JP1.7 A model-based high resolution temperature cllimatology for New York

Monday, 20 June 2005
Brian N. Belcher, Northeast Regional Climate Center, Ithaca, NY; and A. T. DeGaetano

Archived Rapid Update Cycle (RUC) Model initializations are used as a basis to develop high-resolution gridded daily maximum and minimum temperature climatologies for New York State. Since Cooperative Network station data are not used in the RUC initializations, these observations provide a means to ground-truth the model data. First, the gridded RUC temperatures are adjusted to the elevation of each Cooperative Network station in the spatial domain. This adjustment is based on the lapse rate specified in each hourly RUC initialization. Then a spline is fit to the set of 24 hourly elevation-adjusted temperature to obtain daily maximum and minimum temperatures analogous to those measured by a Cooperative observer. The daily maximum (and separately minimum) temperatures at each 40 (or 20) km RUC grid are then interpolated to the location of the Cooperative network station using a multi-quadric interpolation procedure and the bias (model-observed) computed. This process is repeated for all Cooperative stations within the spatial domain of the climatology. The resulting bias field is then interpolated back to each RUC grid, adjusting the model data based on the independent daily Cooperative Network data. In application, we hope to produce gridded temperature climatologies at a 1 km resolution. Here, the procedure would be similar, with the elevation-adjusted RUC data interpolated to digital elevation model grids and the Cooperative Network adjustments applied to these points. To test this methodology, cross-validation is applied by withholding a Cooperative Network station from the bias field and subsequently estimating the known temperature based on the interpolation techniques. This produces favorable results, with little interpolation bias and mean absolute errors near 1°C. The procedure can be improved further, particularly on days with strong radiation cooling, by applying a secondary adjustment related to the mean model bias at surrounding stations with similar topographic characteristics.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner