Model Verification of Short Range High Impact Weather in Central Florida Christopher Hicks Department of Marine and Environmental Systems Florida Institute of Technology Melbourne, Florida 32901 July 18th, 2007
Outline • Motivations • Objectives • Methods • Background Info • The NAM-218 model • The weather stations • Results Analysis • Temperature Comparison • Precipitation Comparison • Summary • Questions
Motivations • The predictability of current operational models need to be evaluated. • To figure out the difference of inland and coastal forecasts in the model.
Why Forecast Models Are Used? • Models are able to provide guidance for weather forecasters. • There’s also less chance of error in the calculations of forecast models. • With supercomputers, running forecast models can take a very short time.
Objectives • The North American Mesoscale (NAM-218) model’s ability to predict high impact weather in Central Florida. • Compare model data to observed weather data around Central Florida. • Discover how well the NAM-218 model is able to predict temperature and precipitation. • Examine whether the model is better at predicting weather events along coastal regions or inland regions of Florida.
Methods • Collected archive data from NOAA’s National Operational Model Archive & Distribution Systems (NOMADS) site. • The forecast model that was used is the 12 UTC (8:00 AM EDT) run on June 18, 2007 which goes out to 8:00 PM EDT on June 21, 2007. • Total 84-hour forecast with 3-hour interval was used to compare to the observed data that was collected.
The North American Mesoscale Model (NAM-218) • The NAM-218 is a forecast model that is widely used here in the United States. • Uses a grid forecasting model with a resolution of 12 km (7.5 miles) which makes it one the highest resolution models used in the United States. • Run by the National Centers for Environmental Prediction. • Makes an 84 hour forecast 4 times a day (00, 06, 12, 18 UTC).
Weather Stations Used in Project • Weather stations set up by Florida Tech: • Satellite High School (Satellite Beach, FL) • Harmony High School (Harmony, FL) • Deseret Ranch (Rural Eastern Osceola County) • Fay Park (Port St. John, FL) • Permanent weather stations: • Vero Beach, FL (KVRB) • Orlando, FL (KMCO) Deseret Ranch Weather Station (Picture courtesy of Kate Howard)
Temperature at Satellite High • Satellite High is less than a mile from shoreline. • The NAM forecast is slightly cooler than observed by about 2 to 3 degrees for the first 48 hours.
Temperature at Vero Beach • Vero Beach is a little farther inland than Satellite High but still close to the shoreline. • Difference between forecast and observed is about 2 degrees for the daily high and about 7 degrees for the overnight low in the first 24 hours.
Temperature at Harmony High • Temperature forecast for Harmony High improves in the early part of the forecast window compared to both Satellite High and Vero Beach. • Forecasted diurnal high for the 19th is off by only 1 degree. • Accuracy of model still degrades the farther into the future it goes.
Temperature at Orlando • The model’s forecast for Orlando becomes even more better than Harmony High School. • Forecast is only off by a degree or less in the first 48 hours. • In this case, the data supports the idea that model forecasts for temperature improves as farther inland you go.
Precipitation at Satellite High • Precipitation that occurs on the 20th is underestimated by .6 inches. • The model predicted the rain event occurring approximately 6 hours earlier than the observed.
Precipitation at Vero Beach • Precipitation for the 20th only off by about a tenth of an inch. • However the forecast model predicted later than what was observed, instead of earlier.
Precipitation at Harmony High • Forecast is very close to two of the three forecast day that have rain. • After 48 hours, timing of the rain events degrades.
Precipitation for Orlando 0.02 inches observed • Forecast for Orlando was off by over 2.5 inches for the 20th of June. • Observed precipitation levels for Orlando significantly lower than other three weather stations. • Overall, it’s possible that the model misinterpreted where the thunderstorm convection would go and how fast it moved.
Summary • In this case, the model is able to predict temperature for inland area much better than coastlines. • Precipitation is still misinterpreted for all regions no matter how close to the coast the weather station is. • Precipitation forecasts can be improved once thunderstorm development can be well simulated for the region. • According to this case, the NAM-218 is good for short range (0-48 hours) forecasts in the Central Florida region.
References • NOAA National Operational Model Archive & Distribution System (2007) http://nomads.ncdc.noaa.gov/data/meso-eta-hi/200706/20070618/ • Wikipedia (2007) North American Mesoscale Model. http://en.wikipedia.org/wiki/North_American_Mesoscale_Model • Weather Underground (2007) History. http://www.wunderground.com/history/airport/KVRB/2007/6/18/DailyHistory.html • Google (2007) Google Earth. http://earth.google.com • University of Utah Department of Meteorolgoy (2007) MesoWest Data http://www.met.utah.edu/mesowest/