NEWS NEWSWorld Climate Research Programme__WCRP

Vol. 12, No. 3 August 2002

Global Energy and Water Cycle Experiment

Percentage of precipitation over land that originatedas continental evaporation, annually averaged over 15years of simulation.

ANALYZING THE REGIONALHYDROLOGICAL CYCLE

CLIMATE FEEDBACKS AS A PROBLEMOF COMPLEX, NON-LINEAR DYNAMICS

William B. RossowNASA/Goddard Institute for Space Studies

Filipe AiresColumbia University

The response of the climate to a change inforcing (or an imposed perturbation from equi-librium) is determined by internal processesthat alter the relation between the forcing andthe response. These processes are called feed-back processes, in analogy with electrical circuittheory, where internal components interact to am-plify or damp the system's response to forcing.This concept was applied to early studies of theclimate represented by very simple, global energy-

(Continued on Page 3)

• Joint GRP/WCRP Climate Feedback Workshopset for November.

• CliC/GEWEX Workshop on Solid Precipitationsets strategy for the future.

CLIMATE FEEDBACK:CLOUDS COUPLE THE ENERGY

AND WATER CYCLE(IN A NONLINEAR PROCESS)

GEWEX CSE SOURCES OFPRECIPITATION USING GCM WATER

VAPOR TRACERS

Michael G. Bosilovich,1 Yogesh Sud,2

Siegfried D. Schubert,1 Gregory K. Walker,2

1NASA/Data Assimilation Office2NASA/Climate and Radiation Branch

Koster et al. (1986) used passive tracers to followincoming atmospheric water from surface evaporationthrough the atmosphere, until it was precipitated. Inthis way, the geographical source of water for allprecipitation could be identified. While these simula-tions were very coarse (8o x 10o) and short duration(only one season long), the work demonstrated a meth-odology of numerical calculation of the local and remotesources of precipitation within the model's simulation.

(Continued on Page 6)

What's New

• GLASS/PILPS-C1 Experiment linking net CO2,

LH and SH underway in the Netherlands.

• GEWEX-IAHS Workshop builds links to waterresource managers.

August 20022

COMMENTARY

PAUL F. TWITCHELLRETIRES FROM IGPO

Paul D. Try, Director, IGPO

Over the last 12 years, Paul Twitchell has beenassisting in developing and promoting GEWEX in awide variety of venues and hasbeen the driving force behindthe GEWEX Newsletter. Be-cause Paul has always beeninvolved in following the latestdevelopments in science andtakes on any activity with greatdedication and enthusiasm, I findit somewhat difficult to believethat he will actually be retiringand slowing down. Havingserved for many years as a program manager withthe Office of Naval Research (ONR), Paul has al-ways been excited about new research results andhas a history of supporting promising new research.Paul received his Ph.D. in Oceanography from theUniversity of Wisconsin and in addition to his servingwith ONR, he has been a professor at the NavalAcademy and also supported the US Air Force AirWeather Service as an Air Force Reservist, retiringas a Colonel.

So, with his third retirement (ONR, AF Reserve,and now from the International GEWEX Project Of-fice), Paul will be moving back to his home in Wellesley,Massachusetts, but has said he will still be availableto assist us in any way he can. I'm sure Paul willcontinue to follow developments within GEWEX, aswell as the other areas of research he has alwaysfollowed—polar lows, air-sea transfer, solar-terrestrialinteractions, and satellite remote sensing of the atmo-sphere and ocean. Within the IGPO planning activitieswe have always counted on Paul to keep us up todate with international research since Paul has devel-oped a remarkable number of contacts around theworld following his many years at ONR trackingthese activities.

I have known and worked with Paul for over 30years in various situations and have never know any-one with such a high degree of dedication and loyalty.This is hard to replace. Whatever the task, Paul hasalways been ready and willing to take it on, even ifit is not the most desirable activity. A true friend—we will miss his contributions.

Paul, best wishes to you and Eunice in your"latest" retirement.

ContentsPAGE

Climate Feedbacks As A Problem ofComplex, Non-Linear Dynamics 1

GEWEX CSE Sources Of Precipitation UsingGCM Water Vapor Tracers 1

Commentary - Paul F. Twitchell RetiresFrom IGPO 2

Editor's Farewell 2

Global Precipitation Climatology Project(GPCP) Products Available From The SurfaceReference Data Cemter 5

MAGS Special Issue of Atmosphere-Ocean 7

Coupling Chemistry And Physics In TheTerrestrial Biosphere: The PILPS-C1Experiment 8

WCRP Workshop on Determination of SolidPrecipitation in Cold Climate Regions 9

GEWEX - IAHS Workshop On The Applicationof GEWEX Scientific Research to WaterResources Management 10

Meetings Calendar 11

EDITOR'S FAREWELL

An activity in 1990 at the International GEWEXProject Office (IGPO) addressed methods of communi-cating the GEWEX scientific objectives, plans, and results.A product of those early discussions was the establish-ment of GEWEX News. The first issue was published inearly 1991 and a format developed for the November1991 issue that is not too different from this August 2002issue.

Now with over 11 years as editor, I will be movingon to other GEWEX related activities. I wish to thankall those who contributed science or news articles toGEWEX News, and particularly thank you for your patiencewith me. The production of GEWEX News has alwaysbeen a team effort, including guidance from the GEWEXScientific Steering Group, specifically Dr. Moustafa Chahineand Profesor Soroosh Sorooshian, and the Director ofIGPO, Dr. Paul Try. The team included for many yearsat IGPO Ms. Carmelitta Riley and now Ms. Erin McNamarafor typing, graphics, and layout assistance. In 1995Ms. Dawn Erlich added significantly in producing GEWEXNews in her role as Assistant Editor. The final excellentediting and supervision over printing was from the begin-ning accomplished by the Publications Division of Scienceand Technology Corporation (STC) under the professionalleadership of Ms. Diana McQuestion.

I am confident the IGPO/STC team will continue toproduce and improve GEWEX News.

Paul F. Twitchell, Ph.D., Editor

3August 2002

balance models. These simple models (see figurebelow) required that the state of the climate systembe represented by one state variable (annual, globalmean surface temperature, T

S) and be in near-

equilibrium with the solar heating (the external forcing,

F acts only on TS, even though it is a function of

TS and X

i, and that the X

i do not interact. These

are crucial and very strong simplifying assump-tions. Multiplying this expression by the systemgain,

and assuming for small time intervals and constantF = F

0 that

gives the familiar expression (cf. Peixoto and Oort,1992; Curry and Webster, 1998):

where

The classical feedback factors (Hansen et al., 1984)are then defined by

Note that each feedback factor is the product ofthree partial first derivatives (in this case). If thereare not too many variables and the relationshipsamong them are very simple (usually linear), thenthe effects of the feedbacks can be calculatedusing these expressions as is still commonly donein current climate studies. However, to obtain theabove expressions, several very strong assumptionsabout the climate had to be made: (1) a stricthierarchy of dependence is assumed (i.e., the forc-ing acts only on T

S, all the internal quantities are

functions of TS only and the internal quantities do

not interact), (2) the external forcing, F0, must be

constant, (3) the system gain, G, must be constant(i.e., a linear dynamical system), and (4) the inter-nal variables, X

i, are linear functions of T

S only.

An important consequence of the last two assump-tions is that the feedback factors are also constant.

If the dynamical system has many variablesand the relationships among them are more compli-cated than linear, such a simple analysis no longerworks. In particular, for the climate and our mod-els of climate, none of the assumptions mentionedare true, even approximately. For example, theoriginal energy balance models assumed that ter-

GF = �Ts

Hi�i

G�Ts =

1 - GF0

Hi�i

Gf = �i

fi =

F( t0 ����������0(t0 ������ + �i

�F(t0 + ��) �Xi(t0 + ��)

�Xi(t0 + ��) �Ts(t0)�Ts(t0)

Feedback loops in system in parallel: F is externalforcing, G is linear gain of system, T

S is surface tempera-

ture (or diagnosed variable). H and X represent feedbackcoefficients and internal variables, respectively.

CLIMATE FEEDBACKS (Continued from Page 1)

S0); and that one or more feedbacks involve internal

variables (Xi) that are functions of T

S only and are

independent of each other. Note also that the larg-est part of the climate's response to solar heating isan increase of temperature until the cooling by ter-restrial radiation balances the solar heating. Thus,the net radiative balance or small deviations from itis not strictly the forcing for climate since it in-cludes a major part of the climate response. Themost studied feedbacks are cloud- and ice/snow-induced changes of planetary albedo (altering theforcing) and cloud- and water vapor-induced changesof the atmospheric opacity to terrestrial radiation(altering the response).

If we consider the evolution of the planetaryenergy balance after a small perturbation from ra-diative equilibrium (or a small change in forcing),the change in the net radiative heating (solar heatingminus terrestrial cooling), F, because of the feed-backs at time, t

0 + 2∆t, can be expressed as

(1)where F

0 is the strictly external part of the forcing.

This expression is obtained only if we assume that

G =�F

�Ts

F Ts

sTXH ∂∂= /11

FTG s ∂∂= /

sTXH ∂∂= /33

sTXH ∂∂= /22

s

i

ii T

X

X

FH

∂∂

∂∂=

August 20024

restrial radiation changes were linear in TS (as-

sumption 3), but it has become common to calculatethe system gain without feedbacks using

which is "more physically accurate" but not con-stant as required for the usual calculation offeedbacks. In other words, the simple Planck rela-tionship for blackbody radiation already introducesa state-dependence to the climate's response: theclimate is not as sensitive to the same forcing at alltemperatures. Moreover, the actual situation weare confronted with is a climate that is changing inresponse to time-varying forcing (where human-induced changes are considered to be externalforcing), violating assumption (2). Finally, the pro-cesses involving clouds, at the least, couple theenergy and water cycle (see figure on front page)in ways that cannot be represented by independentand linear functions of T

S, violating assumptions (1)

and (4). The most important consequence is thatthe climate feedbacks are state-dependent, implyingboth time and location dependence: they cannot beestimated as if they are constant "global" values.

What approach can we use instead to un-derstand climate? Trying to answer this questionraises two others: what do we actually want orneed to know about the climate; and how dowe evaluate how well we know it?

Returning to an evaluation of how a general,multivariate, nonlinear dynamical system respondsto forcing suggests possible answers to these ques-tions by highlighting the fundamental role played bythe sensitivities (the partial first derivatives of eachvariable by the others) in the system's dynamics(remember that the classical feedback factors wereproducts of these sensitivities). If we examine, ingeneral, the changes at time t

0 + 2∆t in a multi-

variate system described by the state variables X(arranged as a vector, where bold symbols arevectors and matrices), then expression (1) becomes

(2)

where the approximation is that, for small ∆t, thechanges in all quantities are sufficiently accuratelyrepresented by the first derivatives (the sensitivities).

Expression (2) is much more complete than ex-pression (1) because the forcing, F, can act (inprinciple) on any of the state variables, X, andbecause all of the variables are (in principle) func-tions of all the others. The possibility of non-linearfunctional relationships among the state variablesalso means that the sensitivities are a function ofthe state of the system. Hence, the third term inexpression (2), which resembles the feedback termin expression (1), is not constant: the feedbackscan, themselves, be functions of the state of theclimate and vary in time (and with location). Thiscreates the main problem of trying to understandhow the climate works: since we can only observethe climate's behavior "in place" with all processesoperating and interacting, it is very difficult to de-termine the state-dependent relationships among allthe variables involved. We do not have the oppor-tunity to simplify the system by conductingexperiments, except in our climate models, whereone variable at a time can be perturbed and theresponse measured.

As a result, the answers to our second twoquestions could be that we want/need to deter-mine the sensitivities of the system and theirstate dependence; and that we can demonstratetheir accuracy by comparing how well we candetermine the (short-term) time-evolution ofthe system as a function of the climate's state.Returning to our first question: How can we deter-mine the sensitivities given observations of the climatestate at various times?

There are a number of powerful statistical analysistools that might be applied to the problem of deter-mining the climate sensitivities from observations,for example traditional autoregression models suchas Auto-Regressive Moving Average. However,the one we are investigating (Aires and Rossow,2002), neural networks, has three major advantages:it can represent very complex and nonlinear rela-tionships, the sensitivities of the variables and theirstate dependence can be determined from the neu-ral network in a straightforward way, and it canefficiently handle a very large number of variables.The analysis procedure has three steps: (1) theneural network is "trained" to represent all of thevalues X (t

0 + ∆t) given the values X (t

0), both

obtained from extensive observations of the climate's

�X(t0 ������

�X(t0 + ��)�X(t0 �������� F(t0 �������� F(t0 + ��) +

�X(t0 ������

�X(t0 + ��)

�X(t0 + ��)

�X(t0)F(t0)

( ) 34 41

sss

TTT

G σσ =∂

∂=−

5August 2002

time variations, (2) the accuracy of this representa-tion is tested on other observations not used for thetraining, and (3) the sensitivities are calculated fromthe parameters of the "trained" neural network. Es-sentially, the training of the neural network performsa multivariate, nonlinear regression fit to the obser-vational data set. Once trained, the parameters ofthe neural network can be used to calculate thesensitivities as a function of X (t).

Aires and Rossow (2002) illustrate the resultsfrom such an analysis approach by applying it toLorenz' simple general circulation model (Lorenz,1984). The sensitivities determined by the analysisof 200,000 observations of the model's time evolu-tion are compared to the values calculated directlyfrom the equations of the model. The neural net-work sensitivities are indistinguishable from the analyticones: quantitatively, the rms differences at eachtime are one to two orders of magnitude smallerthan the state-dependent variations of the sensitivi-ties (except for one that is zero). A crucial resultis that the state-dependence of the sensitivi-ties in the Lorenz system dynamics, which is asimple analog of the climate system, producestime-dependence of the sensitivities: in otherwords, the feedbacks in such nonlinear dynamicalsystems are time-dependent.

This analysis approach is attractive because italso naturally defines several properties of the dataset that are required for an accurate result: (1) thetime (space) resolution of the dataset must be highenough that the approximation used to obtain ex-pression (2) is accurate, i.e., the time (space) intervalhas to be small enough that all the variations in X(t) are represented accurately by only the firstderivatives, (2) the range of climate states coveredby the observations has to be statistically "com-plete", i.e., the data have to provide an adequatesample of all the states of the climate, all thepossible relations of X (t), and (3) the variablesincluded have to represent the complete state ofthe climate, otherwise the relations inferred by theanalysis will be distorted by the variations of the"hidden" variables.

The example discussed here is meant tosuggest a line of research; it is not yet a so-lution to our problem. Much more work is neededto investigate other approaches. For the neural net-work approach, research is needed to determinehow to relate the sensitivities to physical processes.In particular, dealing with the effects of an "incom-plete" observational prescription of X (t) is a major

obstacle. At a minimum, application of the sameanalysis procedure to climate model output, limitedto the same list of variables as available from ob-servations, provides a much more "dynamic" andgeneral way to compare the behavior of a model tothe real thing.

A GEWEX Radiation Panel/Working Group onCoupled Modelling Workshop on Climate Feedbackswill be held 18-20 November 2002 in Atlanta, Geor-gia.

References

Aires, F., and W.B. Rossow, 2002. Inferring instantaneous,mutlivariate and nonlinear sensitivities for the analysis of feed-back processes in a dynamical system: Lorenz model case study.Quart. J. Roy. Meteor. Soc., in press.

Curry, J.A., and P.J. Webster, 1998. Thermodynamics of At-mospheres and Oceans, Academic Press, London, 467 pp.

Hansen, J., A. Lacis, D. Rind, G. Russell, P. Stone, I. Fung, R.Ruedy and J. Lerner, 1984. Climate sensitivity: Analysis offeedback mechanisms, Climate Processes and Climate Sensitiv-ity, Maurice Ewing Series, No. 5, Amer. Geophys. Union,Washington, D.C., 120-163.

Lorenz, E., 1984. Irregularity: A fundamental property of theatmosphere. Tellus, 36A, 98-110.

Peixoto, J., and A.H. Oort, 1992. Physics of Climate, AmericanInstitute of Physics, New York, 520 pp.

GLOBAL PRECIPITATIONCLIMATOLOGY PROJECT (GPCP)

PRODUCTS AVAILABLE FROM THESURFACE REFERENCE DATA CENTER

In 1987, the Surface Radiation Data Center(SRDC) was established under GPCP at the NationalClimatic Data Center in Boulder, Colorado. In 1998,the SRDC was transferred to the EnvironmentalVerification and Analysis Center (EVAC) at theUniversity of Oklahoma, 710 ASP Avenue, Suite 8,Norman, Oklahoma, 72009. One of the tasks of theSRDC is to collect surface measurements of precipi-tation on a variety of temporal and spatial scales anddevelop products that can then be used by the"satellite community" to verify their precipitationalgorithms.

SRDC products are available to any researcherand can be downloaded at http://www.evac.ou.edu/srdc. Currently online are the up-to-date Pacificraingauge data, a digitized version of Taylor's rainfallatlas which contains Pacific island gauge prior to1971. For the Tropical Rainfall Measuring Mission(TRMM) ground validation see http://trmm-fc.gsfc.nasa.gov/trmm_gv/index.html.

August 20026

In this methodology, each source requires a new three-dimensional prognostic array in the General CirculationModel (GCM), which is often not feasible with limitedcomputational resources. Since that study, the tracer meth-odology has seen limited use (Druyan and Koster, 1989;Numagati, 1999). In recent years, there has been in-creasing focus on the atmospheric water cycle, especiallywith respect to the intensity and climate change of theregional water cycle (Morel, 2001). The water tracersprovide a diagnostic link between evaporation, pre-cipitation, moisture transport and the timescale thatwater resides in the atmosphere.

Recently, we have adapted the passive tracer meth-odology to the NASA Data Assimilation Office (DAO)Finite Volume GCM (FVGCM) to simulate the move-ment of regional sources of water (following Koster etal., 1986; and documented by Bosilovich and Schubert,2002, in the NASA GEOS GCM). These passive tracersare termed Water Vapor Tracers (WVT) because theysimulate the model's water vapor prognostic variable atthe model time step. The model dynamics and physicscompute tendencies for the WVT in proportion to themodel's water vapor. While the WVTs evolve accordingto the model dynamics and physical parameterizations,they are entirely passive, in that they do not affect thesimulated hydrological cycle. Evaporation within a limitedregion is used as the source for a WVT. The bottomfigure on page 12 identifies 12 large-scale source re-gions. Each color-coded region represents a continentalor oceanic source of water, in the form of evaporation,to the atmosphere. Following Bosilovich and Schubert(2002), we can diagnose the amount and location ofprecipitation that falls because of evaporation from eachregion.

The FVGCM uses semi-Lagrangian advection that isparticularly useful for tracer calculation (Lin and Rood,1996). The model uses the NCAR CCM3 physicalparameterizations. We have run the FVGCM at 1o x1.25o resolution for 15 years using real time varying Sea

Surface Temperatures (SST) from 1986–2000. In thispaper, we present the simulation of large-scale con-tinental and oceanic sources of water for precipitationin GEWEX Continental Scale Experiments (CSE).The area of each CSE is defined identically to Roads etal. (2002) (see bottom figure on page 12). The tablebelow shows the annual contribution to each CSE pre-cipitation from the large-scale geographical region'sevaporation. An exact estimate of precipitation recyclingcannot be identified in this table because the sourceregions are larger than the CSEs. However, in all casesthe local continental source of water is a majorcontributor to the precipitation. Some CSEs are rela-tively simple to understand, such as the Mackenzie RiverBasin (GEWEX Mackenzie Study [MAGS]) where wa-ter comes either from the North American continent orfrom the Pacific Ocean. The Baltic Sea Experiment(BALTEX) seems to be more complicated with manysources of water contributing to precipitation, includingEuropean continental, North and Tropical Atlantic Oce-anic and Polar (note that the Mediterranean Sea is includedin Polar WVTs for convenience). The GEWEX AsianMonsoon Experiment (GAME)-Tibet and GAME-Siberiaare the only two CSEs where the continental sources ofwater exceed oceanic sources. The figure on the right onpage 1 shows the percentage of precipitation from con-tinental sources over land. In general, coastal regionsshow less continental precipitation, especially where wewould expect on-shore flow from the oceans. Whenaveraged over all land points, 1.08 mm/day of precipita-tion from continental sources falls on land, while 1.50mm/day precipitation from oceanic sources falls on land.

While the annual budgets of the moisture sources areuseful, mean annual cycles can describe the seasonalvariations of the moisture transport. On the back page,the figure shows the mean annual cycles of the majorsources of each CSE. In the Mississippi River (GEWEXContinental International Project [GCIP]) and MacKenzieRiver basins, the North American continental sourcedominates during summertime, while in winter the PacificOcean source dominates. Likewise, BALTEX shows atransition from the local continental sources in summer to

Percentage of precipitation that occurs in each GEWEX CSE from each of the source regions shown on page 12. Thesum of continental and oceanic sources is included for convenience. CSE annual mean precipitation is included in mm/day. Boldface indicates values greater than 10%. The percentages are computed from time averaged WVT precipitationdivided by time averaged total precipitation.

GEWEX CSE SOURCES OF PRECIPITATION(Continued from Page 1)

7August 2002

oceanic sources in winter. With its maximum Mississippicontribution in late summer, the tropical Atlantic Oceanhas a smaller annual cycle than the Pacific Ocean.However, summertime precipitation is larger than winter,so the tropical Atlantic has a larger impact on the annualbudget than the Pacific Ocean (Table on page 6). Con-tinental sources for Amazonian precipitation are largethroughout the year, but the oceanic sources vary withthe seasonal change of the easterly flow. In the GAMETropics region, the moisture sources shift from the Pa-cific Ocean to the Indian Ocean. GAME-Siberia is largelydominated by the summertime Asia continental sources.

The figure on the back page suggests that insome regions, significant amounts of water are trans-ported very long distances. For example, the Asiacontinental source for the Mackenzie basin in summer,and the North American source for BALTEX musttraverse entire oceans. While a map of moisture trans-port may suggest the possibility that these are potentialsources, the tracers provide a quantitative diagnostic.This also raises the question of residence time ofthe atmospheric water in the GCM. To evaluate theresidence time, we initialized a special WVT equal to theinitial atmospheric water content, but provide no sourceat the surface. This allows the precipitation to deplete theWVT atmospheric water content in time without beingreplenished. Again, this WVT is a diagnostic and thesimulated precipitation and hydrologic cycle continue nor-mally. The figure in the next column shows the timeseries of the WVT water content as it is depleted.Fitting the data points with an exponential curveshows that the e-folding time of the atmosphericwater is 9.2 days. The more traditional way of deter-mining the residence time of water in the atmosphereuses a time constant equal to atmospheric water overprecipitation (Trenberth, 1998). For this period, the globalmean precipitation is 3.33 mm/day and the global meantotal precipitable water is 24.9 mm, which suggests amoisture depletion estimate (e-folding time of the watercontent) of 7.5 days. Trenberth (1998) points out that thisestimated value of depletion is quite sensitive and ne-glects moisture transport (inherently included in the WVTestimate).

The WVTs provide a diagnostic tool to evaluate thehydrologic cycle in atmospheric numerical models. Thediagnostic considers the instantaneous evaporationand precipitation rates as well as transport pro-cesses. Such diagnostics should be useful in evaluatingthe water cycle of extreme conditions such as flood anddrought, as well as the intensity of regional water cyclesin climate change experiments. Of course, the quality ofthe WVT diagnostics depends on the veracity of theGCM simulation. At present, we are implementing theWVT diagnostics in the NASA DAO Data AssimilationSystem to evaluate real data case studies and the impactof water vapor assimilation on the hydrologic cycle. Thesediagnostics may be useful in other studies, such as syn-

optic meteorology, mesoscale meteorology and paleocli-matology. It may also be possible to validate the WVTdiagnostic data with precipitation isotope data. (This workis supported in part under the GEWEX Americas Pre-cipitation Project (GAPP), Pan-American Climate StudyWarm Season Precipitation Initiative, and the NASAGlobal Water and Energy Cycle Program.)References

Bosilovich, M.G., and S.D. Schubert, 2002. Water vapor tracers asdiagnostics of the regional hydrologic cycle, J. Hydromet., 3, 149-165.

Druyan, L.M., and R.D. Koster, 1989. Sources of Sahel precipitation forsimulated drought and rainy seasons. J. Climate, 2, 1438-1446.

Koster, R.D., J. Jouzel, R. Suozzo, G. Russell, W. Broecker, D. Rind, andP. Eagleson, 1986. Global sources of local precipitation as determinedby the NASA/GISS GCM. Geophys. Res. Let., 13, 121-124.

Lin, S.-J., and R.B. Rood, 1996. Multidimensional flux form semi-lagrangian transport schemes. Mon. Wea. Rev., 124, 2046-2070.

Morel, P., 2001. Why GEWEX? The agenda for a global energy andwater cycle research program. GEWEX News, Vol. 11., No. 1.

Numagati, A., 1999. Origin and recycling processes of precipitatingwater over the Eurasian continent: Experiments using an atmosphericgeneral circulation model. J. Geophys. Res., 104, 1957-1972.

Roads, J., M. Kanamitsu and R. Stewart, 2002. CSE Water and EnergyBudgets in the NCEP-DOE Reanalysis II. J. Hydromet., 3, 227-248.

Trenberth, K.E., 1998. Atmospheric moisture residence times and cy-cling: Implications for rainfall rates and climate change. Climatic Change,39, 667-694.

Model simulated globally averaged WVT with no evapo-rative sources (daily average) divided by the initial water(dots) and the exponential fit of the model data (line).This indicates an average global residence time of 9.2days from this simulation.

MAGS SPECIAL ISSUE OFATMOSPHERE-OCEAN

The Canadian Meteorological and Oceanographic Society(CMOS) journal, Atmosphere-Ocean, has published a Mack-enzie GEWEX Study (MAGS) special issue with a collectionof papers that address key aspects of the water and energyfeatures of the Mackenzie River basin. This special issueis concerned with the 1994/95 water year, a period charac-terized by one of, if not the lowest, discharge from theMackenzie River into the Arctic Ocean on record. Copies ofthis issue (Volume 40, No. 2, June 2002) may be obtainedthrough the CMOS web site at: http://www.meds-sdmm.dfo-mpo.gc.ca/cmos/pubs.html or from Dr. Ronald Stewart,Climate Processes and Earth Observation Division, Meteo-rological Service of Canada, 4905 Dufferin Street, Downsview,Ontario M3H 5T4 Canada, E-mail: Ron.Stewart@ec.gc.ca.

August 20028

COUPLING CHEMISTRY AND PHYSICSIN THE TERRESTRIAL BIOSPHERE:

THE PILPS-C1 EXPERIMENT

Nicolas ViovyFrench Atomic Energy Commission

In the past, models of the land surface have beendeveloped for two different purposes: as boundary con-ditions for the atmosphere (with an emphasis on biophysicalprocesses), and for a better understanding of the carboncycle (with an emphasis on biochemical processes). Theseseparate developments have lead to relatively indepen-dent intercomparison projects, in the framework of WCRP(e.g. Project for Intercomparison of Land Surface Pa-rameterization Schemes (PILPS), Henderson-Sellers etal., 1993, 1995) and International Global-Biosphere Pro-gram programs, (e.g. Potsdam model intercomparison,Cramer et al., 1999; EMDI, Olson et al., 2001).

But it is now established that geochemical and physi-cal processes are highly coupled at the land surface, ontime scales of seconds to centuries. For instance, it hasbeen demonstrated that the feedback of atmosphericCO

2 concentration on stomatal conductance induces

changes in the hydrologic cycle, with importantconsequences for the climate, and vice versa (Sell-ers et al., 1996; Betts et al., 1997). The recent evolutionof land surface representations, for the study of both thecarbon cycle and climate, have lead to improvements inthe coupling between biophysical and biogeochemicalprocesses. At the same time, since 1995 several siteshave been instrumented for the continuous measurementof fluxes of both net CO

2 and energy. Thus, we are able

now to go further in our understanding of the couplingbetween the CO

2 and water cycles, by comparing simu-

lated fluxes with in situ data collected in the frame offluxnet project.

The PILPS-C1 experiment, an initiative of theGEWEX/Global Land-Atmosphere Systems Study(GLASS), was launched in May 2002, with the aimof comparing land surface model fluxes of net CO2,latent and sensible heat with in situ measurementsat a selected site. This site, called Loobos, is a part ofthe Euroflux network. Loobos is a 80-year-old coniferousforest located in the center of The Netherlands. Fluxesof CO

2, latent and sensible heat have been measured

there continuously for the years 1997 and 1998 using theeddy correlation method. The flux tower is 27 meters inheight, and measurements are made every 30 minutes.

In the comparison of in situ net CO2 fluxes with

model simulations, one major problem concerns soil car-bon spinup. In fact, a part of the CO

2 net flux arises

from decomposition of soil organic matter which can bemore than one century old. Thus, a valid comparisonbetween the simulated and observed CO

2 fluxes requires

that the model be initialized over a period of severalcenturies. This in turn requires a knowledge of both

climate and land use during the spinup period. Suchknowledge is generally not available. An interesting fea-ture of the Loobos site is that the forest was planted 80years ago on sand (i.e., on soil containing no organicmatter), meaning that the initial condition of the soilcarbon pool is known. Moreover, near the Loobos sitethere is a meteorological station which has been record-ing data since the beginning of the 20th century. Thus,Loobos offers a unique opportunity for testing modelsimulations of the carbon cycle. Can our models simulatethe carbon sink observed on this site?

As explained previously, different types of land sur-face schemes have been developed, some more orientedtoward biophysical, others toward biogeochemical pro-cesses. The carbon-type intercomparisons and biophysicalintercomparisons have largely happened independently andinteractions between these groups have been limited.PILPS-C1 offers the first opportunity to bring thesegroups together using carbon. As a result, participation inthe project is open to a wide scientific community, includ-ing those working with traditional land surface schemes,and workers utilizing more carbon-oriented models.

In summary, the two main scientific questions ad-dressed by this project are: (1) What is the ability ofmodels of different types to correctly reproduce bothbiophysical and biogeochemical processes; and (2) Tak-ing into account the long-term history of the site, are themodels able to reproduce the observed sink of carbon?

The 2002 timeline for the project is as follows:(1) Submission of the forcing data in June to the partici-pants, (2) Deadline for submission of results in Augustand (3) Convene a workshop for analysis of preliminaryresults in November.

As of now, nine groups are participating in PILPS-C1, but new groups are encouraged to join the project.

ReferencesBetts R.A., P. Cox, S.E. Lee, F.I. Woodward, 1997. Contrasting physi-ological and structural vegetation feedbacks in climate change simulations.Nature, 387, 796-799.

Cramer W., D.W. Kicklighter, A. Bondeau, B. Moore, III, G. Churkina,B. Nemry, A. Ruimy, A.L. Schloss, & Participants of "Potsdam '95",1999. Comparing global models of terrestrial net primary productivity(NPP): Overview and key results. Global Change Biology, 5, (Suppl.1).

Dolman A.J., E.J. Moors, J.A. Elbers, 2002. The carbon uptake of a midlatitude pine forest growing on sandy soil. Agricultural and ForestMeteorology, 11, 157-170.

Henderson-Sellers A., Z. Yang, R. Dickinson, 1993. The Project forIntercomparison of Land Surface Parameterisation Schemes. Bull.Amer. Met. Soc., 74, 1335-1349.

Henderson-Sellers A., A. Pitman, P. Love, P Irannejad., T. Chen,1995. The project for Intercomparison of Land Surface Parameter-ization Schemes (PILPS): Phase 2 and 3., Bull. Amer. Met. Soc. 76:489-503.

Olson R.J., J.M.O. Scurlock, S.D. Prince, D.L. Zheng, and K.R.Johnson (eds.) 2001. NPP Multi-Biome: NPP and Driver Data forEcosystem Model-Data Intercomparison.

Sellers P., L. Bounoua, G. Collatz, D. Randall, D. Dalich, S.L. Berry,I. Fung, C. Tucker, C. Filed, T. Jensen 1996. Comparison of radia-tive and physiological effects of doubled atmospheric CO

2 on climate.

Science, 271, 1403-1406.

9August 2002

WCRP WORKSHOP ONDETERMINATION OF SOLID

PRECIPITATION IN COLDCLIMATE REGIONS

Fairbanks, Alaska, USA9–14 June 2002

Paul LouieMeteorological Service of Canada

WORKSHOP/MEETING SUMMARIES

Over 50 invited scientists representing 13 countriesparticipated in this international workshop organized byCliC and GEWEX and hosted by the University of AlaskaFairbanks. The workshop was sponsored by: WMO/WCRPand WMO/Global Climate Observing System programmes;Meteorological Service of Canada; NOAA Arctic Re-search Office and NOAA/NASA GEWEX AmericasPrediction Project; and the International Arctic ResearchCentre and the Water and Environmental Research Cen-tre both at the University of Alaska Fairbanks.

The objectives of this workshop were:

• Review the current status of measuring ordetermining precipitation, especially solid pre-cipitation, in cold climate regions;

• Identify gaps and issues; and

• Recommend actions that will allow us to deter-mine precipitation over a range of time andspace scales for climatological and hydrologicalanalyses, regional water budgets, validation andprocess experiments and models.

Invitees brought a wide range of expertise in coldclimate precipitation, representing both in situ measure-ment and remote sensing techniques; the development ofprecipitation adjustment techniques and implementation ofthem on regional and global scales; major field programs;global data archives; and the modeling community. Therewas representation of the major cold regions of the worldincluding North and South America, Scandinavia, Eurasia,China and the Arctic and Antarctic.

The workshop was organized in five sessions whichincluded both oral and poster presentations. The sessiontopics were: 1) Precipitation Measurement, 2) Measure-ment Errors and Adjustment Procedures, 3) PrecipitationData for Major Projects, 4) Regional and Global Precipi-tation Analysis, and 5) Precipitation Modeling and ModelValidation. A short special session was convened topresent related information on Major Projects. Theseincluded presentations on the research and observationactivities of GEWEX and a very comprehensive descrip-

tion of the scientific agenda of the Global PrecipitationMeasurement (GPM) Mission.

The workshop programand the abstracts for all thepresentations are availableat: http://acsys.npolar.no/meetings/precip/ws.htm.The proceedings for theworkshop will be publishedby WMO/WCRP by the endof 2002.

Three breakout sessionswere organized to discuss more

fully the topics of: Precipitation Measurement; Measure-ment Errors and Adjustments; and Global PrecipitationData Sets. The groups were tasked to identify gaps andissues and to recommend actions.

The working group on Precipitation Measure-ment had three sub groups to address ConventionalMeasurement Methods, Alternative Strategies andNew Technologies. Some of the common issuesidentified by these sub working groups were:

• The impact of automation on precipitationmeasurement and related challenges; and

• The need to blend (fuse or combine) data fromdifferent sources (in situ, model, satellite)

Some of the common recommendations included:

• Establish a WCRP working group to developguidelines on the minimum station densityrequired for climate research studies on solidprecipitation in cold climate regions; and

• Conduct urgently needed research to deter-mine how to obtain climate quality data fromautomated weather observing systems—needto define and attribute "climate"-quality tooperational weather observing systems/sites.

The working group on Measurement Errors andAdjustment Procedures addressed several questions,including:

• What are the errors in gridding and creatinggridded products?

• What must be done to provide consistentadjustments to facilitate comparisons amongnational, regional and global climatologies?

Findings from this group included:

• There is real value in reporting adjusted pre-cipitation and there is a continuing need forongoing intercomparisons.

Barry Goodison openingWCRP CliC/GEWEXworkshop.

August 200210

The working group dealing with Global PrecipitationData Sets addressed such questions as:

• What research and development is needed to pre-pare for these new measurement and data systems(for example, the Satellite Global Precipitation Mea-surement Mission);

• What are the limitations in merging satellite and insitu products; and

• Is there a role for a data-model mix in producingglobal precipitation data products?

Some of the research needs identified by thisgroup included:

• Development of models for downscaling to topog-raphy, blending strategies for data assimilation (e.g.define error characteristics), and interpolation todetermine large-scale precipitation patterns in theabsence of data.

Some recommendations put forward by this group:• Develop a strategy for exploiting new technolo-

gies in the development of algorithms and modelsfor third and fourth generation precipitationclimatologies; and

• Use daily precipitation as a building block forprecipitation climatologies.

The closing session provided an opportunity for fur-ther discussion of issues by participants. The CliCImplementation Plan (available at the ACSYS/CliC website,http://clic.npolar.no) had identified issues, needs and pro-posed actions and those identified by GEWEX GHPwere reviewed in the context of the workshop discussion.This workshop provided a good start at tackling theseissues. Some issues that were not fully covered or still inneed of resolution were also identified. Examples include:

• Precipitation measurement in mountainous regionsand Antartica;

• Solid precipitation over sea ice and ice coveredArctic Ocean;

• How to improve the linkage to the GCM commu-nity and determining modeling needs.

After three and half days of presentations andintense discussions, there were several afternoon/eveningfield trips to local research sites. On the followingday, about half of the participants traveled to Barrowto visit the Climate Monitoring and DiagnosticsLaboratory, which included the University of Alaska-Fairbanks – Japan Frontier Institute snowfall/blowingsnow observation site; the U.S. Department of EnergyAtmospheric Radiation Measurement (ARM) site; atour of the Barrow Arctic Science Consortium facilities;the Weather Office at Barrow; and a tour of theAutomatic Weather Observation System site at Bar-row Airport.

The Workshop was held in conjunction with the 3rdInternational Conference on Water Resources and Envi-ronment Research (ICWRER) at the Dresden Universityof Technology. It was organized by the GEWEX WaterResources Applications Project (WRAP) Committee andthe International Association of Hydrological Sciences(IAHS)/World Meteorological Organization (WMO) WorkingGroup on GEWEX. The broad objective of the Workshopwas to initiate a dialogue with water managers on theirneeds and the GEWEX data/model products that areavailable to address those needs. This was intended toidentify useful forecast/modeling products for water man-agers; understand how these are used in decision-making;and determine preferred product delivery mechanisms

The Workshop program consisted of three keynotepresentations, six presentations on water resource appli-cations in GEWEX Continental Scale Experiments (CSE)and a panel discussion with the keynote speakers fromthe Workshop and the ICWRER Conference. The key-note presentations were:

• GEWEX and Water Resources Applications ProjectOverview: Lawrence Martz, Chair WRAP.

• Relevance of Predictions (short-, medium-, long-term) of Water Availability for Water ResourcesManagement: Martin Kaupe and Mathias Schmitt,Wassergewinnung (W), GEW RheinEnergie AG,Germany.

• Current Research on the Application of Me-teorological Forecasts to Water ResourcesManagement: John Schaake, Jr., NOAA NationalWeather Service, USA.

The CSE presentations were:

• Toward Water Resources Assessment and Man-agement in Thailand with GAME-T Datasets:Shinjiro Kanae (GAME).

• GEWEX Related Water Resources Applicationsin the Baltic Sea Drainage Basin – Contributionfrom BALTEX: Sten Bergström (BALTEX).

GEWEX – IAHS WORKSHOP ONTHE APPLICATION OF GEWEX

SCIENTIFIC RESEARCH TO WATERRESOURCES MANAGEMENT

Dresden, Germany22–26 July 2002

Lawrence Martz1 and Alan Hall2

1University of Saskatchewan, Canada2IAHS, Australia

11August 2002

• The Value of Seasonal Climate Forecasts forReservoir Management: Aris Georgakakos (GAPP).

• Application of Seasonal Climate Forecasts toWater Management in the Tennessee River:Ruby Leung (GAPP).

• Hydrometeorological data availability issues inwest Africa: a challenge to understand the AfricanMonsoon): Christian Depraetere (CATCH).

• Restoring Ice-Jam Floodwater to a Drying DeltaEcosystem, Terry Prowse: presented by LawrenceMartz (MAGS).

The Panel Discussion component of the Workshopinvolved the Workshop keynote speakers (with Gert Schultzsubstituting for Kaupe and Schmitt) and keynote speakersfrom the ICWRER Conference—Uri Shamir (Israel),PeteLoucks (USA), and Keith Hipel (Canada). The paneladdressed wide-ranging issues:

• The uncertainty of forecasts and the need formeasures to describe this uncertainty.

• Integration of hydrological and climate models toreduce complexity and model parameterizationrequirements.

• Water managers find it difficult to respond to highlevels of uncertainty.

• As water becomes more scarce, reliable watersupply together with the opposite problem of floodprotection are seen as key issues.

• The potential value of the Global Soil Wetnessproducts and the Prediction of Ungauged Basins(PUBS) initiative.

• The value of hydrological observations to theCoordinated Enhanced Observing Period (CEOP).

• One day workshops with water managers andwater scientists at regional/local/basin scale by theCSEs as a way of developing greater use ofGEWEX products.

The Dresden Workshop presentations and discus-sions will be reviewed at the upcoming WRAP meetingin New York on September 9, 2002. They will also serveas useful base for the second WRAP Workshop plannedfor the IUGG General Assembly in Sapporo in July 2003(JWH02 - The Role of GEWEX Hydrological Science inImproved Water Resources Management).

GEWEX/WCRP MEETINGSCALENDAR

For calendar updates, see the GEWEX Web site:http://www.gewex.org

2–6 September 2002—WMO/WWRP INTERNATIONALCONFERENCE ON QUANTITATIVE PRECIPITATIONFORECASTING, Reading, UK.

9–13 September 2002—8TH SESSION OF THE GEWEXHYDROMETEOROLOGY PANEL AND ASSOCIATED MEETINGS(WRAP, WEBS, CEOP), IRI, Palisades, New York, USA.

30 September – 2 October 2002—GSWP-2 KICKOFF WORKSHOP,COLA, Calverton, Maryland, USA.

1–3 October 2002—GOES USERS’ CONFERENCE II, NIST, Boul-der, Colorado, USA.

2–4 October 2002—GLASS PANEL MEETING, COLA, Calverton,Maryland, USA.

8–9 October 2002—ISLSCP FUTURE STRATEGY AND INITIATIVEIII MEETING, Washington, DC, USA.

9–10 October 2002—CEOP SATELLITE DATA INTEGRATIONMEETING, Tokyo, Japan.

10–19 October 2002—34TH COSPAR SCIENTIFIC ASSEMBLY (SpecialSession on Properties of the Earth-Atmosphere-Ocean System as Inferredfrom the New Generation of Earth Science Satellites), Houston, Texas,USA.

6–10 November 2002—8TH ANNUAL MAGS MEETING ANDSCIENCE COMMITTEE MEETING, Jasper, Canada.

12–15 November 2002—2ND INTERNATIONAL ATMOSPHERICMODEL INTERCOMPARISON PROJECT (AMIP) CONFERENCE,Meteo-France, Toulouse, France.

18–20 November 2002—GRP/WGCM WORKSHOP ON CLIMATEFEEDBACKS, Atlanta, Georgia, USA.

18–22 November 2002—WGNE/GMPP MEETING, Météo-France,Toulouse, France.

6–10 December 2002—AGU FALL MEETING, San Francisco, California,USA. Special theme session on research in climate and hydrology inthe Southern Hemisphere.

15–17 January 2003—US-JAPAN WORKSHOP ON CLIMATECHANGE, Irvine, California, USA.

20–25 January 2003—15TH SESSION OF THE GEWEX SSG, Bangkok,Thailand.

9–13 February 2003—83RD AMERICAN METEOROLOGICALSOCIETY ANNUAL MEETING, Long Beach, California, USA.

17–21 March 2003—WCRP JOINT SCIENTIFIC COMMITTEEMEETING, Reading, UK.

2–4 April 2003—CEOP SECOND FORMAL INTERNATIONALIMPLEMENTATION PLANNING MEETING, Berlin, Germany.

30 June – 11 July 2003—XXIII GENERAL ASSEMBLY OFTHE INTERNATIONAL UNION OF GEODESY AND GEO-SCIENCES (IUGG), Sappora, Japan.

GEWEX NEWSPublished by the International GEWEX Project Office (IGPO)

Dr. Paul D. Try, DirectorEditor: Dr. Paul F. TwitchellMail: International GEWEX Project Office1010 Wayne Avenue, Suite 450Silver Spring, MD 20910, USA

WWW Site: http://www.gewex.org

Assistant Editor: Dawn P. ErlichTel: (301) 565-8345Fax: (301) 565-8279E-mail: gewex@gewex.org

August 200212

Mean annual cycle of the dominant sources of water that occurred as precipitation in each of the CSEs. Colorscorrespond to the geographical regions as shown below.

LINKING CLOUD FEEDBACK, WATER VAPOR SOURCES, AND TERRESTRIALBIOSPHERE TO THE ENERGY AND WATER CYCLE (SEE ARTICLES ON PAGES 1 AND 8)

The source regions of water vapor forprecipitation from continental and oce-anic regions. NA, North America; SA, SouthAmerica; Eur, Europe; Afr, Africa; Asa, Asia-Australia; Npa, North Pacific; Spa, SouthPacific; Nat, North Atlantic; Tat, TropicalAtlantic; InO, Indian Ocean; and Pol, Polar(both north and south are also included).Figure above and to left are referenced inarticle "GEWEX CSE Sources Of Precipita-tion Using GCM Water Vapor Tracers" be-ginning on page 1.