Design and Evaluation of Regional Weather Monitoring Networks

ABSTRACT IRRIGATION scheduling on a regional level is often supported by a weather monitoring network for estimating reference evapotranspiration, E^r- The estimation error for E^^ can be used as a quantitative basis for designing new networks or modifying exisiting networks. Kalman filtering provides a convenient means for both estimating E^^ in multistation networks and quantifying the estimation error. The procedure explicitly considers spatial and temporal correlation of the E^r stochastic process as well as measurement error. Application of the Kalman filtering approach to network design and evaluation is illustrated using an existing network in Colorado. The procedure is used to develop contour maps of estimation error for proposed network configurations. These maps are then used to evaluate adequacy of station density and locations.