The cloud field produced in a grid box of a global climate model by such parameterization equations is obviously a crude simplification of the real clouds. It cannot capture either the spatial structure at subgrid scale or the associated variability of cloud properties. Therefore, the calculation of radiative transfer through the clouds is also simplified. So the first question that we need to answer is: Given a specified three-dimensional field of cloud properties, can we compute with sufficient accuracy the solar and terrestrial radiative flux transfer and associated atmospheric heating rates through the clouds? A so-called soda-straw view of the radiative transfer problem provides the conceptual framework for gathering data to answer the question. One imagines an isolated hollow straw, a few kilometers in diameter, vertically spanning the atmosphere from its top to the ground. This kind of conceptual sampling corresponds to the sparse sampling character of the available data. One simply can't cover the ground densely with expensive measuring facilities. Solar radiation that hits the top of a straw is either reflected back, transmitted all the way through to the bottom of the straw, or absorbed along its length. In the opposite direction, radiation from Earth's surface is likewise either reflected, absorbed, or transmitted by the straw. How the radiation hitting a conceptual straw at either end is partitioned among reflection, absorption, and transmission is a function of the atmosphere's state. The radiatively significant atmospheric constituents are gas molecules, particles, and water in the form of cloud droplets, ice crystals, and precipitation.

The Lamont site is the central facility of the ARM observatory. It is designed to sample continuously all the components of the radiation budget at Earth's surface and all the relevant constituents in the atmosphere above the site. The atmospheric observables include the presence of clouds, their vertical extent, and their radiative properties. The extent of the clouds is measured with active sensors--radar and lidar. But measuring their radiative properties also requires the use of passive sensors. Beyond the question of our ability to compute radiation through clouds with sufficient accuracy lies a second, more difficult, question: Given the large-scale predicted or measured states of the atmosphere, can we predict the statistical properties of the local cloud fields? In other words, can we make the diagnostic connection between the large-scale atmospheric fields and the subgrid-scale clouds? The ARM program has two modeling approaches to that problem. The first we call single-column modeling. A conceptual column is much like a soda straw, except that its diameter, corresponding to a grid spacing, is several hundred kilometers wide. Because different altitudes in the atmosphere couple rapidly with each other through vertical convection and boundary-layer processes, most model cloud parameterizations are designed to treat each vertical column as an almost independent entity. Horizontal transport across the column's lateral boundary, by contrast, occurs in such models only by advection, that is, by large-scale horizontal wind.

Board of Writers:

Redactor : Thomas P. Ackerman and Gerald M. Stokes

Information Culler : Mohammad.R.Kamali