A long-term, consistent, high-resolution climate dataset for the North American domain, as a major improvement upon the earlier global reanalysis datasets in both resolution and accuracy, is presented.

[1] Results are presented from the multi-institution partnership to develop a real-time and retrospective North American Land Data Assimilation System (NLDAS). NLDAS consists of (1) four land models executing in parallel in uncoupled mode, (2) common hourly surface forcing, and (3) common streamflow routing: all using a 1/8°grid over the continental United States. The initiative is largely sponsored by the Global Energy and Water Cycle Experiment (GEWEX) Continental-Scale International Project (GCIP). As the overview for nine NLDAS papers, this paper describes and evaluates the 3-year NLDAS execution of 1 October 1996 to 30 September 1999, a period rich in observations for validation. The validation emphasizes (1) the land states, fluxes, and input forcing of four land models, (2) the application of new GCIP-sponsored products, and (3) a multiscale approach. The validation includes (1) mesoscale observing networks of land surface forcing, fluxes, and states, (2) regional snowpack measurements, (3) daily streamflow measurements, and (4) satellite-based retrievals of snow cover, land surface skin temperature (LST), and surface insolation. The results show substantial intermodel differences in surface evaporation and runoff (especially over nonsparse vegetation), soil moisture storage, snowpack, and LST. Owing to surprisingly large intermodel differences in aerodynamic conductance, intermodel differences in midday summer LST were unlike those expected from the intermodel differences in Bowen ratio. Last, anticipating future assimilation of LST, an NLDAS effort unique to this overview paper assesses geostationary-satellite-derived LST, determines the latter to be of good quality, and applies the latter to validate modeled LST.

[1] Three objective techniques used to obtain gauge-based daily precipitation analyses over global land areas are assessed. The objective techniques include the inverse-distance weighting algorithms of Cressman (1959) and Shepard (1968), and the optimal interpolation (OI) method of Gandin (1965). Intercomparisons and cross-validation tests are conducted to examine their performance over various parts of the globe where station network densities are different. The gauge data used in the examinations are quality controlled daily precipitation reports from roughly 16,000 stations over the global land areas that have been collected by the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center (CPC). Data sources include daily summary files from the Global Telecommunication System (GTS), and the CPC unified daily gauge data sets over the contiguous United States (CONUS), Mexico, and South America. All three objective techniques are capable of generating useful daily precipitation analyses with biases of generally less than 1% over most parts of the global land areas. The OI method consistently performs the best among the three techniques for almost all situations (regions, seasons, and network densities). The Shepard scheme compares reasonably well with the OI, while the Cressman method tends to generate smooth precipitation fields with wider raining areas relative to the station observations. The quality of the gauge-based analyses degrades as the network of station observations becomes sparser, although the OI technique exhibits relatively stable performance statistics over regions covered by fewer gauges.