Search

A comprehensive series of global datasets for land-atmosphere models has been collected, formatted to a common grid, and released on a set of CD-ROMs. This paper describes the motivation for and the contents of the dataset. In June of 1992, an interdisciplinary earth science workshop was convened in Columbia, Maryland, to assess progress in land-atmosphere research, specifically in the areas of models, satellite data algorithms, and field experiments. At the workshop, representatives of the land-atmosphere modeling community defined a need for global datasets to prescribe boundary conditions, initialize state variables, and provide near-surface meteorological and radiative forcings for their models. The International Satellite Land Surface Climatology Project (ISLSCP), a part of the Global Energy and Water Cycle Experiment, worked with the Distributed Active Archive Center of the National Aeronautics and Space Administration Goddard Space Flight Center to bring the required datasets together in a usable format. The data have since been released on a collection of CD-ROMs. The datasets on the CD-ROMs are grouped under the following headings: vegetation; hydrology and soils; snow, ice, and oceans; radiation and clouds; and near-surface meteorology. All datasets cover the period 1987-88, and all but a few are spatially continuous over the earth's land surface. All have been mapped to a common 1 [degree] x 1 [degree] equal-angle grid. The temporal frequency for most of the datasets is monthly. A few of the near-surface meteorological parameters are available both as six-hourly values and as monthly means.

A physically based model is used to derive downward solar irradiance at the surface of the earth and surface albedo from Meteosat satellite measurements in the wavelength range between 0.40 and 1. 1 0 μm. The model takes into account Rayleigh and Mie scattering, water vapor and ozone absorption. No threshold setting is necessary to distinguish between clear and cloudy conditions, thereby avoiding the problem of its arbitrary nature and to some extent allowing quicker and easier data processing. Comparison of noontime satellite estimates with analogous hourly solar irradiance measurements obtained from the French pyranometer network shows a correlation coefficient of 0.92 and a rms error of 109 W m−2 (20% of the mean solar irradiance). The maximum error occurs for values of insulation around 300 W m−2 and can be mostly attributed to the different spatial and temporal samplings of the systems being compared. Mean monthly estimates of the solar irradiance between 1100 and 1300 UT, at different locations, fit pyranometer measurements with a rms error of 37 W m−2 (6%). The method is finally used to product maps of solar irradiance albedo and net solar flux at the surface over France. These maps can be easily derived from the whole Meteosat frame, with resolution ranging from a few kilometers to several dozens of kilometers, which are scales suitable for agrimeteorology and climate studies respectively. [ABSTRACT FROM AUTHOR]