Development of a 103-year high-resolution climate data set for the conterminous United States

1. INTRODUCTION Currently, the only high-quality, high-resolution temperature and precipitation data sets for the continental United States suitable for use on climatological time scales are for mean values. None yet exist that represent sequential monthly values over an extended historical period. Such data sets would enable, for example: transient ecological, hydrological, and natural resource modeling for use in global change assessment; analysis of local and regional climate variations; analysis of frequency, duration, and spatial patterns of extreme climatological events; and investigation of relationships between climatological variability and large-scale forcing mechanisms (e.g., ENSO or QBO). A small number of gridded sequential data sets are under development, but they are at a coarse resolution or have a limited spatial extent. The only long-term sequential data sets currently available for the U.S. are for individual sites or are based on simple averages of any available stations within major climatic divisions. The site data are sparse and located primarily in valley bottoms, and the climatic division data suffer from inherent biases in sensor placement and availability over time. The current project, under NOAA/NASA funding, involves development of 103 years (1895-1997) of gridded monthly precipitation and maximum and minimum temperature at a 4-km resolution for the contiguous United States. PRISM, a knowledge-based interpolation system was used to produce the gridded climate datsets (Daly et al. 1994, 1997, 2002). The project involves development of serially complete monthly data for approximately 8,000 stations in the U.S. An important part of this work has been to develop and apply a semi-analytical, spatially-intelligent quality control system, based on PRISM, for monthly station observations.