The Prediction of Rapidly Deepening Cyclones by NMC's Nested Grid Model: Winter 1989–Autumn 1991

View More View Less
  • 1 Meteorological Operations Division, Forecast Branch, National Meteorological Center, Camp Springs, Maryland
  • | 2 Meteorological Operations Division, Techniques Development Group, National Meteorological Center, Camp Springs, Maryland
© Get Permissions
Full access

Abstract

A study of rapidly deepening cyclones (RDC) produced by the National Meteorological Center's (NMC) Nested Grid Model (NGM) was conducted over a three-year period from the winter of 1988/89 through the autumn of 1991. The axis of RDCs was observed over the western Atlantic 0cean from off the mid-Atlantic coast northeastward to the southern tip of Greenland, and a smaller axis was observed over the eastern Pacific over the Gulf of Alaska and southern Alaska. Relatively few RDCs occurred over the eastern Pacific during the winter and spring, with the maximum distribution occurring during the autumn.

The ability of the NGM to forecast RDCs was a function of both the forecast length and the season. The NGM's ability to simulate RDCs was greatest at 12 h and steadily decreased with forecast length. Similarly, the false alarm rate (FAR) was lowest at 12 h and steadily increased with forecast length through 48 h. The ability of the NGM to detect RDCs was greatest in the winter and decreased during the spring and fall.

There was a smaller decrease in probability of detection in the middle of the forecast cycle during the winter and spring that may be attributed to spinup problems in the NGM during the initialization of the model. Similar trends in the FAR were noted, with a decrease in FAR beyond 12 h during the winter seasons.

The results from this study showed the NGM was too slow to deepen RDCs at all 12-h forecast periods, with the pressure errors increasing with forecast length. The NGM also had a cold bias in the 1000–500-mb thickness forecasts over the RDCs. However, the NGM showed exceptional skill in correctly forecasting the sign of the 12-h pressure change for the RDCs. During the three winter seasons the NGM rarely misforecast the sign of the 12-h forecast pressure change during rapid cyclogenesis.

Over the western Atlantic the NGM was too slow to move the RDCs to the east The overall position errors for RDCs were approximately 10% smaller than the position errors for all cyclones in the NGM at all forecast periods.

An examination of two RDC events revealed significant differences in the NGM's ability to forecast the rapid deepening. During the ERICA IOP 4 cyclone, the NGM forecast the cyclone fairly well, showing its bias of being too slow to deepen the RDCs and too slow to move it eastward. Much poorer skill in forecasting the 4 January 1992 cyclone off the coast of the Carolinas occurred with the NGM having significant problems resolving subgrid-scale processes as the storm deepened rapidly as it crossed the warm Gulf Stream waters.

Abstract

A study of rapidly deepening cyclones (RDC) produced by the National Meteorological Center's (NMC) Nested Grid Model (NGM) was conducted over a three-year period from the winter of 1988/89 through the autumn of 1991. The axis of RDCs was observed over the western Atlantic 0cean from off the mid-Atlantic coast northeastward to the southern tip of Greenland, and a smaller axis was observed over the eastern Pacific over the Gulf of Alaska and southern Alaska. Relatively few RDCs occurred over the eastern Pacific during the winter and spring, with the maximum distribution occurring during the autumn.

The ability of the NGM to forecast RDCs was a function of both the forecast length and the season. The NGM's ability to simulate RDCs was greatest at 12 h and steadily decreased with forecast length. Similarly, the false alarm rate (FAR) was lowest at 12 h and steadily increased with forecast length through 48 h. The ability of the NGM to detect RDCs was greatest in the winter and decreased during the spring and fall.

There was a smaller decrease in probability of detection in the middle of the forecast cycle during the winter and spring that may be attributed to spinup problems in the NGM during the initialization of the model. Similar trends in the FAR were noted, with a decrease in FAR beyond 12 h during the winter seasons.

The results from this study showed the NGM was too slow to deepen RDCs at all 12-h forecast periods, with the pressure errors increasing with forecast length. The NGM also had a cold bias in the 1000–500-mb thickness forecasts over the RDCs. However, the NGM showed exceptional skill in correctly forecasting the sign of the 12-h pressure change for the RDCs. During the three winter seasons the NGM rarely misforecast the sign of the 12-h forecast pressure change during rapid cyclogenesis.

Over the western Atlantic the NGM was too slow to move the RDCs to the east The overall position errors for RDCs were approximately 10% smaller than the position errors for all cyclones in the NGM at all forecast periods.

An examination of two RDC events revealed significant differences in the NGM's ability to forecast the rapid deepening. During the ERICA IOP 4 cyclone, the NGM forecast the cyclone fairly well, showing its bias of being too slow to deepen the RDCs and too slow to move it eastward. Much poorer skill in forecasting the 4 January 1992 cyclone off the coast of the Carolinas occurred with the NGM having significant problems resolving subgrid-scale processes as the storm deepened rapidly as it crossed the warm Gulf Stream waters.

Save