18th Conference on Weather and Forecasting, 14th Conference on Numerical Weather Prediction, and Ninth Conference on Mesoscale Processes

Wednesday, 1 August 2001: 11:10 AM
Brent L. Shaw, NOAA/FSL, Boulder, CO; and E. R. Thaler and E. J. Szoke
Poster PDF (363.3 kB)
Recently, the NOAA Forecast Systems Laboratory (FSL) has developed a new, enhanced version of the Local Analysis and Prediction System (LAPS). The revised version generates real-time, dynamically balanced analyses of the atmosphere using all available data (including satellite imagery and radar data) and includes an analysis of cloud and precipitation species within the three-dimensional volume. This new analysis scheme is being used to initialize a slightly modified version of the NCAR/PSU MM5 model in a "hot start" mode. That is, at the model initialization time, full and accurate analyses of rain, snow, cloud liquid, cloud ice, and precipitating ice are inserted into the initial conditions along with the LAPS balanced wind, temperature, and humidity fields. The goal of this initialization scheme is to improve the short-range predictability of clouds and precipitation by reducing or even eliminating the model spin-up problem without having to resort to much more computationally intensive model initialization techniques. The details of the analysis technique and quantitative verification of the forecast results are covered in related papers.

Although quantitative verification is a useful tool to evaluate a model, an equally important component is how useful the model is to the operational forecaster. For example, a state-of-the-art NWP model may provide the most accurate forecast, but may be so computationally expensive that it is not available in a timely matter. Because the FSL scheme described above is relatively inexpensive computationally, such a configuration could be run in a local office to complement the NCEP guidance. However, this would only be worthwhile if the system is deemed by the operational forecasters as adding value to their forecast process. Since the Boulder NWS WFO is collocated with FSL, a collaborative effort ensued to do such an evalation. Since December 2000, the Boulder WFO has been receiving the grids from the FSL real-time system. The model produces a 24-hour forecast every six hours and is typically available to the forecaster approximately 2.5 hours after the initial conditions valid time. Thus, accuracy of the short term forecast is paramount.

Thus far, feedback from the WFO has been extremely positive. The ability of the forecasters to display and interact with the forecast grid on their operational AWIPS system has been a critical factor in obtaining meaningful feedback. This paper presents the details of the model configuration and how the model is currently being used within the operations.

Additionally, anecdotal evidence of its strengths and weaknesses as well as actual forecast examples will document these results. The preliminary conclusion is that a system patterned after the FSL prototype is not only feasible, but can also provide significant benefits to a local forecast office.

Supplementary URL: