3.5 Evaluation of NOAA/NCEP's North America Mesoscale (NAM) 12-km and 4-km High-Resolution Nest (NAM4) Forecast for a typical Southern Temperate Deciduous Forest

Tuesday, 24 January 2017: 5:00 PM
612 (Washington State Convention Center )
William Pendergrass, NOAA/OAR/ARL/ATDD, Oak Ridge, TN; and J. McQueen, G. Dimego, and M. Ek

An important aspect of the application of weather forecasts is the assessment and incorporation of the uncertainties associated with the forecast guidance. Decreasing model grid resolution to minimize model uncertainties does not necessarily translate to a better, or more accurate, forecast. Given that there is an underlying unsolvable variability, the dominant question then becomes how to determine to what extent increased deterministic detail improves the accuracy of forecasts.

The intensively instrumented NOAA/ATDD Chestnut Ridge forest-meteorology tower located in Oak Ridge, TN provided an opportunity to examine the performance of NOAA/NCEP’s 12km North America Model (NAM12) and 4-km High Resolution Nest numerical weather predictions against observations from the Chestnut Ridge tower. Above canopy observed winds, temperatures, heat and momentum fluxes were compared against hourly meteorological parameter fields generated from the 12z update cycle for both the 12km and 4km forecast products over the period 5/20/2016 through 7/19/2016. For the full dataset, the analysis suggests a mean absolute 10m wind speed error for the 12-km NAM forecast of 0.95 m/s; the mean absolute 10m wind speed error for the 4-km NAM Nest was 0.91 m/s. Analysis of 2m temperature errors found little statistical difference between 12-km and 4-km models; nighttime mean absolute errors roughly 1.25 0C with daytime error peaking at 3.5 oC. Drag coefficients between the two models were quite comparable and typically with 25% of observed values with the exception of the morning transition period when the relative difference between models and observations exceeded 250%.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner