viewpoints on satellite data needs fromour community and integrating them into clear state-ments of priority for the space agencies.

3August 2004

RECENT NEWS OF RELEVANCE TO GEWEX

PLANS ADVANCE FORTHE 5TH INTERNATIONAL GEWEX

SCIENCE CONFERENCEPlans have been progressing for the 5th Inter-

national Scientific Conference on the Global Energyand Water Cycle, June 20–24, 2005. The abstractsubmission and registration pages are complete (http://www.gewex.org/5thconf.htm). The AmericanMeteorological Society is now a cosponsor andnotices about the Conference will be published ina forthcoming Bulletin of the American Meteoro-logical Society issue.

AMMA PROJECT RECEIVES EU FUNDINGThe African Monsoon Multidisciplinary Analy-

ses (AMMA) Project has been successful in securinga large grant from the European Union to cover thecosts of people, radar networks, flux towers, lidarsand aircraft time. AMMA draws heavily on theFrench science community to provide leadershipfor the Project. AMMA plans to seek to beupgraded to Continental Scale Experiment status atan upcoming GEWEX Hydrometeorology Panelmeeting.

WORKSHOP ON TRENDS INGLOBAL WATER CYCLE VARIABLES

TO SUPPORT IPCCThis workshop, to be held November 3–5 in

Paris, France, brings together invited experts to dis-cuss secular trends in the global water cycle and theirpossible significance for assessments of climate change.Trends in the historical record will be consideredwith an emphasis being placed on those in the past20 to 30 years that can be seen globally in satelliteand integrated satellite in situ data products. Formore information about the Workshop, see the website at http://www.gewex.org/trendswkshp.htm.

PAN-WCRP MONSOON MEETINGAs a result of a decision at the JSC meeting to

improve the coordination among monsoon activitiesin WCRP, a meeting was held at the CLIVARconference in May 2004 to discuss options. As aresult of those discussions, it has been agreed thata workshop will be held on June 15–17, 2005 inIrvine, California to deal with the various aspectsof monsoons. A planning committee is now beingset up to prepare for this workshop.

INTERVIEWS HELD FOR THEEUROPEAN GEWEX COORDINATORThere was strong response to the advertisement

for a joint position for a person to work 50 per-cent of their time for the European Space Agencyand half their time as the European GEWEX coor-dinator. Interviews were held in late June 2004 anda qualified candidate has been offered the position.We hope to be able to introduce you to thesuccessful candidate in the next Newsletter.

GEWEX SUMMER EXECUTIVE MEETINGOn July 21–22, 2004, the Chair of the GEWEX

Scientific Steering Group, the GEWEX Panel Chairs,and the International GEWEX Project Office staffmet at the University of Maryland, Baltimore County,to review progress in GEWEX and to begin toprepare for the January 2005 SSG meeting. TheExecutive meeting was preceded by a 1-day brief-ing on GEWEX for the U.S. federal programmanagers at the National Academy of Sciences.The active interaction between program managersand GEWEX program leaders provided useful feed-back for the program. Central activities takingplace over the next 6 months include a series ofassessments for GEWEX Radiation Panel prod-ucts, a review of the role and future of the ContinentalScale Experiments at the upcoming GEWEX Hy-drometeorology Panel meeting and a workshop ontrends in global data sets. GEWEX also plans tofocus on the development of an implementationstrategy for Phase II over the next 6 months.

SOIL MOISTURE –WHAT IS YOUR DEFINITION?

(See page 11)

WATER CYCLE ROADMAP WORKSHOPHELD IN SEATTLE

As a follow-up to the 2003 US/Japan bilateralmeeting on the water cycle, a number of scientistsfrom Japan, the USA, and to a lesser extent Eu-rope met in Seattle on July 25–27, 2004 to discussa roadmap for water cycle observations and tocomment on the draft version of the 10-year GEOSSImplementation Plan. Organizers for the workshopincluded Norman Miller, Toshio Koike, Eric Woodand Rick Lawford.

August 20044

ENSEMBLE STREAMFLOWFORECASTING IN SNOWMELT

DOMINATED RIVER BASINS

Martyn Clark1, Lauren Hay2,Andrew Slater1, Kevin Werner3,

David Brandon3, Andrew Barrett1, SubhrenduGangopadhyay1, and Balaji Rajagopalan1

1CIRES, University of Colorado2USGS Water Resources Division

3Colorado Basin River Forecast Center

A common solution to the streamflow forecast-ing problem is to (1) run a hydrologic (orland-surface) model up to the start of the forecastperiod to estimate basin initial conditions, and (2)run the model into the future, with an ensemble offorecast inputs, to produce an ensemble of fore-casted streamflow. The River Forecast Centers inthe United States have used this “EnsembleStreamflow Prediction” (ESP) method for manyyears. The skill of the ESP approach depends onuncertainties in: (1) local-scale forecasts used todrive the hydrologic/land-surface model; (2) themodel’s estimate of basin initial conditions; and(3) the hydrologic/land-surface model itself. Thisarticle summarizes our research on all three aspectsof the streamflow forecasting problem.

The operational use of Numerical Weather Pre-diction (NWP) model output for hydrologicapplications is hampered by large biases in surfaceclimate fields, as well as poor (local-scale) forecastskill in many regions. Clark and Hay (2004) evalu-ated potential hydrologic applications ofmedium-range NWP model output. Before usingthe NWP forecasts as input to their hydrologicmodel, they used regression-based methods to re-move NWP model biases and extract (local)basin-scale forecast information. Results showedthe skill of NWP-based streamflow forecasts to bestrongly dependent on the hydro-climate regime(see figure at the bottom of page 16). In snow-melt-dominated basins, the ensemble streamflowforecasts had lower error, and higher probabilisticskill, than the climatological control case in whichhistorical data were used to construct the forecastensemble. In rainfall-dominated basins, the post-processed NWP forecasts added little skill overthe climatological control case. Work is ongoingto develop alternative downscaling approaches thatimprove over the simple regression-based methods(Gangopadhyay et al., in press; Gangopadhyay etal., under review).

As an extension of the Clark and Hay paper,Werner et al. (under review) conducted a feasibilitystudy to assess the use of medium-range NWPmodel output for operational streamflow forecastingapplications using the Colorado River Headwaterswith the state of Colorado as the test-basin(20,850 km2). The Colorado River Headwaters ismodeled by the Colorado Basin River ForecastCenter as 26 connected sub-basins, meaning thatforecasted precipitation and temperature fields mustreplicate the observed space-time variability in sta-tion time series. To meet this requirement, Clark et al.(2004a) implemented and tested a heuristic en-semble re-ordering method developed by JohnSchaake at the National Weather Service (NWS)Office of Hydrologic Development. This method waseffective in reproducing the observed spatial corre-lation structure, the temporal persistence, and theinter-variable correlations. The resultant streamflowforecasts had lower error and higher probabilisticskill than the climatological control case in whichhistorical data was used to construct the forecastensemble (Werner et al., under review).

For seasonal time scales, Clark et al. (2004b)and Werner et al. (in press) developed genericmethods to condition streamflow forecasts on proba-bilistic climate forecasts. The first approach isbased on a weather generator that produces alter-nate (conditioned) weather sequences for use inhydrologic/land-surface models (Clark et al., 2004b).The approach has three main steps: (1) condition-ally re-sample years from the historical record, suchthat the Cumulative Distribution Function (CDF) fromthe re-sampled set of years is equivalent to theCDF from a selected probabilistic forecast; (2) re-sample (daily, 6-hourly) data from the conditionedsubset of years to construct a desired number of(ensemble) weather sequences; and (3) re-orderthe ensemble members to preserve the observedspace-time variability.

The second approach (Werner et al., in press)involves forcing a hydrologic model with uncondi-tioned weather sequences from historical years,assigning weights to the weather sequence (streamflowtrace) associated with each historical year – basedon climate indices or probabilistic climate forecasts– and using these weights to construct a modified(conditioned) probabilistic streamflow forecast. Thismethod has been tested for three sub-areas of theColorado River basin: (1) the Salt River (Arizona);(2) the Upper Colorado River (western Colorado);and (3) the Green River (southwest Wyoming). Themethod produced multi-season streamflow fore-

5August 2004

casts with higher probabilistic skill than operationalmethods, with most significant forecast improve-ments evident in the Salt River basin where ElNiño/Southern Oscillation (ENSO) signals are strong.We are currently implementing these new methodsin the NWS operational streamflow forecasting sys-tems.

A considerable amount of forecast skill in snow-melt-dominated river basins is derived fromknowledge of the amount and extent of the accu-mulated snowpack. Thus, updating the model’sestimate of basin snowpack conditions at the startof the forecast period has potential to greatly in-crease the skill of the streamflow forecast. Barrett(under review) used estimates of snow-coveredarea from visible satellite imagery to update modelestimates of basin snow-covered area for two head-water basins in the Upper Colorado River.Application of a direct insertion technique for snowupdating actually degraded the skill of streamflowsimulations. Exclusion of forested pixels pro-duced streamflow simulations with lower error thanthe case when all pixels were included for updat-ing, which is symptomatic of the low accuracy ofvisible satellite estimates of snow-covered area underforest canopies. Quantifying errors in snow-cov-ered area products presents an outstanding challengeto using satellite products in hydrologic forecastmodels.

In an alternative approach, Slater and Clark (inpreparation) used station observations of SnowWater Equivalent (SWE) in an ensemble Kalmanfiltering scheme for snow updating. Model error isestimated by running SNOW-17 with ensembleforcings of precipitation and temperature (Clark

and Slater, in preparation) – the error in modelstates (i.e., SWE) is computed based on the model’sensemble spread. Observation errors (i.e., errorsin SWE) are taken as the cross-validated errorfrom the SWE interpolation. Results show thatsnow updating using the ensemble Kalman filterprovides improved snowpack simulations for sta-tions in the upper Colorado river basin (see figurebelow).

The hub of a model-based streamflow-forecast-ing system is the hydrologic/land-surface modelitself. Model performance depends on both tech-niques for estimating model parameters, as well asmodel structure. Hay (in preparation) developedan hierarchal approach to parameter estimation.Hay disaggregated the hydrologic modeling systeminto modules (solar radiation, evapo-transpiration,surface runoff, groundwater recharge, etc.), identi-fied calibration/validation data sets for each module(e.g., groundwater recharge estimated throughhydrograph separation), and calibrated each mod-ule in a stepwise/iterative manner. This approachwas tested over 67 basins across the contiguousUSA, and is found to result in considerable im-provements over a priori parameter estimates. Workis continuing to address issues of parameter uncer-tainty and model complexity.

This article has summarized recent work indeveloping experimental streamflow forecastingmethods for snowmelt-dominated river basins. Mostof our work has focused on developing local-scaleforecasts of model forcings (e.g., precipitation andtemperature), for forecast lead times from days

Results of snow updating at Grizzly Peak, Colorado for the water year 1989/90. Ensemble results are gray lines,while truth is plotted as asterisks, showing: a) the ensemble of cumulative precipitation forcing; b) raw ensemblesimulations from SNOW-17; c) snow updating results from the Ensemble Kalman Filter – the circles are ourestimated observations, complete with error estimates.

August 20046

through to seasons. These methods have resultedin tangible increases in forecast skill, in both ex-perimental and operational applications. Other,more embryonic, research has focused on improv-ing estimates of basin initial conditions, andaddressing parameter and structural issues in hy-drologic and land-surface models. Early resultsindicate that more attention in these areas willresult in significant increases in the skill of streamflowforecasts.

References

Barrett, A. P. (under review). Integrating remotely sensed snowcover with a distributed hydrologic model. Paper submitted toHydrologic Processes.

Clark, M. P. and L. E. Hay, 2004. Use of medium-range weatherforecasts to produce predictions of streamflow. J. Hydrometeor.,5, 15-32.

Clark, M. P., S. Gangopadhyay, L. E. Hay, B. Rajagopalan, andR. L. Wilby, 2004a. The Schaake Shuffle: A method to recon-struct the space-time variability of forecasted precipitation andtemperature fields. J. Hydrometeor., 5, 243-262.

Clark, M. P., S. Gangopadhyay, D. Brandon, K. Werner, L. E.Hay, B. Rajagopalan, and D. Yates, 2004b. A resampling proce-dure to generate conditioned daily weather sequences. WaterResources Research, 40, W04304, doi:10.1029/2003WR002747.

Clark, M. P., and A. G. Slater (in preparation). A method forgenerating high-resolution ensemble grids of surface climate fields.Paper in preparation for submission to the J. Appl. Meteor.

Gangopadhyay, S., M. P. Clark, B. Rajagopalan, K. Werner, andD. Brandon (in press). Effects of spatial and temporal aggrega-tion on the accuracy of statistically downscaled precipitation estimatesin the Upper Colorado River basin. Paper in press in J. Hydrometeor.

Gangopadhyay, S., M. P. Clark, and B. Rajagopalan (under re-view). Statistical downscaling using k-nearest neighbors. Paperunder review in Water Resources Research.

Hay, L. E., (in preparation). An hierarchal approach to parameterestimation. Paper in preparation for submission to theJ. Hydrometeor.

Slater, A. G., and M. P. Clark (in preparation). Snow dataassimilation via ensemble Kalman methods. Paper in preparationfor submission to the J. Hydrometeor.

Werner, K., D. Brandon, M. P. Clark, and S. Gangopadhyay (inpress). An evaluation of approaches for using climate indices forseasonal volume forecasting with the ensemble streamflow predic-tion system of the NWS. Paper in press in the J. Hydrometeor.

Werner, K., D. Brandon, M. P. Clark, and S. Gangopadhyay(under review). Incorporating Medium-Range Numerical WeatherPrediction Model Output into the Ensemble Streamflow PredictionSystem of the National Weather Service, Paper submitted to theJ. Hydrometeor.

ISLSCP INITIATIVE II UPDATE

J. Collatz1, F. Hall2,E.B. deCoulston3, and D. Landis3

1NASA GSFC, 2JCET/GSFC, 3SSAI/GSFC

The International Satellite Land Surface Clima-tology Project (ISLSCP) sponsored the productionof the first co-registered, peer-reviewed, documentedinterdisciplinary data collection called Initiative I,which included global, monthly surface meteorol-ogy, vegetation, soils, surface routing and runoff,atmospheric radiation data and clouds for 1987and 1988 at a 1-degree spatial resolution.

The ISLSCP Initiative II collection is greatlyexpanded in both time and spatial resolution. Ini-tiative II is a 10-year core global data collectionspanning the years 1987 to 1995 with improvedspatial and temporal resolution (one-quarter to 1degree) using improved data generation algorithms.In addition, Initiative II includes some data setsspanning the 18-year period, 1982–1999. Thedata collection includes additional carbon and so-cioeconomic data sets uniquely designed to supportglobal carbon cycling studies. Initiative II pro-vides a comprehensive collection of high priorityglobal data sets in a consistent data format and Earthprojection.

The full Initiative II collection consists of 47different data types with 230 different parametersshown below (number of parameters in parentheses).

• Carbon (15)• Vegetation (29)• Hydrology, Topography and Soils (39)• Near-Surface Meteorology (95)• Snow and Sea Ice (4)• Oceans (1)• Radiation and Clouds (45)Each of the 47 data sets and its documentation

has undergone two peer reviews, one by reviewersfamiliar with the data set (but not the producer of it)and a second reviewer, a potential user who has noprevious experience with it. These reviews are nearlycomplete to be followed by an evaluation of theentire collection prior by the Initiative II staff and asmall group of users. This collection-overview evalu-ation will include an evaluation of the individual dataseries against each other and independent data sourcesand an evaluation of the value of the ISLSCP IIcollection as a whole. Initiative II data sets are nowin production and scheduled for completion mid2005. Much of the data is already available at http://islscp2.sesda.com/.

7August 2004

LAND SURFACE MODELINGIN BALTEX

Heinz-Theo Mengelkamp

Institute for Coastal ResearchGKSS Research Center Geesthacht GmbH

A primary objective of the Baltic Sea Experi-ment (BALTEX) is the understanding and simulatingof the exchange processes between the Earth’ssurface and the atmosphere. Focus is on the im-provement of land-surface models (LSM) in weatherforecasts and regional climate models, and the in-clusion of hydrological processes through couplingthe atmospheric models with river routing schemes.The exchange processes between the sea surface/sea ice and the atmosphere play a dominant roleover the Baltic Sea area. In order to test andvalidate LSMs, measurement campaigns have beenundertaken over open sea, at coastal areas andover land during all seasons, and include continuouslong-term observations at the Coordinated EnhancedObserving Period (CEOP) sites.

The Oestergarnsholm air-sea research projecthas led to a new understanding of the marineatmospheric boundary layer (MABL): a) the MABLis strongly influenced by the sea state; b) forgrowing sea conditions (young waves), the wavestravel slower than the wind velocity and the turbu-lence structure of the MABL resembles the boundarylayer over land; c) for a more mature sea or mixedsea, some waves are travelling faster than the wind,and the MABL starts to deviate from the boundarylayer over land; and d) for swell conditions, domi-nated by long waves travelling faster than the wind,the MABL is significantly different from the boundarylayer over land (Smedmann and Hoegstroem, 2004).

Measurement campaigns over land are concen-trated on the surroundings of the BALTEX referencesites in CEOP at Norunda, Sweden (forested ar-eas); Sodankylae, Finland (snow covered surfaces);and Lindenberg, Germany (heterogeneous terrain).In May and June 2003, a major field experimentwas carried out around the Lindenberg site as partof the Evaporation over Grid and Pixel Scale (EVA-GRIPS) project under the auspices of the GermanClimate Research Program, DEKLIM. EVA-GRIPSfocuses on the determination of the area-averagedevaporation and sensible heat flux over a heteroge-neous land surface, both from experimental dataand from modelling activities. The purpose of EVA-GRIPS is the verification and implementation of

surface flux averaging strategies into the land sur-face schemes of the climate model, REMO/ECHAM(Regional), and the German Weather Service fore-cast model LM (Local) and as a control, the LSMSEWAB (Mengelkamp et al., 2001). The LSMs areobjectively calibrated using the MOSCEM algorithmof the University of Arizona (Gupta et al., 1998)with flux data from 12 sites over the EVA-GRIPSexperimental area around the Lindenberg observa-tory of the German Weather Service. Compositeflux data are estimated from the point data byweighing them with the respective grid fraction ofthe vegetation type. The time series of these com-posite fluxes are used to verify averaging strategiesand to deduce grid specific effective parameterthrough calibration. The mosaic approach has shownthe best agreement between observed and simu-lated grid representative fluxes.

The alliance to the GEWEX community is main-tained through regular participation of the LSMSEWAB in the Project for Intercomparison of Land-surface Parameterizations (PILPS; Henderson-Sellerset al., 1995) activities. In PILPS Phase 2e, coupledland-surface hydrological models were compared withrunoff and discharge from the Thorneaelv river inSweden into the Baltic Sea. This exercise providedthe opportunity to compare the LSS SEWAB and acoupled routing scheme (Lohmann et al., 1996).

The SEWAB/routing scheme model system wasoriginally developed as a hydrological model forthe Odra watershed. It is coupled to the mesoscalemodel MM5 for flood forecast purposes. Besidesa variety of river routing schemes for various wa-tersheds around the Baltic Sea, the conceptualhydrological model, HBV (Bergström and Forsman,1973), and the distributed model, LARSIM (Rich-ter et al., 2003) cover the whole Baltic Sea catchmentarea.

At the end of BALTEX Phase I in 2005, coupledatmospheric – hydrological model systems will havebeen extensively validated for single catchmentsand for the entire Baltic Sea area.

References

Bergström, S. and A. Forsman, 1973. Development of a con-ceptual deterministic rainfall-runoff model, Nordic Hydrology,4, 147-170

Gupta, H. V., S. Sorooshian and P. O. Yapo, 1998. Towardimproved calibration of hydrologic models: Multiple andnoncommensurable measures of information, Water Res. Res.,34, 4, 751-763.

(Continued on page 15)

August 20048

ULTRA-HIGH RESOLUTIONOBSERVATION-DRIVEN LAND

MODELING NEEDED TO ENABLE THEDEVELOPMENT OF GLOBAL CLOUD

RESOLVING EARTH SYSTEM MODELS

Paul R. Houser, Mike Bosilovich,Christa Peters-Lidard, and Wei-Kuo Tao

NASA Goddard Space Flight Center

The relatively crude representation of cloudprocesses in global climate models has long beenrecognized as a primary source of uncertainty inclimate change predictions. It is well known thatthe high-resolution time and space complexity ofland surface phenomena have significant influenceon atmospheric boundary layer turbulence and cloudprocesses. Scale truncation errors, unrealistic physicsformulations, and inadequate coupling between sur-face fluxes and the overlying atmosphere causeserious systematic errors. Standard subgrid-scaletiling approaches are not able to adequately repre-sent observed heterogeneity and boundary-layerinteraction (Bosilovich, 2002), which means thatfine-scale model representations are required. Theinability to explicitly account for the considerablespatial and temporal variability in terrain topogra-phy, surface properties, rainfall, and net surfaceradiation constitute an organic weakness of currentclimate models and a cause for substantial errorsin model simulations of near-surface climateover land.

Therefore, the land surface research communitymust progress toward a fully process-scale resolv-ing model of land surface hydrology, atmosphericdynamics, and cloud processes over the globaldomain (Tao et al., 2003). We must integrate allobviously interdependent land-atmosphere processesinto a common ultra-resolution (100s of meters)framework for Earth system modeling, through fu-sion of traditional land surface hydrology moduleswith boundary-layer turbulence and cloud processmodules. Decisions regarding the model formula-tions must be guided to the greatest extent possibleby the use of observations, as prescribed input,data assimilation constraints, validation or relevanceto applications or decision-making.

To this end, we envision two, eventually con-vergent, paths toward fully interactive land-atmospherecoupling: (1) In the near-term, implement traditionalcloud parameterization and atmospheric turbulenceschemes and implicitly couple those (through theatmospheric circulation) to patch-based land mod-

els at highest possible resolution; and (2) In thelong-term, develop true global process-resolvingcoupled land-atmosphere models in a phased ap-proach, starting with (a) off-line land-cloud processresolving studies, then progressing to (b) land-cloud super-parameterizations based on samplingthe relevant process scales, (c) nested land-cloudresolving models in a Global Circulation Model(GCM) framework, and finally (d) to a true globalultra-high-resolution global cloud-land process re-solving model. These two paths will eventuallyconverge when computing power allows the reso-lution of the Earth system model to overlap theresolution of the global cloud resolving model.

The unprecedented availability of new globalland-surface remote sensing data over the pastdecade should be a fundamental driver for thedevelopment of new scientific understanding andmodeling innovations. Land data assimilation sys-tems have been developed that use sophisticatedland surface models to ingest satellite and ground-based observations, as parameters, forcing, anddata for assimilation, in order to produce the bestpossible fields of land surface states and fluxes.The multi-institution North American Land DataAssimilation System (NLDAS) project was the firstto embrace this concept (Mitchell et al., 1999). Itssuccess led to the development of Global LDAS(GLDAS) (Houser and Rodell, 2002; Rodell et al.,2004; http://ldas.gsfc.nasa.gov) through the jointeffort of scientists at the National Aeronautics andSpace Administration's (NASA) Goddard SpaceFlight Center and the National Oceanic and Atmo-spheric Administration's (NOAA) National Centersfor Environmental Prediction (NCEP).

The 1/4-degree resolution, high quality, near-real time and retrospective output fields that haveresulted from GLDAS (see the figure at the bottomof page 1), the first of their kind, are providing thebasis for global scale studies of the hydrologicalcycle and meteorological processes. In addition,GLDAS surface state fields are being tested inweather and climate model initialization studies atNCEP and NASA’s Global Modeling and Assimi-lation Office (GMAO), where GLDAS soil moisturefields have been shown to improve the predictabil-ity of seasonal precipitation, and are being testedas input to water management decision support sys-tems.

The Land Information System (LIS) Project(http://lis.gsfc.nasa.gov) has streamlined andparallelized the GLDAS code and has executed 1-km resolution, global simulations using 3 differentland models on high performance computing plat-

9August 2004

forms. LIS incorporates Assistance for Land sur-face Modeling Activities (ALMA) and Earth SystemModeling Framework (ESMF) standards to facili-tate inter-operation with other Earth system models.LIS also employs a Grid Analysis and DisplaySystem – Distributed Oceanographic Data System(GrADS-DODS) server framework which allowsfor seamless access to large observational data-bases. LIS is currently being coupled to the WeatherResearch and Forecasting (WRF) and GoddardCumulus Ensemble (GCE) models to explore sur-face-layer feedback effects due to assimilation.However, there remain great challenges in repre-senting process-scale land-surface dynamics in Earthsystem models, such as the need for LSM-compat-ible groundwater, glacier, ice sheet, and wetlandmodels and schemes for simulating the effects ofdams, agriculture, and irrigation on land surfacehydrology.

New global remote sensing observations providethe foundation for the development of a new gen-eration of Earth system models that will explicitlyresolve weather and climate relevant physical, chemi-cal and biological processes, in order to improvedramatically the understanding and prediction ofweather and climate. This will require, among otherthings, an ultra-high-resolution observation-driven landsurface model with process-scale hydrology andbiogeochemistry dynamics that is implicitly coupledto high-resolution boundary-layer turbulence and cloudmicrophysics parameterizations. These innovationswill be invaluable for a wide range of applications,including satellite data assimilation, observation sys-tem design, weather forecasting and climate simulation.

References

Bosilovich, M. G., 2002. On the use and validation of mosaicheterogeneity in atmospheric numerical models. Geophys. Res.Lett., 29, 10.1029/2001GL013925.

Houser, P. and M. Rodell, 2002. GLDAS: An Important Contri-bution to CEOP. GEWEX News, May 2002, 2,8,9,16.

Mitchell, K. E., et al. 2004. The multi-institution North AmericanLand Data Assimilation System (NLDAS): Utilizing multiple GCIPproducts and partners in a continental distributed hydrological modelingsystem, J. Geophys. Res., 109, D07S90, doi:10.1029/2003JD003823.

Rodell, M., P. R. Houser, U. Jambor, J. Gottschalck, K. Mitchell,C.-J Meng, K. Arsenault, B. Cosgrove, J. Radakovich, M. Bosilovich,J. K. Entin, J. P. Walker, D. Lohmann, and D. Toll, 2004. TheGlobal Land Data Assimilation System. Bull. Amer. Meteor. Soc.,85 (3), 381-394.

Tao, W.-K. and co-authors, 2003. Microphysics, radiation andsurface processes in the Goddard Cumulus Ensemble (GCE) model,Meteorology and Atmospheric Physics, 82, 97-137.

THE ROLE OF FINE-SCALELANDSCAPE AND SOIL MOISTURE

VARIABILITY IN THE INITIATION OFDEEP CONVECTION

Fei Chen, Thomas T. Warner, Stanley B.Trier, and Kevin W. Manning

National Center for Atmospheric Research

Understanding the feedback between land-surfacevariability and precipitation has been a central GEWEXissue because of its potential benefit in improvingclimate predictability. In the summer, mesoscale bound-aries play a critical role in the initiation of heavyprecipitation. The zones of enhanced convergencealong these boundaries have been recognized as areasof deep-convection initiation. The origin of thesemeso-scale boundaries includes synoptic-scale fronts,outflows from previous storms, orographic features,and differential surface heating. The differential heat-ing can be enhanced by heterogeneities in land-surfaceconditions. The land surface may have differing im-pacts, depending on atmospheric conditions. Small-scaleground features, such as vegetation, hillslopes, andurban or industrial areas can also have subtle impactsthat can determine the exact boundary and intensityof storms.

Chen et al. (2001) examined the impact of theland surface and terrain on the 1996 flash flood thatstruck Colorado's Buffalo Creek watershed. Amongtheir findings: the wildfire burn area received particu-larly heavy rainfall during the flood, possibly becausethe denuded site transferred more heat into the atmo-sphere than an area filled with living trees would.That extra heat would enhance a local storm byadding to the buoyancy of updrafts. In a studyfocusing on heavy precipitation associated with adryline in the south central United States, Trier et al.(2004) found that fine-scale (L~10 km) boundary-layer circulations that directly trigger deep convectionare confined within a mesoscale region containing adeeper and more unstable Planetary Boundary Layer(PBL), and that this region is a result of a surfacesensible heat-flux maximum over dry soils. Theyutilized the soil moisture fields from a high-resolutionland data-assimilation system (HRLDAS) and fromthe National Centers for Environmental Prediction(NCEP) Eta Model Data Assimilation System (EDAS)to initialize the MM5 model. The simulation initializedwith EDAS soil moisture did not initiate deep con-vection along the dryline in Texas (see figure at thetop of page 16) as shown in the satellite imagebecause of subtle differences in soil moisture and inthe subsequent evolution of the boundary layer.

(Continued on page 15)

August 200410

UPDATE ON THE SECOND GLOBALSOIL WETNESS PROJECT (GSWP-2)

Xiang Gao1, Paul A. Dirmeyer1,and Taikan Oki2

1Center for Ocean-Land-Atmosphere Studies,Calverton, Maryland, USA, 2Institute of Indus-

trial Science, University of Tokyo, Japan

The Second Global Soil Wetness Project(GSWP-2) is the large-scale uncoupled modelingactivity of the Global Land-Atmosphere SystemStudy (GLASS), one of three parallel efforts underthe GEWEX Modeling and Prediction Panel (GMPP).GSWP-2 is closely linked to the 10-year (1986–1995) International Satellite Land Surface ClimatologyProject (ISLSCP) Initiative II land surface data set(http://islscp2.sesda.com/), but takes a step for-ward with corrections to the systematic biases inthe National Centers for Environmental Prediction(NCEP)-Department of Energy (DOE) reanalysisfields made by hybridization of the 3-hourly analy-sis with global observationally based gridded datasets at lower temporal resolution (Zhao and Dirmeyer,2003, ftp://grads.iges.org/pub/ctr/ctr_159.pdf). LandSurface Scheme (LSS) simulations in GSWP-2 areconducted at a spatial resolution of 1 degree. Sofar, baseline simulation results from 16 modelshave been submitted. A major product of GSWP-2 will be global multi-model land surface analyseson daily and monthly intervals from January 1986–December 1995. These multi-model results will bemade available in 2005.

A new thrust for GSWP-2 is to link landsurface modeling better to applications in remotesensing. GSWP-2 intends to expand the validationand assimilation capabilities of current LSSs be-yond those few areas where in situ data are readilyavailable. This is done by coupling LSSs with avalidated L-band microwave emission model (MEB;Pellarin et al., 2003) to simulate forward brightnesstemperature as observed from microwave radiom-eters. Several participating LSSs have been examinedfor their performance in simulated brightness tem-perature, which is assessed by comparisons withairborne measurements from several large-scalecampaigns held during the GSWP-2 period. Suchan assessment constitutes a useful prototype ofLSS validation on a global scale when future sat-ellite-based L-band radiometry data are available,such as the NASA Hydrosphere State (HYDROS)mission and the European Space Agency (ESA)Soil Moisture Ocean Salinity (SMOS) mission. Thefigure at the top of page 1 shows an example ofa global 1-degree map of synthetic L-band H-

polarized brightness temperature as it would beseen by HYDROS. In the future, comparison ofbrightness temperature produced by coupled mod-els with satellite-based L-band radiometry data willhelp us to understand better the performance ofvarious LSSs, as well as provide a reference forquality assurance of remotely sensed brightnesstemperature imagery.

Modeling sensitivity studies in GSWP-2 involvere-integrating the LSSs over the 10-year periodwith alternative boundary conditions or forcings totest the response of the models to uncertainties inthose data sets. The sensitivity studies within eachGSWP participating group are now under way.They include sensitivity to choices in meteorologi-cal forcing data, land surface parameters (e.g.,global vegetation class maps), and surface vegeta-tion data (e.g., climatological vegetation). Most ofthe participants have shown strong interest in in-vestigating sensitivity to choice of the near-surfacereanalysis product, and choice of precipitation datasets. A new experiment is also added where LSS-native or default parameters are used for soil andvegetation instead of GSWP-2 global fields as aLSS is often developed and calibrated to use spe-cific data sets.

GSWP-2 is a project in progress, and a num-ber of scientific results and data sets will emergeover the next 1-2 years. The latest results fromthis project will be presented at the GSWP-2 sci-ence workshop in Kyoto, Japan on 13–15 September2004 (http://hydro.iis.u-tokyo.ac.jp/Info/GSWP/); ata special session of the 85th AMS Annual Meetingto be held in San Diego, California on 9–13 Janu-ary 2005; and at the 5th GEWEX InternationalScientific Conference in Orange County, Califor-nia, 20–24 June 2005.

Complete descriptions, access to forcing data,and the latest information concerning the evolvingof the GSWP2 project are available at http://www.iges.org/gswp/. This project is supported byNational Aeronautics and Space Administration GrantNAG5-11579.

References

Zhao, M., and P.A. Dirmeyer, 2003. Production and Analysisof GSWP-2 near-surface meteorology data sets. COLA Tech-nical Report, 159, 38pp.

Pellarin, T., J.-P. Wigneron, J.-C. Calvet, M. Berger, H. Douville,P. Ferrazzoli, Y.H. Kerr, E. Lopez-Baeza, J. Pulliainen, L.P.Simmonds, and P. Waldteufel, 2003. Two-year global simulationof L-band brightness temperature over Land. IEEE Trans.Geosci. Remote Sens., 41, 2135-2139.

11August 2004

SOIL MOISTURE—MUDDY PROSPECTSFOR A CLEAR DEFINITION

Paul Dirmeyer

Center for Ocean-Land-Atmosphere StudiesCalverton, Maryland, USA

Soil moisture is a key state variable connectingthe land surface to climate variability and predict-ability. However, not only is soil moisture sparselymeasured around the globe, there is also little agree-ment on what is meant by soil moisture.

According to the International Glossary ofHydrology, “soil moisture” is defined as the mois-ture contained in the portion of the soil which isabove the water table, including water vapor, whichis present in the soil pores. Contrast this to “soilwater” which is defined as water suspended in theuppermost belt of soil, or in the zone of aerationnear the ground surface, that can be dischargedinto the atmosphere by evapotranspiration. TheGlossary of Meteorology describes “soil moisturecontent” as the amount of water in an unsaturatedsoil, expressed as a volume of water per unit vol-ume of porous media, or as a mass of water perunit oven-dry mass of soil.

None of these “official” definitions are veryprecise or satisfying. Part of the problem is thatsoil moisture is a quantity that lies at the interfaceof several disciplines, each with its own history andjargon. Nevertheless, for the purposes of scientificresearch, more precise definitions are needed. Soilmoisture as a measurable quantity can be charac-terized in one of three broad categories: in terms ofa water mass (expressed in units of mass, volume,density or depth), as a physical index (usually ona scale of 0-1 or 0-100% based on a ratio ofvolumes or weights of water to water + soil), oras an observational index (based on a remotesensing retrieval algorithm, electrical conductivity,pulse transmission, or some other measurement ofa physical property which is known to vary with soilmoisture).

The Global Soil Wetness Project (GSWP) hasinformally adopted the convention of referring tothe water mass definitions as “soil moisture,” whichalways have associated units such as kgm-3 or mm.Gravimetric measurements of soil moisture, wherea soil core is weighed, baked to remove moisture,then weighed again, produce a measurement of soilwater mass. Models of the land surface typically

calculate closed water budgets, so they usuallytreat soil moisture in terms of volumetric quantity,which is 0 for totally desiccated soils, and equal tothe total volume of the soil pore space when thesoil is saturated. Volumetric soil moisture can beeasily converted to water mass by multiplying bythe density of water, usually assumed constant.

In GSWP, physical indices are referred to as“soil wetness” (always dimensionless). Many mod-els and measurement methods tabulate water in thesoil as either a volumetric soil wetness (volumetricsoil moisture divided by the total volume of the soilsample), or a saturation ratio (ranging from 0 to 1from desiccated to saturated, equal to the volumet-ric soil wetness divided by the soil porosity). Thereexist nodes for defining indices other than totaldesiccation and saturation, adding to the confusionin this category. At the dry end, other anchorpoints for indices include the wilting point (whichitself has several definitions and means of estima-tion) and the lowest measured value. The wiltingpoint is considered to be the soil wetness levelbelow which plants can no longer extract waterfrom the soil, and is used as the base for estimating“plant available water”, which itself can be ex-pressed as either a “moisture” or a “wetness.” Atthe wet end of the scale, the field capacity (theamount of water a soil column can hold by capillaryretention against gravity in the absence of evapora-tion), the level of zero vegetation water stress andthe highest measured value are also used. Mostindices set their lower node equal to 0, and theupper node to 1 or 100%. GSWP-1 defined its“Soil Wetness Index” as ranging from 0 at wiltingpoint to 1 at field capacity. So-called “bucket” soilmodels implicitly cover only this range, while moresophisticated soil column models in land surfaceschemes with vegetation and baseflow are capableof soil wetness outside of these bounds, particularlywhen runoff (another muddy quantity) is parameter-ized rather arbitrarily as a function of soil wetness.

Observational indices are artifacts of electronicinstruments and serve as proxies for direct soilmoisture measurements such as gravimetric assays.They may be expressed as voltages, impedances,radiative brightness temperatures, or ratios of bright-ness temperatures, to name a few. Observationalindices can be calibrated to correspond to physicalindices or water mass measurements, or cross-calibrated with indices from other electronicinstruments. However, many factors can affectcalibration, making transferability difficult, especiallywhen soil moisture is being measured by remote

August 200412

GEWEX RELEVANT PUBLICATIONSOF INTEREST

This is a new section to alert readers topapers of possible interest.

Global Soil Moisture from SatelliteObservations, Land Surface Models,and Ground Data: Implications for

Data Assimilation

Reference: Reichle, R. H., R. D. Koster, J. Dong,A. A. Berg, 2004. J. Hydrometeorology, Vol. 5,No. 3, 430-442.

Summary/Abstract: Three independent surface soilmoisture data sets for the period 1979–87 are com-pared: 1) global retrievals from the ScanningMultichannel Microwave Radiometer (SMMR); 2)global soil moisture derived from observed meteoro-logical forcing using the NASA Catchment LandSurface Model; and 3) ground-based measurementsin Eurasia and North America from the Global SoilMoisture Data Bank. Time-average soil moisturefields from the satellite and the model largely agreein the global patterns of wet and dry regions.Moreover, the time series and anomaly time seriesof monthly mean satellite and model soil moistureare well correlated in the transition regions be-tween wet and dry climates where land initializationmay be important for seasonal climate prediction.However, the magnitudes of time-average soil moistureand soil moisture variability are markedly differentbetween the data sets in many locations. Absolutesoil moisture values from the satellite and the modelare very different, and neither agrees better withground data, implying that a “correct” soil moistureclimatology cannot be identified with confidence fromthe available global data. The discrepancies be-tween the data sets point to a need for bias estimationand correction or rescaling before satellite soil moisturecan be assimilated into land surface models.

sensing. Many assert the future of soil wetnessmeasurement lies in satellite sensors in the micro-wave band. Several specific frequencies at variouspolarizations have been proposed and are beingdeveloped. However, all of these sense only thefirst few millimeters of water, whether it be onthe soil or in overlying vegetation. Problemsinclude the effects of surface temperature, rough-ness, ice and vegetation that influence the registeredbrightness temperature and the need to translatethe sensed radiation into a water quantity.

So what is the correct definition for soil mois-ture? It depends on what you are trying toaccomplish. For example, for the agriculturalplanner concerned about crop water stress, soilmoisture is a real and precise quantity with awell known, direct impact on a tangible commod-ity. Indeed, water mass is a fairly incontrovertiblequantity, albeit very difficult to measure adequately.In contrast, many weather and climate modelerstreat soil moisture as a means to an end – afictional quantity in an equation for calculatingsurface energy and moisture fluxes. In theirview, the land surface model gives reasonableheat fluxes to the atmosphere and it is relevantif the underlying soil state variable validates wellagainst observations. Hydrologists, who worryabout accurate streamflow predictions, find thisapproach unacceptable. Given the complexities anduncertainties in soil structure, and its effects onvertical and lateral water movement, it may notbe possible to have a single useful definition.However, we do need more discussion and docu-mentation of this issue to have a commonunderstanding of what is meant when we use theterm. For more information on the basics of soilmoisture, see Chapter 6 of Yu et al. (1993).

Reference

Yu, C., C. Loureiro, J.-J. Cheng, L. G. Jones, Y. Y. Wang, Y.P. Chia, and E. Faillace, 1993. Data Collection Handbook toSupport Modeling Impacts of Radioactive Material in Soil.Environmental Assessment and Information Sciences Division,Argonne National Laboratory, Argonne, Illinois, USA [http://web.ead.anl.gov/resrad/documents/data_collection.pdf], 158 pp.

Impacts of Land Surface ModelComplexity on a Regional Simulation of

a Tropical Synoptic EventReference: H. Zhang, J. L. McGregor, A.Henderson-Sellers, J. J. Katzfey, 2004. J. Hy-drometeorology, Vol. 5, No. 1, pp. 180–198.

Summary/Abstract: The complexity of landsurface models has tended to grow as our knowl-edge of land surface processes has increased andour appreciation of the importance of land surfaceprocesses in the weather and climate system hasdeveloped. However, results from the Project for

INTERESTED IN SUBMITTING ANARTICLE TO BE PUBLISHED IN

GEWEX NEWS?

See the guidelines for submitting articles at:

http://www.gewex.org/newsguidelines.htm

13August 2004

Temporal Interpolation of Global SurfaceSkin Temperature Diurnal Cycle Over

Land Under Clear and Cloudy Conditions

Reference: Aires, F., C. Prigent, and W. Rossow,2004, J. Geophys. Res., 109, D04313, doi:10.1029/2003JD003527.

Summary/Abstract: The surface skin temperatureis a key parameter at the land-atmosphere interface.An accurate description of its diurnal cycle wouldnot only help estimate the energy exchanges at theinterface, it would also enable an analysis of theglobal surface skin diurnal cycle and its variabilitywithin the last 20 years. Two main applications ofsuch a data set can be identified. First, the sur-face skin temperature could be used to evaluatesurface models and could also provide an addi-tional constraint on latent and sensible heat fluxcalculations. The second main application of asurface skin temperature data set is related toclimate change studies. Incomplete space and timecoverage complicates the analysis of global trendsfrom the in situ meteorological network; however,skin temperature can be estimated from satelliteinfrared radiance observations.

This study is based on the 3-hourly surfaceskin temperature estimated by the International SatelliteCloud Climatology Project (ISCCP) from the infra-red measurements collected by the polar andgeostationary meteorological satellites. The diurnalcycle of surface skin temperature is analyzed al-most globally (60N–60S snow-free areas), using aPrincipal Component Analysis (PDA). The firstthree components are identified as the amplitude,the phase, and the width (i.e., daytime duration) ofthe diurnal cycle and represent 97% of the variabil-ity. PCA is used to regularize estimates of thediurnal cycle at a higher time resolution. A newtemporal interpolation algorithm, designed to workwhen only a few measurements of surface tempera-ture are available, is developed based on the PCArepresentation and an iterative optimization algo-rithm. This method is very flexible: only temperaturemeasurements are used (no ancillary data), no sur-face model constraints are used, and the time andnumber of measurements are not fixed. The perfor-mance of this interpolation algorithm is tested forvarious diurnal sampling configurations. In particu-lar, the potential to use the satellite microwaveobservations to provide a full diurnal surface tem-perature cycle in cloudy conditions is investigated.

by parameter tuning in simplified land surfaceschemes.

Intercomparison of Land Surface ParameterizationSchemes (PILPS) have revealed poor agreementamong current land surface schemes in represent-ing key surface processes such as surface waterand energy partitioning when forced with the samemeteorological forcing. Three questions then arise:(i) do land surface processes matter in the numeri-cal model simulations of weather and climate? Then,if assured in the affirmative, (ii) how complexshould land surface schemes be in current numeri-cal models, particularly when considering that therepresentations of known important atmosphericprocesses such as convection and clouds in nu-merical models are still simple compared with reality?and (iii) can such complexity be represented byparameter tuning?

A multimode Chameleon Surface Model (CHASM)with different levels of complexity in parameteriz-ing surface energy balance is coupled to a limited-areamodel (DARLAM) to investigate the impacts ofcomplexity in land surface representations on themodel simulation of a tropical synoptic event. Alow pressure system is examined in two sets ofnumerical experiments to discuss the following. (i)Does land surface parameterization influence re-gional numerical weather simulations? (ii) Can thecomplexity of land surface schemes in numericalmodels be represented by parameter tuning? Themodel-simulated tracks of the low pressure centerdo not, overall, show large sensitivity to the differ-ent CHASM modes coupled to the limited-areamodel. However, the landing position of the sys-tem, as one measurement of the track difference,can be influenced by several degrees in latitude andabout one degree in longitude. Some of the trackdifferences are larger than the intrinsic numericalnoise in the model estimated from two sets ofrandom perturbation runs. In addition, the landingtime of the low pressure system can differ byabout 14 h. The differences in the model-simulatedcentral pressure exceed the model intrinsic numeri-cal noise and such variations consistent with thedifferences seen in simulated surface fluxes. Fur-thermore, different complexity in the land surfacescheme can significantly affect the model rainfalland temperature simulations associated with thelow center, with differences in rainfall up to 20 mmday-1 and in surface temperature up to 2°C. Explic-itly representing surface resistance and bare groundevaporation components in CHASM produces themost significant impacts on the surface processes.Results from the second set of experiments, inwhich the CHASM modes are calibrated by param-eter tuning, demonstrate that the effects of thephysical processes represented by extra complexityin some CHASM modes cannot be substituted for

August 200414

4TH STUDY CONFERENCE ONBALTEX

24–28 May 2004Gudhjem, Bornholm, Denmark

Hans-Jörg IsemerGKSS Research Centre, Germany

More than 110 scientists from 15 countriesgave 78 oral and 36 poster presentations at theConference, which was organized jointly by RisøNational Laboratory, the Technical University ofDenmark and GKSS Research Centre. The Confer-ence was organized into the following sessions:Improved understanding of water and energy cycleprocesses: (remote sensing applications, diagnosticstudies and field experiments, numerical model-ling); trends and variability in the regional climateduring the past two centuries; development andvalidation of advanced modeling tools for regionalclimate studies; projection of future climate changeat river catchment and basin scales during the 21st

century; and applications for water resources man-agement, including extreme events, long-term changesand studies on air and water quality.

Speakers at the opening session, including Pro-fessors Sven-Erik Gryning (Risø) and Hartmut Graßl(MPI Hamburg), stressed that the Conference washeld at an important time in the BALTEX programme.After about 10 years of successful basic researchcontributing to new data sets, better process un-derstanding and model development of the waterand energy cycles of the Baltic Sea basin, BALTEXhas recently defined revised objectives for Phase IIof the programme. A future focus of BALTEX willbe on applications of its past achievements toother fields, both in scientific research and be-yond, where knowledge of the water and energycycle is of fundamental importance. Nearly 30 percentof all presentations given were alluding already toaspects of the revised BALTEX objectives for

Phase II, such as investigations on past and futureclimate, as well as contributions to water resourcemanagement.

The following statements summarize majorachievements of the BALTEX programme that werehighlighted at the Conference:

• Improved understanding of the processes in theair land surface, air-sea surface, and air-sea iceboundary layers through the exploitation of avariety of new and experimental data sets andmodels.

• Improved precipitation data sets for both thesea and over land, based on both gauge andradar data.

• New measurements and improved understand-ing of the water mass exchanges within thedeep basins of the Baltic Sea and of turbulentmixing in the Baltic Sea.

• A better understanding of major Baltic Seainflow events and an improved modeling ofextreme inflow events.

• The water and heat budgets of the Baltic Seahave been investigated with increasing detail;the accuracy of the individual terms needs tobe addressed, and further efforts towards ana-lyzing the budgets, as well as their long-termtrends and variability for the entire Baltic Seabasin are still necessary.

• Two coupled model systems, RCAO andBALTIMOS, are progressing from their de-velopment and validation phases to applications,the potential of these systems has been dem-onstrated.

• The necessity of studying and understandingprocesses at different scales was highlighted.

• The need for a stronger focus on uncertaintyassessment, including ensemble runs of futurescenarios is evident.

The next BALTEX Conference is scheduled tobe held in 2007 on the Estonian island Saaremaa.

WORKSHOP/MEETING SUMMARIES

Participants at the 4th Study Conference on BALTEX.

15August 2004

GEWEX/WCRP MEETINGS CALENDARFor the complete listing of meetings, see the

GEWEX web site (http://www.gewex.org)

GEWEX NEWSPublished by the International GEWEX Project Office

Richard G. Lawford, DirectorDawn P. Erlich, Editor

Mail: International GEWEX Project Office1010 Wayne Avenue, Suite 450Silver Spring, MD 20910, USA

Tel: (301) 565-8345Fax: (301) 565-8279

E-mail: gewex@gewex.orgWWW Site: http://www.gewex.org

Henderson-Sellers, A., A. Pitman, P. Love, P. Irrannejad, andT. Chen, 1995. The project for intercomparison of land-surfaceparameterization schemes (PILPS): Phases 2 and 3. Bull. Amer.Meteor. Soc., 94, 489-503.

Lohmann, D., R. Nolte-Holube, and E. Raschke, 1996. Alarge scale horizontal routing model to be coupled to landsurface parameterization schemes. Tellus, 48A, 708-721.

Mengelkamp, H.-T., Kiely, G., and Warrach, K., 2001. Evalu-ation of hydrological processes added to an atmosphericland-surface scheme, Theor. Appl. Climatol, 69, 199-212.

Richter, K.-G., P. Lorenz, M. Ebel, and D. Jacob, 2003. Analysisof the water cycle of the BALTEX basin with an integratedatmospheric hydrological ocean model, Proc. 4th Study Conf. onBALTEX, Bornholm, May 2004, 154-155.

Smedmann, A.-S. and U. Hoegstroem, 2004. The marine boundarylayer – new findings from the Oestergarnsholm air-sea interac-tion site in the Baltic Sea, Proc. 4th Study Conf. on BALTEX,Bornholm, May 2004, 41-42.

LAND SURFACE MODELING IN BALTEX(Continued from page 7)

Results from these and other recent researchstudies provide some hope that the careful treat-ment of land-surface physics and soil moisture inconvection-resolving models can lead to increasedrainfall predictability. In particular, this should beachievable by improving: 1) the representation ofland surface processes, 2) the initialization of soilproperties, and 3) the specification of various veg-etation characteristics by combining modeling, newremote sensing capabilities, and data-assimilationtechniques. It is, however, a more daunting chal-lenge to incorporate the complex mesoscaleinteractions among the land surface, the boundarylayer, and clouds, in large-scale climate models.These interactions play a critical role in determin-ing the timing and location of convection initiation,and the intensity of precipitation, but operate onsubgrid scales. Our understanding of these compli-cated interactions is still primitive, but a coordinatedresearch program, within GEWEX, on coupled land-atmosphere modeling and comparison of simulationswith field data will be a major step forward.

ReferencesChen, F., T. Warner, and K. Manning, 2001. Sensitivity oforographic moist convection to landscape variability: A Study ofthe Buffalo Creek, Colorado, flash-flood case of 1996. J. Atmos.Sci., 58, 3204-3223.

Trier, S., F. Chen, and K. Manning, 2004. A study of convec-tion initiation in a mesoscale model using high-resolution landsurface initial conditions. Mon. Wea. Rev., in press.

THE ROLE OF FINE-SCALELANDSCAPE AND SOIL MOISTURE

(Continued from page 9)

30–31 August 2004—GAPP PRINCIPAL INVESTIGA-TORS MEETING, Boulder, Colorado, USA.

13–15 September 2004—GSWP-2 SCIENCE WORK-SHOP, Kyoto, Japan.

13–16 September 2004—10TH MEETING OF THE GEWEXHYDROMETEOROLOGY PANEL, Montevideo, Uruguay.

15–17 September 2004—GLASS PANEL MEETING,Kyoto, Japan.

17–18 September 2004—CEOP MONSOON MEETING,Montevideo, Uruguay.

21–23 September 2004—GCSS SCIENCE PANEL MEET-ING, NASA GISS, New York, NY, USA.

23–25 September 2004—WCRP OFFICERS, CHAIRSAND DIRECTORS MEETING, Geneva, Switzerland.

11–15 October 2004—20TH SESSION OF THE CAS/JSCWGNE/8TH SESSION OF THE GMPP, Exeter, UK.

18–19 October 2004—GRP WORKING GROUP ONDATA MANAGEMENT AND ANALYSIS (WGDMA),Kyoto, Japan.

20–22 October 2004—15TH SESSION OF THE GEWEXRADIATION PANEL, Kyoto, Japan.

3–5 November 2004—IGWCO/GEWEX/UNESCOWORKSHOP ON TRENDS IN GLOBAL WATER CYCLEVARIABLES, Paris, France.

1–5 December 2004—GAME INTERNATIONAL SCIENCEPANEL MEETING AND 6TH INTERNATIONAL STUDYCONFERENCE ON GEWEX IN ASIA AND GAME, Kyoto,Japan.

20–24 June 2005—5TH INTERNATIONAL SCIENTIFICCONFERENCE ON THE GLOBAL ENERGY AND WA-TER CYCLE, Orange County, California, USA.

August 200416

THE ROLE OF FINE-SCALE LANDSCAPE AND SOIL MOISTURE VARIABILITY IN THEINITIATION OF DEEP CONVECTION

Simulated 3-h precipitation from 1500 to 1800 CST 19 June 1998, color coded at 4-mm intervals starting at0.5 mm, and 1500 CST 19 June surface water vapor mixing ratio in blue contours, with 3-g kg1 intervals startingat 6-g kg-1, for the MM5 3-km domain: (a) MM5 simulation initialized with HRLDAS soil moisture. (b) MM5simulation initialized with NCEP EDAS soil by moisture. (c) GOES-8 visible satellite imagery for 1645 CST 19 June1998. See article on page 9.

The Ranked Probability Skill Score for probabilistic streamflow forecasts for each forecast lead time and month,using synthesized ensembles from downscaled Numerical Weather Prediction (NWP) model output (MOS), and theclimatological control case in which historical data were used to construct the forecast ensemble (ESP). Resultsare shown for four hydrologically diverse river basins in the contiguous United States: (a) the Alapaha (southernGeorgia, 3626 km2), (b) the Animas (southwestern Colorado, 1792 km2), (c) the Cle Elum (Central Washington,526 km2), and (d) the east fork of the Carson (on the California/Nevada border, 922 km2). Note that downscaledNWP model output improves forecast skill in the Animas, Cle Elum, and the east fork of the Carson during thespring snowmelt period when streamflow in these basins is highest and most variable. See article on page 4.

ENSEMBLE STREAMFLOW FORECASTING IN SNOWMELTDOMINATED RIVER BASINS