Gerald M. Stokes* and The Atmospheric Radiation Stephen E. Schwartz+ Measurement (ARM) Program: Programmatic Background and Design of the Cloud and Radiation Test Bed

Abstract ence of clouds and the role of cloud radiative feed­ back. The United States Global Change Research The Atmospheric Radiation Measurement (ARM) Program, sup­ Program (USGCRP) (CEES 1990) identified the sci­ ported by the U.S. Department of Energy, is a major new program entific issues surrounding climate and hydrological of atmospheric measurement and modeling. The program is in­ systems as its highest priority concern. Among those tended to improve the understanding of processes that affect atmospheric radiation and the description of these processes in issues, the USGCRP also identified the role of clouds climate models. An accurate description of atmospheric radiation as the top priority research area. ARM, a major activity and its interaction with clouds and cloud processes is necessary to within the USGCRP, is designed to meet these re­ improve the performance of and confidence in models used to study search needs and is an outgrowth and direct continu­ and predict climate change. The ARM Program will employ five (this ation of the Department of Energy's (DOE's) decade­ paper was prepared priorto a decision to limit the numberof primary measurement sites to three) highly instrumented primary measure­ long effort to improve general circulation models ment sites for up to 10 years at land and ocean locations, from the (GCMs) and other climate models, providing reliable Tropics to the Arctic, and will conduct observations for shorter simulations of regional and long-term climate change periods at additional sites and in specialized campaigns. Quantities in response to increasing greenhouse gases. to be measured at these sites include longwave and shortwave Planning for the ARM Program began in the fall of radiation, the spatial and temporal distribution of clouds, water vapor, temperature, and other radiation-influencing quantities. There 1989; a description of the initial program is presented will be further observations of meteorological variables that influ­ in the ARM Program Plan (U.S. Department of Energy ence these quantities, including wind velocity, precipitation rate, 1990). Since its publication, the scientific issues have surface moisture, temperature, and fluxes of sensible and latent been further delineated, and there has been substan­ heat. These data will be used for the prospective testing of models tial progress in designing and implementing the facili­ of varying complexity, ranging from detailed process models to the highly parameterized description of these processes for use in ties necessary to conduct the measurements required general circulation models of the earth's atmosphere. This article to meet the scientific objectives of the program. This reviews the scientific background of the ARM Program, describes paper outlines the scientific background for the ARM the design of the program, and presents its status and plans. Program, the objectives of the program, how the design of the field facilities respond to these objec­ tives, and the current status of the program. 1. Introduction

The Atmospheric Radiation Measurement (ARM) 2. Background Program is a major program of atmospheric measure­ ment and modeling intended to improve understand­ A variety of models have been developed to simu­ ing of the processes and properties that affect atmo­ late the physical processes that occur in the earth's spheric radiation, with a particular focus on the influ- atmosphere as part of attempts to understand climate and possible climate changes due to human activity. 'Global Studies Program, Pacific Northwest Laboratory, Richland, One such class of models, GCMs (U.S. Department of Washington. Energy 1985; Simmons and Bengtsson 1988; Cubasch +Department of Applied Science, Brookhaven National Laboratory, and Cess 1990; Randall 1992), may be viewed as part Upton, New York. of a hierarchy of models and modeling systems, Correspondingauthoraddress: Dr. Gerald M. Stokes, Pacific North­ ranging from detailed process models to multireservoir west Laboratories, Battelle Boulevard, P.O. Box999, Richland, WA 99352. climate models (e.g., Hoffert et al. 1980) to GCMs. In final form 10 March 1994. This hierarchy is characterized by a spectrum of ©1994 American Meteorological Society complexity of the description of the physical pro-

Bulletin ofthe American Meteorological Society 1201 cesses being modeled and by a wide range in spatial, The completely empirical approach is confounded by temporal, and wavelength resolution. It attempts to the large number of types of situations for which consolidate large amounts of knowledge about many parameterizations must be developed. Also, the atmospheric processes to gain insight into the interac­ parameterizations may inadvertently incorporate fea­ tion among these processes on regional to global tures of the current climate into models of future scales and ultimately into the processes that are climates. On the other hand, the completely theoreti­ responsible for controlling the earth's climate. cal approach may be limited by the current state of An important feature of GCMs is how they capture theoretical understanding or may result in descriptions the interaction between the processes that control the of such complexity that it will be hard to incorporate vertical transport of energy and water and the large­ representations of the processes into large-scale cli­ scale circulation of the atmosphere. It is convenient to mate models. An important feature of many recent think about the treatment of the vertical transport attempts to parameterize atmospheric processes in processes in terms of grid cells. In this view, the climate prediction models is the emphasis on the use properties characterizing a region of space (a volume of physical models rather than empirical data as the or a surface area) are represented in the model by basis for the parameterization. values ascribed to a single point representing that grid Currently, two particularly important basic atmo­ cell. These properties are the various scalar and spheric processes requiring improved description in vector properties assigned to the cells-for example, climate models are 1) the transfer of radiation within temperature, water mixing ratio, wind velocity, or the atmosphere including the spectral dependence of pressure. Because a model can conveniently handle this radiation transfer, and 2) the processes respon­ only a limited number of grid cells, a single grid cell in sible for cloud formation, maintenance and dissipa­ the model must represent a considerable volume or tion, and the prediction of the associated cloud prop­ area of the planet. Many meteorological phenomena erties, particularly their radiative properties. Accurate important to weather and climate, as well as associ­ description of cloud processes in climate models is ated radiative processes, occur on scales that are too critical because of the large influence of clouds on the small to be resolved by the resulting grid mesh in earth radiation budget (Ramanathan et al. 1989) and current climate models. the possibility that cloud properties may change as There is concern over the ability of models to climate changes (Lindzen 1990). Both cloud and accurately represent phenomena that exhibit sub­ radiative processes present particularly significant stantial variability at scales finer than are resolved by challenges in attempting to parameterize them on the the grid spacing of the model. Therefore, a major subgrid scale. challenge to GCMs and related models is to capture Liquid water clouds exhibit negligible supersatura­ the essence of these subgrid phenomena in the tion, and clouds form when air containing water vapor model with reasonable computational simplicity and is cooled below its dewpoint. Therefore, modeling the without introducing systematic errors that manifest presence of clouds as a function of location is straight­ themselves in unrealistic global-scale phenomena. forward in a model whose grid-cell dimensions are The means of achieving this goal is through a process small relative to the dimensions of the clouds (e.g., that is often referred to as "parameterization" (Randall Clark 1979; Kogan 1991). The problem arises in 1989, 1992). translating this understanding to the description of Understanding the best approach to parameteriza­ cloud formation in a model whose grid-cell dimensions tion and the accuracy that can be expected for the may be more than 100 km on a side and 50 hPa in the various approaches is a major challenge of climate vertical, manytimes the dimensions of certain types of modeling. These approaches may be viewed as falling clouds. along a continuum. At one end ofthe continuum would In these models, the key issue for parameterization be a purely empirical approach: observing the phe­ becomes identification of the conditions under which nomena to be parameterized, empirically developing clouds form and the appropriate representation of the parameterizations, and testing these parameterizations properties of the resulting clouds that are important for by comparing model output with observations. At the other aspects ofthe model. These aspects include the other end of the continuum would be a theoretically associated latent heat release, the radiative proper­ based approach: acquiring an understanding of the ties of the clouds, and the development of precipita­ physics of the controlling processes, developing tion. The global climate sensitivity of a GCM to a parameterizations of that physics, and again testing change in radiative forcing depends heavily on the parameterizations by comparing model output with cloud parameterization scheme employed (Mitchell et observations. al. 1989; Cess et al. 1990). The development of cloud There are problems with both types of approaches. parameterizations is further complicated because

1202 Vol. 75, No.7, July 1994 macroscopic properties such as the persistence and surements will be sufficiently comprehensive to allow radiative properties of clouds, and therefore the verti­ testing of parameterizations through the direct com­ cal distribution of water vapor, can be influenced by parison of field observations with model calculations microphysical properties and processes, such as drop­ of the radiation field and associated cloud and aerosol number density and size distribution, and precipitation interactions. development (Twomey 1977, 1991; Albrecht 1989; The two primary ARM objectives are the following: Baker and Charlson 1990). 1) To relate observed radiative fluxes in the atmo­ Parameterization is also required to account for the sphere, spectrally resolved and as a function of posi­ wavelength dependence of radiant energy transfer. tion and time, to the atmospheric temperature, com­ While the radiative absorption, emission, and reflec­ position (specifically inclUding watervaporand clouds), tion processes associated with the earth's atmo­ and surface radiative properties. sphere and surface exhibit a highly structured wave­ 2) To develop and test parameterizations that length dependence, the computational requirements describe atmospheric water vapor, clouds, and the of using high spectral resolution cannot be approached surface properties governing atmospheric radiation in in climate models, again necessitating parameteriza­ terms of relevant prognostic variables, with the objec­ tion. The Intercomparison of Radiation Codes used in tive of incorporating these parameterizations into gen­ Climate Models (ICRCCM) (Ellingson and Fouquart eral circulation and related models. 1991) clarified the status of the parameterization of The achievement of these objectives should lead to radiative process. That study showed that one can the improvement of the treatment of atmospheric begin from a high-resolution radiative model and ef­ radiation in climate models, explicitly recognizing the fectively determine a useful and consistent param­ crucial role of clouds in influencing this radiation and eterization in the clear-sky case. However, without the consequent need for an accurate description ofthe actual measurements against which to compare the presence and properties of clouds in climate models. output of the original high-resolution model, it is not possible to discern the accuracy of the parameter­ izations oreven to decide which parameterizations are 4. Approach most accurate. Both the original model and the result­ ing parameterizations must be validated with a sys­ The number of processes pertinent to the transfer tematic program of observation. This is particularly of radiation in the atmosphere that must be repre­ true when one moves from the consideration of clear­ sented in climate models is large, and any given sky radiative transfer to the cloudy-sky case. In this process can in principle be represented in a variety of situation, the parameterization involves not only the possible ways. The requirement for a program such as physics of scattering and cloud radiative properties ARM is to test many candidate representations of a but a variety of geometrical considerations as well. variety of processes and identify those that are most At the outset of the ARM Program, it was clear that suitable for use in climate models. ARM will attempt to a comprehensive program of modeling and observa­ achieve this by using data to test models and tion directed at the validation of radiative transfer in parameterizations operated in a predictive mode, clear skies, the further development of radiative trans­ rather than by relying exclusively on phenomenology fer codes for use in the presence clouds, and more and empirical parameterization. Acquisition of data effort in predicting the presence and properties of necessary for model development and testing will be clouds would greatly advance climate modeling. accomplished by establishing and maintaining several sites, whose spatial extent is comparable to the size of a typical GCM grid cell, approximately 200 km on a 3. ARM program objectives side. At each of these sites, continuous measure­ ments will be made, for 7-10 years, of atmospheric The overall goal of the ARM Program is to develop radiation and of the atmospheric and surface proper­ and test parameterizations of important atmospheric ties influencing transfer of radiation in the atmosphere. processes, particularly cloud and radiative processes, These measurements will be used to develop and test for use in atmospheric models. A central feature of model parameterizations. ARM is an experimental test bed for the measurement The research component of ARM, the actual devel­ of atmospheric radiation and the properties controlling opment and testing of specific models and parameter­ this radiation. A principal objective of this test bed is to izations, is the province of the ARM Science Team. develop a quantitative description of the spectral The science team consists of more than 50 research radiative energy balance profile under a wide range of groups whose efforts fall into three broad categories: meteorological conditions. The intent is that the mea- 1) developing and testing parameterizations, 2) devel-

Bulletin of the American Meteorological Society 1203 oping and testing instruments, and 3) participating as A final feature of the ARM approach is its emphasis site scientists. The first group is focused on the actual on continuous operation of the instrumentation at the development and testing of parameterizations. These sites. This emphasis has several motivations. First, investigators are involved in the full cycle of param­ continuous operation permits continuous evaluation eterization development, ranging from the basic delin­ of the performance of parameterizations and process eation of phenomena to be parameterized, to the models, which also must continuously simulate the development of detailed theoretical models to serve planetary atmosphere. This will allow the determina­ as the basis for parameterizations, to the actual test­ tion of whether particular parameterizations or pro­ ing of the parameterizations themselves. Within this cess models exhibit systematic biases when operated group, each investigator defines one or more experi­ over an extended period and the widest possible range ments to be conducted at the ARM sites. An experi­ of conditions. Next, from the investigator's perspec­ ment consists of the comparison of measurements tive, continuously available data allows the acquisition with model output. The model may be initialized with and use of data in a facility mode, offering ready input variables specified by observations at the ARM availability of data and the ability to test new hypoth­ sites and/or by data obtained from other sources-for eses with new data. Continuous measurements also example, the National Weather Service-and opera­ provide an opportunity to identify new technologies tional satellites. that might be incorporated into the observational sys­ The second major activity within the science team tem with relative ease. Finally, continuous operation is the Instrument Development Program (lOP). Inves­ provides the greatest likelihood of a high quality dataset, tigators within the lOP focus on the development and without uncertainties introduced by successive starts testing of instruments that may be suitable for future and shakedowns of the performance of a large suite of deployment to meet measurement reqUirements at instruments. ARM sites. The final component of the science team consists of the site scientists, one for each ARM site. Their 5. The general measurement strategies complete role is described in section 7c, but they also have an active research program, which may involve An important element of the design of the CART the activities of a large group of investigators at their was the specification of the observations that would home institution. Table 1 lists the members of the meet the data requirements of the science team. An science team during the first 3 years of the ARM analysis ofthe various science team projects revealed Program. A more complete description ofthese projects that the various approaches to using CART data could is given in U.S. Department of Energy (1993) (GCR be organized into four classes, referred to as general 1993). measurement strategies (GMS): single-column model, While the objectives of ARM are distinctly focused hierarchical diagnosis, data assimilation, and instan­ on modeled results, the path to these results has a taneous radiative flux. These four strategies capture strong coupling to experiment. The next major ele­ the range of modeling problems addressed by ARM ment of ARM, after the science team, is the Cloud and and are the basis of the design of the CART facilities. Radiation Test bed (CART), which consists of the The following is a description of the four strategies measurement facilities and the process of assembling the data to meet the experimental requirements of the a. Single-column model members of the science team. CART has been de­ One approach to testing the key process models signed so that the observations will support the mea­ and parameterizations in a GCM is to extract a singll3 surement requirements of multiple research groups vertical array of cells from the model and operate the with the same data streams. CART may thus be model in what is often called single-column mode. This viewed as a facility for the prospective testing of subset of the model retains much of the physics that models in a shared data environment. CART will must be represented in climate models and offers a consist of several observing facilities or sites. As convenient approach to testing parameterizations of discussed in section 7b of this paper, the need for the physics. To operate models in this mode, it is several sites is dictated by the wide range of geo­ necessary to specify the initial values of the prognostic graphical and meteorological situations that must be variables within the cell: temperature, water vapor, accurately represented by climate models. The sites wind velocity, cloud amount, radiation field, precipita­ have been selected to allow the observation of a tion, and soil moisture. It is also necessary to provide sufficiently wide range of meteorological situations, the boundary conditions for the column and their time permitting models to be tested under virtually all evolution: wind velocity and thermodynamic variables climatically relevant conditions. at the lateral boundaries, surface emittance and re-

1204 Vol. 75, No.7, July 1994 TABLE 1. Science team projects and principal investigators. Thistable provides a listing ofthe investigators within the ARM Science Team that have been selected through the peer review process over the first 3 years of the program. The site scientists and their institutions are listed in Table 2. The membership ofthe science team has been organized by General MeasurementStrategy (section 6). Several ofthe listed Principal Investigators (Pis) are involved in several General Measurement Strategies, but they are listed only once.

Principal General measurement strategies investigator Organization

Dataassimilation

Point-area relationships for global climate modeling J. C. Doran Pacific Northwest Laboratory (PNL)

Remote sensing of surface fluxes important to cloud development F.Bames Los Almas National Laboratory (LANL)

Area-representative estimates of surface heat flux R. L. Coulter Argonne National Laboratory (ANL)

Integratedcloud observation and modeling T. P.Ackerman Pennsylvania State University

Development of an integrated data assimilation sounding system W. F. Dabberdt NCAR

Test bed model and data assimilation J.F.Louis Atmosphericand EnvironmentalResearCh, Inc.

Cloudformation

Interfacing between a hierarchy of numerical models C. Y. Kao/J. Leone LANULawrence Livermore National Laboratory (LLNL)

Parameterization ofconvective Clouds, mesoscaleconvective W. R. Cotton ColoradoState University systems, andconvective-generated clouds (CSU)

Modeling clouds and radiation for developing parameterizations O. B. Toon/ NASA Ames Research Center for general circulation models D. L. Westphal

Evaluation of a neW GCM-capable stOChastic cloud/radiation R. N. Bryne Science Applications parameterization InternationalCorporation

Development of a radiative and cloud parameterization W. R. Cotton CSU scheme of stratocumulus and stratus clouds, which include the impact of CCN on cloud albedo

Development ofa GCM stratiform cloud parameterization S.J. Ghan PNL

Development and testing of parameterizations for continental A. J. Heymsfield NCAR and tropical ice cloud microphysical and radiative properties in GCM and mesoscale models

Development ofa cloud parameterization package for the C.Y.Kao LANL improvementofcloud simulationsin climate models

A hierarchial approach to improved cloud-radiation J. T. Kiehl NCAR parameterization forclimate models

Testing an Improvedparameterizationofupper-tropospheric D. A. Randall CSU clouds for use In climate models

Parameterization of GCM subgrid nonprecipitating cumulus R. B. Stull University of Wisconsin, and stratocumulus clouds using stochastic phenomenological models Madison

Bulletin ofthe American Meteorological Society 1205 TABLE 1 continued. Science team projects and principal investigators.

Radiativemodeling

Spectroscopic study of water absorption in the 8-14-J1m atmospheric T. J. Kulp LLNL window measurement of new line and continuum parameters and investigation offar wing phenomena

Instrumentdevelopment program

Radiometric instrumentation

Tethered balloon sounding system for vertical radiation profiles C. D. Whiteman PNL

Development of rotating shadowband spectral radiometers and GCM radiation code test datasets

1206 Vol. 75, No.7, July 1994 TABLE 1 continued. Science team projects and principal investigators.

Radiometricinstrumentation

Development of a shortwave infrared solar spectral radiometer D. G. Murcray University of Denver (UD)

Development of a radiation measurement system P.J. Valero NASA ARC

High spectral resolution radiance measurements H. E. Revercomb University of Wisconsin. Madison

Cirrus and aerosol profilometer for radiometric measurements J. Splnhirne NASA GSFC

Remotesensinginstrumentation

Laser remote sensing of water vapor M. Lapp Sandia National Laboratory (SNL), Livermore

Developmentof a high-resolution L1DAR technology UniVersity of Wisconsin. Madison

Ground-based millimeter wave cloud profiling radar systems (CPRS) R. E. Mcintosh University of Massachusetts, Amherst

Cloud and aerosol characterization. mUltiple remote Sensor KSassen University ofUtah. techniques development Salt Lake City

Development of an integrated data assimilation/sounding system E. R. Westwater/ NOAA Wave Propagation K. S. Gage Laboratory (WPL)/NOAA Aeronomy Laboratory

Passive cloud dynamics measurement, A flow field registration approach SNL, Livermore

Shipboard measurements of the cloud-capped marine boundary layer during FIRE/ASTEX R. A. Kropfli NOAAWPL

A precise passive narrOW-beam filter infrared radiometer C. M. R. Platt Commonwealth ScientifIC and its use with !idar andIndustrialResearch OrganiZation

flectance at the lower boundary, and the top of the meterizations is to take advantage of the fact that atmosphere solar flux. The GCM parameterizations GCMs are part of a hierarchy of models and modeling are then tested by comparing measured and modeled systems whose representation of atmospheric phys­ evolution of the prognostic and other variables, such ics ranges from the highly aggregated to the highly as cloud amount, radiation field, precipitation, and soil specific. GCMs and single-column models are neces­ moisture. Observations at the CART site serve to sarily characterized by highly aggregated physics and determine the boundary and initial conditions of the by coarse spatial and temporal resolution. To under­ prognostic variables and track the temporal eVOlution stand the limitations imposed by such aggregation, of the prognostic variables predicted by the models. and ultimately to improve the models, it is useful, if not necessary, to analyze and interpret observations us­ b. Hierarchical diagnosis ing higher-resolution models. Such analyses can give A second approach to the development of para- partiCUlar insight into how much of the detailed physics

Bulletin of the American Meteorological Society 1207 is required in a model to capture the essential features influence of including observations of water or tem­ of a particular process. perature in the assimilation. This allows attention to be There are several possible meanings of "low reso­ focused on a cloud formation model while minimizing lution" in highly aggregated models. The influence of the effects of the calculation of the velocity fields by low spatial and temporal resolution can be studied continually updating the velocity field with assimilated using higher-resolution models overthe same domain data derived from observation. An approach like this covered by the coarser grid. GCMs use low-"resolu­ offers the opportunity to focus attention on particular tion" physics as well. The higher resolution would then aspects of a given process, while reducing the influ­ involve the use of models with more comprehensive or ence of uncertainties introduced in modeling multiple elaborate physical descriptions. An example might be processes at the same time. the use of a detailed cloud model that includes micro­ physical parameterizations to diagnose the perfor­ d. Instantaneous radiative flux mance of a GCM cloud model that uses only relative The final GMS is a special case of the hierarchical humidity and vertical velocity to infer the formation and diagnosis strategy but is so central to the ARM Pro­ persistence of clouds. The hierarchical diagnosis gram that it is treated separately. Accurate treatment measurement strategy will require observations on a of radiation is essential in climate models; and testing finer spatial grid and may also require measurement of of radiation transfer models is central to the objectives parameters such as aerosol light scattering coeffi­ of ARM. The ICRCCM (Ellingson and Fouquart 1991) cients or cloud microphysical parameters that are illustrated that the state of the art in radiative modeling treated in various detailed process models but are not cannot be advanced by further model intercomparisons treated explicitly in present GCMs. and that it is essential that the intercomparison be supplemented with field observations if further progress C. Data assimilation is to be made. This leads to the requirement that the The third approach recognizes that observations of observations provide an almost instantaneous char­ atmospheric properties (the outputs of individual in­ acterization of the state of the atmosphere under struments) are made at discrete locations and times clear-sky, general-overcast, and broken-cloud condi­ and necessarily sample only limited volumes of the tions that can be used as input into radiative models. atmosphere. However, to test models on a variety of Strictly speaking, the requirement is not for instan­ physical scales, it is necessary to develop methods taneous characterization but concurrent characteriza­ that will allow the output of individual instruments, tion of atmospheric state and consequent fluxes. Any which measure different parameters, to be combined lack of concurrency between observations of the to infer the time-dependent three-dimensional field of atmospheric and surface properties and the radiative meteorological variables. The requirement for these fluxes can thus lead to major problems in the compari­ four-dimensional fields of meteorological data is par­ son of predicted and observed fluxes. In this strategy, ticularly important for testing models of cloud forma­ the set of observations required to predict the radiative tion, maintenance, and dissipation, and the effects of fluxes are the upper boundary condition (the solar cloud fields on atmospheric energy fluxes. Data as­ zenith angle and the incident solar flux), the lower similation is a method for combining observations of boundary condition (the longwave and shortwave different parameters using a physically based model radiative properties of the sUrface), and the radiative as an interpolation vehicle. The principle of the ap­ properties of the intervening atmosphere. The models proach is that the derived four-dimensional field must then predict various integral forms of the intensity of satisfy certain imposed constraints-for example, for the radiation-for example, the flux as a function of continuity and conservation of mass, energy, and wavelength and direction. momentum. Data assimilation is expected to be an important method for establishing the boundary con­ ditions for a single-column model and for developing 6. The cloud and radiation test bed average conditions over the entire ARM site to initial­ ize the models. The field measurement component of the ARM Data assimilation approaches are also useful as a Program is the CART. CART has been designed to technique for focusing attention on a more limited set acquire the observational data necessary to carry out of variables than the full complement of prognostic the GMS, to prepare these data to compare with variables. For example, one might choose to assimi­ model results, and to archive the resulting data streams. late the observations of the velocity field with a mesos­ CART will ultimately include several permanent ob­ cale or single-column model, while allowing the water serving sites and an ability to conduct intensive cam­ vapor and temperature fields to evolve without the paign activities at the fixed ARM sites and elsewhere,

1208 Vol. 75, No.7, JUly 1994 to supplementthe long-term con­ tinuous streams of data. There Hillsboro also is a data processing and "" archiving system, which sup­ ports the above activities, ac­ quires additional related data re­ quired by ARM experiments, and facilitates analysis of these data Haviland WiChita by the science team. The follow­ "Wind Profiler" NEXRAD ~~ ing subsections provide a de­ Neodesha scription of the configuration of "Wind Profiler" an ARM site, a discussion of the locales selected for ARM sites, Coffeyville information about deployment FIRE Site and data management at ARM sites, and a discussion of the use of intensive observational Lamont "Wind Profiler" Central Facility periods (lOPs) and unmanned aerospace vehicles (UAVs) Tulsa • 1993 within the ARM Program. Watershed NErD ,.,.,.,...,.~.,.,.- Study Area Tulsa a. The configuration ofan ARM site Haskell "Wind Profiler" The ARM field measurement NSSL Doppler Radars program will consist principally of coordinated sets of observa- tions ateach of five primary sites. /~~'hom. These sites are the principal ex­ perimental resource of CART. .._-- To satisfy the requirements of Chickasha Watershed o 50 the GMS, each ARM site will Study Area I II III consist of several components, Kilometers as indicated in Fig. 1. The figure shows the design of the first Ef? ,.....-.... Lawton Watershed ARM site, in the southern U.S. Study Area E!l!m Durant Great Plains. The design of each l1li Watershed Study Area site varies depending on the geo- graphical and meteorological at­ tributes of the site and on the Texas particular scientific issues to be emphasized. The components of a site are a central facility, a FIG. 1. The southern Great Plains site near Lamont. network of auxiliary stations, a network of extended observing stations, and a set of the distribution of shortwave and longwave radiation boundary facilities. and the physical conditions that control radiation trans­ The following description of the components of an fer-the instantaneous radiative flux GMS. Observa­ ARM site shows the contributions of several compo­ tions at the central facility provide high spectral reso­ nents of the site to the generation of the observations lution radiometric measurements and a detailed char­ required to support the experiments represented by acterization of the atmospheric column above the the GMS. Although all of the following elements are facility. To meet these requirements, ARM will deploy being developed for the first ARM site, subsequent equipment at the central facility for measuring down­ sites mayor may not include all these elements. welling and upwelling spectral radiance and for char­ acterizing the pertinent local atmospheric properties, 1) CENTRAL FACILITY including temperature and water vapor concentration A critical experimental task of ARM is to measure (both as a function of altitude), cloud fractional cover-

Bulletin of the American Meteorological Society 1209 age, cloud-base altitude, and liquid water path,and the fication and selection of specific sites within these local surface reflectance, temperature, and emissivity. locales. The complete locale selection process is presented in U.S. Department of Energy (1991) and is 2) AUXILIARY STATION NETWORK summarized here. A series of auxiliary stations surround the central The key principle guiding locale selection was that facility within a 10-20-km radius. These stations char­ the set of locales should stress models describing acterize the three-dimensional structure of the atmo­ radiation transfer in the atmosphere and atmospheric sphere in the region that exerts the major influence on properties influencing such radiation transfer by span­ the radiation flux at the central facility. The instrumen­ ning, as much as possible, the domain of radiation­ tation at these facilities provides a reconstruction of influencing attributes. The attributes used in selecting the cloud geometry surrounding the central facility the locales included the following: using all-sky cameras, satellite imagery, and ceilome­ ters. This network is intended to provide support for • latitude the instantaneous radiative flux GMS, particularly as • altitude it relates to three-dimensional radiative transfer is­ • continentality (midcontinent, continental margin, sues. The network also supports studies of cloud ocean) formation as part of the data assimilation and hierar­ • terrain (level, mountainous, etc.; uniform, variegated) chical diagnosis GMS. • surface (water, land; vegetated, desert, snow or ice covered, etc.; uniform, variegated) 3) EXTENDED OBSERVING NETWORK • cloud frequency and type (cirrus, stratus, cumulus, Surrounding the central facility and the auxiliary ground fog, etc.) station network will be an additional network of 16-25 • precipitation: amount, frequency, type extended observing stations. The extended observing • temperature: mean, range area will include a region comparable to that expected • humidity: mean, range for GCM grid cells anticipated over the next decade, • seasonality approximately 200 km x 200 km. The instruments at • concentrations of ozone, pollution aerosols, wind­ the extended stations will collect the basic radiometric blown dust, etc. information, conventional meteorological data, and surface flux data needed to characterize the radiative The goal was to select a set of locales that was as transfer throughout the extended area, giving insight small as possible but that could test models and into the averaging process implicit in many GCM para­ parameterizations over a range of conditions suffi­ meterizations. Only limited vertical information will be ciently wide enough to give confidence in their general collected, with the more extensive and demanding applicability. profiling equipment reserved for the central facility. It was decided that the initial ARM sites should be These facilities will support the single-column model, fairly homogeneous spatially so that the initial test hierarchical diagnosis, and data assimilation GMS. case for models would be uncomplicated by spatial discontinuities caused by mountain ranges and coast­ 4) BOUNDARY FACILITIES lines. However, since climate models must also deal Several GMS, such as the single-column model and with such discontinuities, it was recognized that such four-dimensional dataassimilation, involve modelswhose features either must be present at later ARM sites or operation requires specification of boundaryconditions. be dealt with by campaigns. To maximize the range of The need for these boundary conditions will be met by conditions that would be experienced by the small a set of facilities to support determination of the vertical number of ARM sites, it was decided that the greater profile of the horizontal fluxes of critical quantities such the temporal variability at a given locale, the more as water and energy at the edges ofthe roughly 200-km suitable that locale would be. Thus, a locale that x 200-km ARM site. experiences high seasonal and intraseasonal ranges in temperature, humidity, surface state, and fluxes b. Locale selection was to be preferred to one that did not exhibit such A major step in designing CARTwas selecting sites ranges. for the ARM measurements. A two-stage approach A further key consideration was the logistics of was used. First, a set of generic locations or "locales" establishing and operating an ARM site within the was selected where sites might be located that would locale. Although unfavorable logistics can be over­ collectively meet the ARM scientific requirements come, this leads to increased costs, so that other subject to the constraint of logistical suitability. The things being equal, a logistically favorable locale is second stage, which is continuing, consists of identi- preferred. Sites must be physically, politically, and

1210 Vol. 75, No.7, July 1994 FIG. 2. Geographical distribution of recommended locales. economically accessible, and they should have an types; large intra-annual variability of surface flux adequate infrastructure (roads, power, communica­ properties and weather, including cloud types, tem­ tion, living accommodation, etc.). They must also be perature, and specific humidity and favorable logis­ a sufficient distance from populated areas to preclude tics. These requirements are met by the southern potential negative impacts (e.g., an urban heat-island Great Plains (SGP) locale. This locale also affords the effect). opportunity for synergistic activity with other ongoing Another consideration was the presence of other and planned meteorological projects and facilities. A projects conducting related atmospheric measure­ key facility is the network of wind profiling stations ments, which might offer synergistic interaction with being installed in this area as part of the National ARM. Sites should offer possibilities for sharing of Oceanic and Atmospheric Administration (NOAA) Wind data, facilities, and costs with other agencies and Profiler Demonstration Network; the high density of programs making atmospheric measurements and vertical atmospheric structure data from this network concerned with atmospheric science and climate will be very important to many ARM experiments. change. Another pertinent project is the Global Energy and Based on the above considerations, a set of locales Water Cycle Experiment (GEWEX; WCRP 1992; was recommended for establishment of ARM sites. Chahine 1992), whose Continental Scale Interna­ This set of locales consists of five primary locales tional Program (GCIP) will focus on the greater Missis­ recommended for long-term occupancy, 7-10 years, sippi basin. The SGP is also situated favorably with and four supplementary locales recommended for respect to the orbit of the Tropical Rainfall Measuring short duration or campaign occupancy (months to a Mission (TRMM) satellite, which is expected to pro­ year). Occupation ofthe supplementary locales would vide valuable measurements of key physical and permit testing of ARM models over a wider range of radiative variables. atmospheric and surface conditions than can be en­ compassed at the primary sites. The geographical 2) TROPICAL WESTERN PACIFIC OCEAN distribution of the recommended locales is shown in Ocean locales are crucial for ARM both because of Fig. 2. the large fraction of the earth surface covered by The rationale for recommending establishment of oceans and because of the globally important cloud ARM sites in each of the primary locales is summa­ types and meteorological situations that are found rized here. only at ocean locations. The Tropical Western Pacific Ocean (TWP) is uniquely suited for characterizing 1) SOUTHERN U.S. GREAT PLAINS deep tropical convection responsible for the transport Key requirements for the first ARM locale included of water vapor to the upper troposphere and for the high geographical homogeneity; a variety of cloud tropical cirrus cloud distributions over much of the

Bulletin of the American Meteorological Society 1211 ocean. This locale is well suited for observing cumu­ short-term for intermittent occupation or for research lonimbus clouds and for observing the distribution and campaigns to address specific modeling needs that radiative impacts of fair-weather cumulus clouds. It cannot be adequately addressed with measurements experiences extremely high temperature and specific at the primary sites. The rationale for recommending humidity for an ocean locale and displays substantial the several supplementary locales is summarized interannual variation associated with the EI Nino/ here: Southern Oscillation (ENSO). • Central Australia or Sonoran Desert. The high 3) NORTH SLOPE OF ALASKA temperatures and low specific humidity of the cen­ The North Slope of Alaska (NSA) experiences tral Australian or Sonoran Deserts permit substan­ highly diverse atmospheric and surface properties: tial extension of the range of conditions over which low temperatures, high albedo when covered with ice infrared radiation transfer models can be tested. or snow, moist vegetation and low sun in summer, and Totally clear sky is frequent, enhancing the value of polar night at midwinter. It accordingly experiences a such studies in these locales. In addition, there is an wide range in surface fluxes. Together, the two land identified need to improve parameterizations for locales, the SGP and the NSA, span a wide range of latent and sensible heat exchange at desert sur­ atmospheric, surface-flux, and geographic conditions. faces, particularly dry and moist convection, and The NSA is a region where climate feedbacks relating the coupling between soil moisture and downwind surface and tropospheric temperatures, surface al­ cloud formation and precipitation. bedo, evaporation, cloud cover, and the polar atmo­ • NorthwestUnitedStates-southwestCanada coast. spheric heat sink are expected to be large. A campaign in the northwest United States over several winter months would provide a strenuous 4) EASTERN OCEAN MARGINS test ofthe ability of models to simulate the response Eastern ocean margins represent a prevalent and of clouds to orographic inhomogeneity. This locale climatologically important cloud and meteorological is very inhomogeneous and has abundant stratus/ situation that is quite distinct from that represented by altostratus clouds strongly modulated by the pres­ the TWP. Both the eastern North Pacific and the ence of a mountain range. Wintertime nimbostratus eastern North Atlantic locales exhibit a high frequency occurs much morefrequently on the seaward slopes of low-level marine stratus, a key cloud type governing of the mountains than on the leeward side. the earth radiation budget, and both locales experi­ • AmazonorCongobasin. These are climatologically ence moderate latent heat fluxes. An ARM site in important regions with moderate intra-annual vari­ either ofthese locales would meet the requirements of ability and little annual variability. Deep convection an eastern ocean margin locale; the final choice of occurs in these regions almost daily. Surface latent locale and site will depend on logistical and synergistic heat fluxes are large, and the specific humidity is considerations and results of earlierexperiments such often very high. The Amazon and Congo are good as the First ISCCP (International Satellite Cloud Cli­ locales for testing the accuracy of treatment of the matology Project) Regional Experiment-Atlantic Stra­ water vapor continuum and radiative transfer in tocumulus Transition Experiment (FIRE-ASTEX). penetrating convective clouds. These are also good locales for testing models of exchange of heat and 5) GULF STREAM moisture between the surface and the air and of The Gulf Stream off eastern North America, a warm redistribution of these quantities within the atmo­ western boundary current, exhibits extreme ranges sphere. Also, smoke from biomass burning varies and variability in surface heat fluxes. Cold-air out­ SUbstantially, allowing tests of effects of this breaks in the fall and early winter from the North aerosol on radiation transfer and cloud microphys­ American continent result in air-water temperature ics. differentials that are greater here than anywhere else • Beaufort, Bering, orGreenlandSea. The polarseas globally. This is a key local for observing formation, are key locales for studying important issues asso­ distribution, and radiative properties of marine al­ ciated with the ocean-ice edge. Issues of impor­ tostratus. This locale also provides an opportunity to tance include changes in albedo and surface fluxes observe mature midlatitude cyclonic storms, the only accompanying the growth and decay of sea-ice storm-related cloud systems expected to be resolved and possible albedo compensation of changes in by present and future GCMs. ice cover by the decay or growth of marine stratus clouds. The Beaufort Sea locale is the most appeal­ 6) SUPPLEMENTARY SITES ing logistically. It is frozen in winter and melts in Sites in the supplementary locales are intended for summer, providing manyopportunities for ice-edge

1212 Vol. 75, No. 7, July 1994 TABLE 2. Site management.

Site Site program manager Sitescientist Site operator

southern u.s. Great Plains O. L. Sisterson, ANL P. J. Lamb, University of Oklahoma, Forecast Systems NormanLab., NOAA

Tropical western Pacific Ocean W. E. Clements, LANL T. P. Ackerman, The Pennsylvania State University

North Slope ofAlaska 8.0.Zak, K. H. Stamnes, Sandia, Albuquerque University of Alaska, Fairbanks

Eastern North Pacific R. M. Reynolds, BNL or Atlantic Oceana

Gulf Streama P. Michael, BNL S.Raman. North Carolina State University. Raleigh aThis paper was prepared prior to a recent change of status for these locales from primary to supplementary.

studies. The proximity of the Beaufort Sea to the instrument types are deployed at the central facility; North Slope of Alaska locale may allow a site subsets of these instruments are deployed at the selection strategy that would permit studying both extended and boundary facilities. Several instruments locales. listed in the table are under development within the Instrument Development Program; others have com­ c. Deployment pleted that development stage and are being de­ The planning for and operation of a CART site is ployed as operational instruments. the responsibility of two individuals, the site program manager and the site scientist. The site program d. Data management manager is responsible for planning the site infra­ The ARM data system, referred to as the CART structure, managing the deployment of instrumenta­ Data Environment (CDE), is an essential component tion to the field, and providing operational oversight of of the ARM Program. The data system provides for the site operator. The site scientist is responsible for acquisition and processing of data from the CART developing the detailed scientific mission plan for the sites and data from external sources. It supports the sites and has the day-to-day responsibility for ensur­ combination of these data streams into a form re­ ing that the scientific objectives of the site are being quired by various science team experiments. The met. The latter responsibility leads to the site scientist system also provides for ongoing quality control of the assuming much responsibility for the ongoing quality various processes, archival storage of the data, and control for the data. The site scientists were selected access to the ARM data by the rest of the scientific through a competitive process and have responsibil­ community. ity for conducting a research program at the site and The structure of the data system is shown in Fig. 3. conducting an educational program in conjunction The main components of the system are the site data with site operations. Once a site is established and systems, which provide for acquisition of data at each instruments and facilities are installed, a site operator of the ARM sites and for management of site opera­ assumes the day-to-day responsibility for operations. tions; an experiment center, which brings together The current site program managers, site scientists, information from the sites and other data sources for and site operators are shown in Table 2. processing to meet the particular needs of the science Deployment of instrumentation to the first site, in team; and an archive, which provides forthe long-term the SGP locale near Lamont, Oklahoma, began in storage of the data and access for the broader scien­ May of 1992. Table 3 shows the set of instruments tific community. The experiment center is at the Pacific intended for deployment at the site. The table indi­ Northwest Laboratory in Richland, Washington, and cates the deployment of the instruments at each of the the archive is at Oak Ridge National Laboratory in Oak types of facilities within the site. Most of the different Ridge, Tennessee. The data system uses the national

Bulletin of the American Meteorological Society 1213 TABLE 3. Planned deployment of instruments and observing capabilities at the SGP CART site. Except as indicated, all instrumentsl observations will be at the central facility (CF). Check marks indicate instruments/observations that will also be located at extended facilities (EFs) or boundary facilities (BFs). The notation IDP indicates thatthe instrument is underdevelopment in the Instrument Development Program for possible future deployment at the site.

Central facility Extendedfacility Boundaryfacility

Radiometric

Wind, temperature, andhumidityaloft

Internet as its operational backbone, giving ARM analysis support. The basic architecture of leb was investigators access to a variety of computational conceived to be highly flexible in number and types of resources to support their modeling and analytic re­ input data streams accommodated and in processing quirements. and displaying the data. This flexibility supports the The operational kernel of the CART data system is development of incremental capabilities that target the leb system (Corbet and Mueller 1991) developed specific needs of the ARM application. The leb sys­ by the National Center for Atmospheric Research tem thus provides many of the logical functions re­ (NCAR). This system was originally developed for the quired by CART, both for the site and the experiment collection and analysis of atmospheric data to support data systems. The data system has been designed field campaigns and to provide postcampaign data and sized to meet the initial requirements for modeling

1214 Vol. 75, No.7, July 1994 TABLE 3 continued. Planned deployment of instruments and observing capabilities at the SGP CART site.

Central facility Extendedfacility Boundaryfacility

Cloudobservations

Whole-sky Imaging system (lOP) x (IOP) X Cloud radar (lOP) X

Nea~surlaceozoneandaerosolobservations

Ozone sensor X Integrating X Optical particle counter X Aerosol optical absorption system X Condensation nuclei counter X

Other

X X Energy balance Bowen ratio stationa X X

__ un. _ _ _.. .~__• •• ~ ••••_ •••••••••••• Eddy correlation (25-60mat CF; 3 m at EFs)3 X Temperature and humidity (60 m) . X aEither the energy balance Bowen ratio station or the eddy correlation station, but not both, will be deployed at the EFs.

and instrumentation and can be expanded as neces­ The ARM data archive at Oak Ridge National sary in the future. Laboratory will provide access to ARM data for non­ To meet the scientific objectives of ARM, it is ARM participants. The Oak Ridge facility, which is necessary to combine observations from the CART affiliated with the Earth Observing System Data and sites with data from other sources. Examples of such Information System (EOSDIS) as a Data Analysis and external data include the wind profiler data from the Archieve Center (DAAC), has begun providing data on NOAA demonstration network and data from opera­ a limited basis. More extensive access to ARM data tional satellites maintained by NOAA and the National should begin in late 1994. Eventually all data and data Aeronautics Space Administration (NASA). Satellite products generated by ARM will be available through observations offer critical data for ARM. Top-of-the­ the archive. As an EOSDIS-DAAC, the Oak Ridge atmosphere radiative fluxes and profiling of atmo­ archive should also allow merging of ARM data with spheric conditions in regions beyond the range of data streams from other DAACs. sounds and ground-based sensors are crucial for many ARM projects. The requirement for external e. Intensive operations, campaigns, and interactions data will also include data products that go beyond with other programs data from various observational systems, such as While the ARM observational strategy is to provide meteorological forecasts and cloud statistics from the a baseline set of observations available continuously ISCCP. The data system has been explicitly designed from each of the primary sites, this set of observations to allow ingest of this class of external data, its merger may not always be to fully satisfy the requirements of with the data streams from the ARM sites, and delivery the scientific experiments to be conducted within to the ARM Science Team. ARM. Consequently, the baseline set of observations

Bulletin of the American Meteorological Society 1215 experience that would support laterfield activities. For example, Instruments the Pilot Radiation Observation • Microwave Experiment (PROBE) was con­ Radiometer ducted during November 1992­ Site February 1993 at Kavieng, • Rotating Experiment Shadowband Data Center Papua New Guinea, in conjunc­ Radiometer System tion with Tropical Ocean-Global • Bown Ratio Surface Flux Atmosphere-Coupled Ocean Station Atmosphere Response Experi­ • Balloon Borne ment (TOGA COARE) and pro­ Sounding System vided measurements of radia­ • Surface tion in the tropical western Pa­ Meteorological Observing cific. ARM also supported the Stations Boardman Regional Flux Experi­ ment (BARFEX) (June 1991, near Boardman, Oregon; Doran et al. 1992) that examined subgrid-scale variability of sen- sible and latent heat fluxes and FIG. 3. Schematic representation of ARM data system components with associated data laid the groundwork for similar flows. observations at ARM sites. These pilot projects and the role will be supplemented periodically with lOPs. During they played in the level of ARM are summarized in lOPs, ARM will support several classes of observa­ Table 4. tions, including observations that are too expensive or In the future, ARM plans to host and support the personnel intensive to be conducted continuously, conduct of campaigns at primary or supplementary calibrations or in situ testing of validity of remote ARM sites. The ARM facilities offer an important sensing data, and field testing of new instrument­ operational base on which other observational cam­ ation. paigns can be based. Examples of possible cam­ Possible lOPs include the following or combina­ paigns include a comprehensive flux divergence ex­ tions of the following: periment with multiple aircraft at different altitudes above an ARM site or process-oriented studies similar • increased frequency or density of observations to to FIRE or WISP operating at an ARM site. ARM has test more detailed models of atmospheric pro­ set a pattern of collaboration with othercomponents of cesses the USGCRP and the international research commu­ • measurements of low stratus optical and micro­ nity. As noted previously, the program has benefited physical properties with a tethered balloon significantly from interaction with other programs in its • measurements of cirrus optical and microphysical early development stages. These interactions have properties with aircraft (manned or remotely pi­ included nonclimate-related projects such as WISP, loted) but the vast majority have been climate related, in­ • augmentation or modification of observations to cluding FIRE-cirrus and FIRE-stratus and TOGA allow testing of IDP instruments COARE. • observations coordinated with satellite overpass The establishment of a series of facilities, which will schedules. be occupied for an extended period, makes the CART an attractive base upon which to build more campaign­ The ARM Program also participates in more tradi­ oriented projects. For example, the ARM facilities will tional observational campaigns in collaboration with be a major component of GEWEX field experiments. other programs. ARM participation in campaigns is The ARM-GEWEX collaboration will include partici­ generally directed to specific scientific objectives that pation in the continental program scheduled for the complement or supplement the ARM objectives. Early Mississippi basin GCIP and an intercomparison of in the program, ARM collected observations in con­ water vapor measuring systems GEWEX Water Va­ junction with several projects [FIRE; Spectral Radi­ por Program. ARM facilities may also serve as a basis ance Experiment (SPECTRE); Winter Icing and Storms for the next field deployment of the International Project (WISP)], in an attempt to gain operational Satellite Land Surface Climatology Project.

1216 Vol. 75, No.7, July 1994 f. ARM and unmanned aerospace vehicles Tables 4 and 5. Key programmatic and technical The original ARM Program Plan (U.S. Department milestones shown in Table 5 include critical events of Energy 1990) called for the limited use of aircraft in such as the selection of the science team. As noted in the operation of CART. Late in 1990 the possibility of section 7e, the program has used a series of collabo­ a new kind of airborne measurement became appar­ rative pilot projects to test critical concepts and instru­ ent in what are frequently called "Unmanned Aerial mentation before deploying them in the more perma­ Vehicles" (UAVs) (CalSpace 1992; JASON 1992). nent installations. Many of these pilot deployments These aircraft offer ARM the possibility of long-dura­ were conducted with the cooperation of several other tion flights (greater than 48 h), autonomous operation agencies including NASA, NOAA, the Federal Avia­ over long distances, and considerable payloads (ap­ tion Administration, and the National Science Founda­ proximately 150 kg). The ARM missions proposed for tion. UAVs include multiday missions to measure the flux The central activity of the program is the develop­ divergence at the tropopause at several of the sites, ment of the field facilities. The following is a summary delineate boundary profiles of temperature and hu­ of the status of the activity for five primary locales. midity with , study the radiative transfer associated with deep tropical convection in the central a. Southern Great Plains and western tropical Pacific, and cross calibrate sat­ The SGP site covers approximately 60 000 km 2 ellite sensors. near the center of the locale shown in Fig. 2. The Currently, a UAV program called "ARM-UAV" is central facility is approximately 7 miles southeast of underdevelopmentwith the operational goals outlined the town of Lamont (Fig. 1). As noted earlier, this site above. This program began flight activities in the fall of is in the midst of the densest portion of the NOAA 1993 with a series of demonstration flights to profile demonstration wind profiler network, and the outer longwave and shortwave radiative flux in the lower boundaries of the site are delineated by the six profilers troposphere (below 8 km). It is anticipated that these shown in the figure. The layout of instrumentation in observations will soon be conducted at the SGP site. the central facility is shown in Fig. 4. The installation of instrumentation at the site began in April 1992, with the bulk of the instrumentation to be in place in early 1994. 8. Conclusion: Schedule and status Initial deployment of the instrumentation emphasized the placement of instruments in support of the instan­ The ARM Program is planned to last a decade or taneous radiative flux GMS. Several intensive obser­ more. The schedule of near-term events, pilot deploy­ vational periods have already occurred at the site, and ment, and a history of the program are summarized in several instruments from the Instrument Development

TABLE 4. Pilot deployments.

Pilot effort Location/date Description

WiSP NorthernColorado/winter1990/91 Provided a test of remote sensing equipment including RASS andmicrowave radiometers.

BARFEX 1 Boardman, Oregon/summer 1991 Began the process of understanding the mechanism by which the lower boundary condition of an SCM might be examined.

SPECTRE-FIRE Coffeyville, Kansaslta1l1991 SPECTRE provided a test of the concepts behind the instantaneousradiative flux measurementstrategy.

ASTEX Azores/spring 1992 Limited participation but support of cloud radar and wind profilerdeployment.

BARFEX2 Boardman,Oregon/summer1992 Continuationof BARFEX 1.

PROBE Kavieng, Papua New Guinea/winter 1992/93 Support of TOGA COARE and testing of concepts for radiometric observations in the TWP.

Bulletin of the American Meteorological Society 1217 {l co t a:o

Roads

Roadside Cluster

Not on the Three acoustic sources, quarter 15m from corners of section radar antenna array for the Central Pasture Facility

Central Cluster Instruments Potential additional - Broadband Radiation location for active - remole sensing systems - Sky Imagery (IDP and CART) - Energy Balance Bowen Ratio Surface Fluxes - Microwave Radiometer

FIG. 4. The instrument layout in the central facility at the southern Great Plains site.

Program have been deployed as operational instru­ b. Tropical western Pacific ments. For a full account of the scientific objectives of The siting strategy for the TWP is complicated by the SGP site and current deployment status, see issues such as the heavy instrumentation require­ Schneider et al. (1993). ments associated with studies of deep tropical con-

1218 Vol. 75, No.7, July 1994 vection, the logistical complexity of a site that would be mated radiation measurement system called an "At­ free from effects of land, and the great spatial extent mospheric Radiation and Cloud Station" (ARCS). Both of the processes (e.g., EI Nino). These complications ARCS and ISS are modular field stations built around require some compromises. Currently, the compli­ an air-conditioned seatainer. This approach was tested cated nature of the scientific issues in the TWP has led during PROBE and appears to provide excellent data to a three-element approach to deployment of instru­ in support of the instantaneous radiative flux measure­ mentation. These strategies are built around basic ment strategy. Subsequent activity will depend heavily radiative processes, studies of deep convection, and on the experience gained in the early phases of the processes over the open ocean. The current proposal project and the progress made in other programs, is to field instrumentation in three different experimen­ such as GEWEX and the satellite-borne tropical rain­ tal configurations to meet the scientific goals, rather fall measurement mission. than using the single, multipurpose facility approach adopted in the SGP. c. North slope ofAlaska The first phase of the TWP field efforts will be on On the NSA the conditions and logistics are of Manus Island, Papua New Guinea, the first of an east­ comparable difficulty to the TWP. The current ap­ west chain of island-based stations targeted at radia­ proach is focusing on a sitting strategy more similar to tive processes. A key building block of the strategy will that being pursued in the SGP. This approach would be a combination of the NCAR-NOAA Integrated place something similar to the SGP central facility Sounding System (ISS; NCAR 1993) and an auto- near Barrow, Alaska, with boundary facilities located

TABLE 5. ARM schedule and milestones.

Type of Milestone Milestone Date Description-<:omments

Generalprogrammatic milestones

Approval byCEES December 1989

Selection of first science team members Summer1990 ARM Science Team members are selected and funded for periods of 3 years.

Second round of science team selection Summer 1991

First science team meeting November1990 Las Vegas, Nevada

Second SCience team meeling November 1991 Denver, Colorado

Third science team meeting March 1993 Norman, Oklahoma

Fourth science team meeting Charleston, Southc~r~~~~~

Site selection and deploymentmilestones

Publication of locale recommendation report Aprli1991

Begin occupation of SGP Site­ April 1992 Occupation of the site will take Grant County, Oklahoma place over a 2-year period.

ARM-UAV first demonstration flight Fall199S

Begin occupation of TWP site Fall 1994 Begins with ARCS deployment on Manus Island Papua New Guinea.

Begin occupation of NSA site 1995-1996

Bulletin of the American Meteorological Society 1219 on the tundra and along the coast. The science team, Acknowledgments. The authors acknowledge the support of the which met in March of 1993, agreed that the siting U.S. Department of Energy's Environmental Sciences Division, the strategy would also consider placing instrumentation hard work of the multilaboratory and multi-agency participants, the support of the academic community, and the leadership of the ARM out on the ice for at least part of the year. This strategy management team and the ARM Science Team Executive Commit­ is compatible with the Surface Heat Budget of the tee. Their dedication and support have brought the project to this Arctic Ocean program. Further, the development of stage ofdevelopment. UAVs may make studies of the sea-ice margin more This studywas prepared forthe U.S. Department of Energy under practical than previously expected. In the summer of Contract DE-AC06-76RLO 1830. 1993, instrumentation tests began on the NSA. Early tests emphasized the effect of acoustic sources on References wildlife, but eventually they will be focused on the difficult challenge of hardening instruments for opera­ Albrecht, B. A., 1989: Aerosols, cloud microphysics, and fractional tion in the Arctic. A key to the hardening and eventual cloudiness. Science, 245, 1227-1230. deployment strategy is the development of the ARCS -, et aI., 1994: The Atlantic Stratocumulus Transition Experi­ for the TWP. ment-ASTEX, Bull. Amer. Meteor. Soc., submitted. Baker, M. B., and R. J. Charlson, 1990: Bistability of CCN concen­ d. Eastern North Pacific/Atlantic Ocean and Gulf trations and thermodynamics in the cloud-topped boundary layer. Nature, 345,142-145. Stream locales Calspace, 1992: Report of the workshop on scientific application of In the locale covered by the two Northern Hemi­ remotely piloted aircraft measurements of radiation, water va­ sphere marine stratus zones, which have been called por, and trace gases to climate studies. NASA Ames Research the eastern ocean margin locales, the emphasis has Center and the California Space Institute, San Diego, CA, 31 pp. CEES, 1990: Committee on Earth and Environmental Sciences.Our been twofold. In the Gulf Stream locale, the emphasis Changing Planet: The FY 1991 Research Plan ofthe U. S. Global has been on identifying existing data sources that Change Research Program. U.S. Geological Survey, 250 pp. would allow members of the science team to begin Cess, R. D., et aI., 1990: Intercomparison and interpretation of outlining research interests and issues for this locale. climate feedback processes in 19 atmospheric general circula­ These two sites were moved from primary to second­ tion models. J. Geophys. Res., 95, 16601-16 615. Chahine, M. T., 1992: The hydrological cycle and its influence on ary status in the summer of 1993. This decision allows climate. Nature, 359, 373-380. the program to focus its resources on the first three Clark, T. L., 1979: Numerical simulations with a three dimensional sites. The development of the ARCS concept for the cloud model. J. Atmos. Sci., 36, 2191-2215. Tropics, and its possible extension to the Arctic, sug­ Corbet, J. M., and C. Mueller, 1991: Zeb: Software fordata integra­ gests that a modular cloud and radiative experimental tion, display, and analysis. Proc. of the 25th Int. AMS Conf. on RadarMeteorology, Paris, France, Amer. Meteor. Soc. capability may be more readily deployable than origi­ Cubasch, U., and R. D. Cess, 1990: Processes and modeling. In nally thought. This would make short-term occupation Climate Change: The IPCC Scientific Assessment. J. T. of the secondary sites more practical than classic Houghton, G. J. Jenkins, and J. J. Ephraums, Eds. Cambridge campaign approaches may have allowed. University Press, 69-91. Dabberdt, W. F., et al. 1994: Advances in the Development of an Integrated Data Assimilation and Sounding System. Proceed­ ings of the Third Atmospheric Radiation Measurement Science 9. Summary Team Meeting, U.S. Dept. of Energy, CONF-9303112. 143­ 145. The ARM Program is creating a set of observational Doran, J. C., et aI., 1992: The Boardman regional flux experiment. facilities that should allow the research community to Bull. Amer. Meteor. Soc., 73, 1785-1795. Ellingson, R., and Y. Fouquart, 1991: The intercomparison of focus its efforts on the high priority problems of radiation codes in climate models: An overview. J. Geophys. understanding clouds and radiation in the climate Res., 96, 8925-8927. system. The observational facilities of the ARM Pro­ Hoffert, M. I., A. J. Callegari, and C. T. Hsieh, 1991: The role of deep gram represent a unique resource for meteorological sea heat storage in the secular response to climatic forcing. J. research not only because of the large complement of Geophys. Res., 85, 666-667. JASON, 1992: Small Satellites and RPAs in Global-Change Re­ equipment deployed over an extended period of time search. The MITRE Corporation, McLean, VA, 200 pp. but also because of their close relationship with exist­ Kogan, Y. L., 1991: The simulation of a convective cloud in a 3-D ing facilities and coordination with other field pro­ model with explicit microphysics. Part I: Model description and grams. The deployment of ARM to the first three sensitivity experiments. J. Atmos. Sci., 48, 1160-1189. locales-the southern Great Plains, the tropical west­ Lindzen, R. S., 1990: Some coolness concerning global warming. Bull. Amer. Meteor. Soc., 71,288-299. ern Pacific, and the North Slope of Alaska-should Mitchell, J. F. B., C. A. Senior, and W. J. Ingram, 1989: CO2 and provide a sound base for planning other research climate: A missing feedback. Nature, 341, 132-134. programs in these climatologically interesting and Ramanathan, V., R. D. Cess, E. F. Harrison, P. Minnis, B. R. important areas. Barkstrom, E. Ahmad, and D. Hartman, 1989: Cloud-radiative

1220 Vol. 75, No.7, July 1994 forcing and climate: Results from the earth radiation budget --, 1991: Aerosols, clouds and radiation. Atmos. Environ., 25A, experiment. Science, 243, 57-63. 2435-2442. Randall, D. A., 1989: Cloud parameterization for climate modeling: U.S. Department of Energy, 1986: Projecting the climatic effects of Status and prospects. Atmos. Res., 23, 345-361. increasing carbon dioxide. State-of-the-Art Report, Washington, --,1992: Global climate models: Whatand how, in global warming. D.C., 381 pp. Physics and Facts, B. G. Levi, D. Hafemeister, and R. Scribner, --, 1990: Atmospheric radiation measurement program plan, Eds., American Institute of Physics, 24-45. DOE/ER-0442 (Executive Summary, 19 pp) and0441 (116 pp), Rassmussen, et aI., 1992: Winter icing and storms project (WISP). U.S. Department of Energy, Washington, D.C. Bull. Amer. Meteor. Soc., 73, 951-974. --,1991, April 1991 : Identification, recommendation, and justifi­ Schneider, J. M., P. J. Lamb, and D. L. Sisterson, 1993: Site cation of potential locales for ARM sites. U.S. Department of scientific mission plan for the southern great plains CART site, Energy Report 0494T (Executive Summary, 14 pp) and 0495T, July-December 1993. U.S. Department of Energy Rep.ort, ARM­ 160 pp. 93-001,34 pp. --, 1993: Global change research: Summaries of research in FY Simmons, A. J., and L. Bengtsson, 1988: Atmospheric general 1993. U.S. Department of Energy Report, 0494T (Executive circulation models: Their design and use for climate studies. Summary, 14 pp) DOE/ER-0597T, 216 pp. Physically-Based Modelling and Simulation of Climate and Cli­ WCRP, 1992: World Climate Research Programme. Scientific Plan matic Change. Vol. 2, Nato ASI Series, M. Schlesinger, Ed., for the GEWEX Continental Scale International Project (GCIP). Kluwer Academic, 627-652. World Meteorological Organization; World Climate Program Twomey, S., 1977: Atmospheric Aerosols. Elsevier, 302 pp. Research Series, WCRP-67 WMOITD No. 461,85 pp.

The Representation of Cumulus Convection •In Numerical Models Meteorological Monograph No. 46

Cumulus convection is perhaps the most complex and perplexing subgrid-scale process that must be represented in numerical models of the atmosphere. It has been recognized that the water vapor content of large parts of the atmosphere is strongly controlled by cloud microphysical processes, yet scant attention has been paid to thiS problem in formulating most existing convection schemes. This monograph is the frUit of the labors of many of the leading specialists in convection and convective parameterization to discuss this and other issues. Its topics include: an overview of the problem; a review of "classical" convection schemes in Widespread use; the special problems associated with the representation of convection in mesoscale and climate models; the parameterization of slantwise convection; and some recent efforts to use explicit numerical simulations of ensembles of convective clouds to test cumulus representations. American ©1994 American Meteorological Society. Hardbound, B&W, 246 pp., $65 list\$45 members Meteorological lincludes shipping and handlingl. Please send prepaid orders to: Order Department, American Meteorological Society, 45 Beacon St., Boston, MA 02108-3693. Society

Bulletin of the American Meteorological Society 1221