Station Density Strategy for Monitoring Long-Term Climatic Change in the Contiguous United States

Abstract

The National Oceanic and Atmospheric Administration is establishing the U.S. Climate Reference Network (CRN) to improve the capacity for observing climatic change and variability. A goal of this network is to provide homogeneous observations of temperature and precipitation from benchmark stations that can be coupled with historical observations for detection and attribution of climatic change. The purpose of this study was to estimate the number and distribution of U.S. CRN observing sites. The analysis was conducted by forming hypothetical networks from representative subsamples of stations in an existing higher-density baseline network. The objective was to have the differences between the annual temperature and precipitation trends computed from reduced-size networks and the full-size networks not greater than predetermined error limits. This analysis was performed on a grid cell basis to incorporate the expectation that a greater station density would be required to achieve the monitoring goals in areas with greater spatial gradients in trends. Monte Carlo resampling techniques were applied to stations within 2.5° latitude ? 3.5° longitude grid cells to successively lower the resolution compared to that in the reference or baseline network. Differences between 30-yr trends from lower-resolution networks and full-resolution networks were generated for each grid cell. Grid cell densities were determined separately for temperature and precipitation trends. In practice densities can be derived for any parameter and monitoring goal. A network of 327 stations for the contiguous United States satisfied a combined temperature-trend goal of 0.10°C decade?1 and a precipitation-trend goal of 2.0% of median precipitation per decade.