Atmospheric Radiation Measurement Program



What is ARM?
Climate studies make use of what are known as General Circulation Models (GCMs) in an attempt to understand the processes that influence the earth's climate and, consequently, to evaluate the possible impact of human activity on our climate in the future. One of the major atmospheric science issues of today is the global warming question, and the role that clouds may play in controlling the earth's climate. If indeed the earth's surface temperature does rise over the next few decades, a result of an enhanced greenhouse effect due to the increase of greenhouse gases such as CO2, then one line of reasoning asserts that the higher surface temperatures will lead to an increase in evaporation from the earth's oceans. Increased evaporation means sending more water vapor into the atmosphere, which might in turn lead to an increase in cloudiness. If this increased evaporation leads to the formation of more low-level clouds, such as stratocumulus, this would tend to favor a decrease in surface temperatures as stratocumulus have a high albedo and thus reflect more solar radiation back to space than do other types of clouds. On the other hand, if the increase in atmospheric moisture leads to the formation of upper-level clouds, such as cirrus, then this would tend to increase surface heating since cirrus clouds are great absorbers of infrared radiation, but are mostly transparent to shortwave radiation. Normal amounts of solar radiation would thus continue to reach the earth's surface while there would simultaneously be an increase in infrared emitted by the cirrus back down to the surface. So, either a negative or positive cloud feedback could result depending on whether stratocumulus or cirrus clouds, respectively, end up being produced in greater amounts.

Since the nature of GCMs is such that they have a limited resolution (i.e. limited number of gridpoints) at which the state of the atmosphere is represented, a way must be found to include factors, such as clouds and radiation, that often occur on scales too small to be "seen" by the GCM and yet play major roles in influencing the earth's climate. Capturing and incorporating the effects of subgrid-scale atmospheric phenomena in terms of resolved variables is called parameterization. There are many approaches to parameterization which may be thought of as falling along a continuum, from a purely empirical approach on the one end to a theoretical one on the other. Using the former approach, one would empirically parameterize certain phenomena by based on a curve-fit to observed data. Then the parameterizations would be subjected to testing by comparing the output from the model into which they had been incorporated with observational data. In contrast, the theoretical approach involves a prior understanding of the physics behind the particular atmospheric phenomena to be described, deriving a suitable simple model depicting the physics, and then similarly testing this model as a parameterization by comparison of the model output with observed data.

There are, however, drawbacks to both these approaches. The large number of types of situations requiring parameterization make the empirical approach difficult to implement, while the physical approach can be hampered by an incomplete understanding of the processes involved. Despite the potential shortcomings of the physical parameterization process, it is this approach that is being used with increasing frequency as a means of representing certain atmospheric phenomena in GCMs.

In the quest to improve the performance and credibility of the models being used to study and predict climate change, two important, basic atmospheric processes are the focus of current parameterization efforts:

It is essential that these processes be accurately depicted in GCMs because of the critical role that they play in the earth's radiation budget, as well as the possibility that cloud properties may change as the climate changes. Therefore, a thorough understanding of atmospheric radiation and its interaction with cloud processes is of utmost importance.

Whatever the choice of parameterization schemes, the evaluation of the parameterization by comparison with observational data plays a crucial role in the testing phase, as alluded to above. Thus, the Atmospheric Radiation Measurement (ARM) program has the development and testing of radiation and cloud parameterizations as its overall goal, in an effort to improve on the understanding of processes that affect atmospheric radiation and the description of these processes in climate models. This goal is to be accomplished by the direct comparison of model calculations with a comprehensive set of field observations, obtained under a wide variety of meteorological conditions. The program is supported by the U.S. Department of Energy (DOE) and is an outgrowth of the United States Global Change Research Program (USGCRP; CEES 1990). Accordingly, ARM has been designed to meet the needs of scientific inquiry into areas of major concern, such as the role of clouds, identified by the USGCRP in its objective of understanding climate and hydrological systems.

The main objectives of ARM may be broken down into two areas of activity:

The ultimate objective is the incorporation of these parameterizations into GCMs. It is thus that ARM intends to meet the objective established by the DOE to improve GCMs and provide reliable simulations of regional and long-term climate change in response to increasing greenhouse gases.

There are many ways to parameterize atmospheric processes such as radiative transfer and cloud life-cycles, and, as has been stated, one of the goals of ARM is to sift through these parameterizations to evaluate their effectiveness at describing these processes and suitability for use in a GCM. Thus, a reliable source of meteorological data, obtained under a wide variety of atmospheric conditions, is an important part of this operation. This data would be used in several capacities; to supply input for initialization of the models, to serve as a check for the output of the models in predictive mode, and also to provide forcing for the models. Furthermore, the data should be approximately representative of the areal extent covered by a GCM grid cell, which is typically on the order of 200 km on a side. Finally, it was decided that this observational data should be made available on a continual basis, over the span of several years, while staying within labor and financial constraints. To meet each of these facets of the ARM program, it was decided the labor should be divided up into two spheres of activity. Therefore, the branch of ARM responsible for carrying out the research phase of the project is the Science Team. Their duties encompass the actual process of development and testing of models and parameterizations, as well as instrument development and testing. On the other hand, the Cloud and Radiation Testbed (CART) personnel oversee the set-up and maintenance of the instruments at the observation site, collect and process the raw data, and check to see that the experimental requirements of the members of the Science Team were being met. An explanation of the CART site setup will now be given.

The function of a CART site is to provide the source of field data necessary to meet the goals of the Science Team. As previously mentioned, these goals essentially encompass

Therefore, an empirical data set is required for initializing and running the models, and then comparing the model results to observed data. This data set needs to include both longwave and shortwave radiation fluxes; turbulent fluxes of heat, moisture and momentum; the distribution of radiatively significant particulates, aerosols and gases; cloud types, composition and distribution; a complete thermodynamic description of the air mass; surface fluxes of heat and moisture; and finally any processes (such as precipitation/evaporation or generation of cloud condensation nuclei) that might have an impact on these variables. Additionally, an efficient ingest and archival system for the data streams is needed for distribution of ARM data to the scientific community. An analysis of the Science Team needs, which is referred to as General Measurement Strategies (GMS), was drawn up to specifically address these issues, and has served as a guide for designing and selecting the appropriate CART site.

With the objective of studying radiation transfer in the atmosphere and atmospheric processes that influence it, the choice of a CART site was to be made in light of several radiation-influencing factors, such as latitude/ longitude, altitude, terrain/surface, cloud frequency/type, precipitation, temperature and humidity. Once several potential sites had been identified, three additional criteria were established in the GMS to help narrow down the pool of CART site candidates. The first of these criteria was that the site should exhibit a great temporal variability in terms of weather conditions. Second, ideally the CART site would experience as much atmospheric variation as possible while remaining logistically accessible; i.e. the region would be capable of supplying the roads, power, communications and living accommodations necessary to keep such a site functioning. At the same time, the CART site should be located far enough away from urban areas so that it would not be negatively influenced by urban-generated factors such as air pollution or the heat-island effect. Lastly, if at all possible, it should be situated in proximity to other agencies or projects similarly involved in atmospheric measurements so that constructive interaction might take place.

Although the ARM project calls for the eventual installation of three CART sites at diverse locations around the globe, geographically and climatologically speaking, there is currently only one CART site in operation, and it is located near the town of Lamont in northeastern Oklahoma. The Southern Great Plains (SGP) region of the United States was chosen for this prototype CART site since it meets many of the requirements outlined above. Due to its interior location, it has a continental-type climate and thus meets the criterion of high seasonal ranges in temperature, humidity, and precipitation. Additionally, it is well suited for the purposes of cloud parameterization research since the SGP region experiences deep and vigorous convection in the spring and summer seasons. Further, the terrain is spatially homogeneous which reduces complexities introduced by discontinuities such as mountains and coastlines in model testing. It is not far from urban areas, yet far enough to be relatively undisturbed by the problems associated with them. Finally, there are several on-going projects which offer the possibility of beneficial collaboration in the region:

The CART site itself consists of a central facility, an auxiliary station network, an extended observation network, and boundary facilities as shown in the accompanying figure. The observations at the central facility aim to construct as detailed a characterization of the atmospheric column above the facility as possible. Consequently, more data is collected at the central facility than elsewhere on the CART site, with instruments installed there taking measurements of upwelling and downwelling radiation, cloud fractional coverage, cloud-base altitude, liquid water path, local surface reflectance, temperature and emissivity. Additionally, radiosondes provide atmospheric temperature, humidity, wind speed and direction above the central facility.

Next, in order to gain a three-dimensional perspective on radiative transfer processes over the central facility, the auxiliary stations network supply additional pertinent information through the use of all-sky cameras and ceilometers within a 20 km radius of the central facility. Then, surrounding both the central facility and the auxiliary network is the extended observing network. Its purpose is to obtain radiometric and meteorological information in addition to surface flux data covering an area roughly equivalent to a GCM grid cell. Using this data, additional understanding of radiative transfer processes over the entire area may be used in evaluating GCM parameterizations. Finally, the boundary facilities provide information on the vertical profile of horizontals fluxes of moisture and temperature through the simultaneous launching of radiosondes. Currently, the boundary facilities form a diamond approximately 200 km on a side, and the launch sites are collocated with wind profilers (the original configuration was a triangle). One of the goals of the ARM Experiment Support Team has been to combine the profiler data with those of the radiosondes through objective analysis, to compensate for drift and tracking errors in the radiosonde wind data.

What is an SCM?
The Single Column Model (SCM) is a single vertical array of gridpoint cells taken from a GCM. This GCM subset is run as a model in and of itself, with certain parameters, such as large-scale rising motion, being prescribed by necessity in the SCM (this would naturally arise from the governing equations in a full 3-dimensional GCM). However, it is this very feature makes the SCM a convenient vehicle for testing a particular parameterization scheme. This is because certain atmospheric processes can be studied in an SCM which would otherwise be difficult or impossible to isolate in the 3-dimensional GCM.

In addition to the prescribed boundary conditions mentioned above, the types of data needed as input for an SCM also include initial values of the prognostic variables within the cell, such as temperature, humidity, wind velocity, cloud fraction, planetary boundary layer depth, radiation field, precipitation and soil moisture. Other boundary conditions besides large-scale vertical motion are thermodynamic variables at the lateral boundaries, surface emittance and reflectance at the lower boundary, and the top of the atmosphere solar flux. The connection with the CART site is now evident: it supplies the necessary data for the boundary and initial conditions to run an SCM, and for parameterization evaluation by comparison with the actual state of the atmosphere through tracking the evolution in time of the model's prognostic variables.

What is an IOP?
Under normal operations, the ARM observational strategy calls for continuous atmospheric observational data to be provided by the CART site. In addition, supplemental data is provided during time spans of limited duration known as Intensive Observation Periods (IOPs). Examples of the kinds of data that are provided on a continuous basis would be upwelling and downwelling solar and terrestrial radiation; column-integrated precipitable water vapor and liquid water path; brightness temperature; cloud cover, type, and elevation; sensible and latent heat flux at the surface, plus surface temperature, pressure, precipitation, soil moisture, wind speed and direction. Much of these data are provided by remote sensors at the central facility as well as the extended/auxiliary/boundary network locations. However, the supplemental data are furnished by observations that are too expensive to be implemented continuously and/or require additional personnel to be performed (such as frequent radiosonde launches). These observations are instead conducted periodically throughout the year during IOPs according to a schedule agreed upon by the Science Team.

The purpose of an IOP is two-fold