Effects of Synoptic Surveillance on model forecasts for
16 September 2004 0000 UTC
Ivan.


Quicktime movie of AVN DLM wind model forecast
Quicktime movie of DLM wind data increment


TRACK AND INTENSITY PERFORMANCE

TRACK (km)
MODEL12 h24 h36 h48 h60 h72 h84 h96 h108 h120 h
AVNN 38 111011534174965076159801095
AVNO 24 14 70 74253389340506740847
%IMP37%-27%31%52%39%22%33%18%24%23%
GFNO 36 7522832758987390210301326
GFDL 44 8620930756080189910561420
%IMP-26%-15% 8% 6% 5% 8% 0%-3%-7%
INTENSITY (kt)
MODEL12 h24 h36 h48 h60 h72 h84 h96 h108 h120 h
AVNN-10 -3 2 1 5-17-21-24-24-18
AVNO -4 -1 2 -1 5-17-18-20-14-15
%IMP60%67% 0% 0% 0% 0%14%17%42%17%
GFNO 23 16 11 17 25 4 -1 -9-12
GFDL 22 18 10 17 27 6 0 -7 -9
%IMP 4%-13% 9% 0%-8%-50%100%22%25%

Landfall forecasts

16/0650 UTC 30.2N 87.9W Near Pine Beach, AL 7.0 h into the forecast
MODELLATLONTIMEERRORLOCATION
AVNN30.3988.18 8.5 34.2Dauphin Island, AL
AVNO30.2888.12 7.5 22.9Dauphin Island, AL
%IMP 67% 33%
GFNO30.3388.19 7.0 31.4Dauphin Island, AL
GFDL30.2688.30 7.0 39.0Dauphin Island, AL
%IMP 0% -24%

Figure 1. NCEP 850 - 200 hPa mean wind analysis for 16 September 2004 0000 UTC (Hurricane Ivan).

Figure 2. Track forecasts for the no dropwindsonde (NO or NN) and the all dropwindsonde (AL) runs for the AVN and GFDL models initialized on 16 September 2004 0000 UTC.

Figure 3. Track forecasts for the no dropwindsonde (NO or NN) and the all dropwindsonde (AL) runs for the AVN and GFDL models initialized on 16 September 2004 0000 UTC.

Figure 3. Track forecasts for the no dropwindsonde (NO or NN) and the all dropwindsonde (AL) runs for the AVN and GFDL models initialized on 16 September 2004 0000 UTC.

Figure 3. NCEP Global Ensemble Predction System track forecasts for all cyclones in the Atlantic basin initialized from 14 September 2004 0600 UTC to 15 September 2004 0000 UTC, showing the tracks of Hurricane Ivan.

Figure 4. Ensemble perturbation variance at the nominal sampling time 16 September 2004 0000 UTC from the previous day NCEP ensemble forecast. The green circles represent the dropwindsonde locations.

Figure 5. Variance explained within the verification region (large red circle) for observations taken at the sampling time 16 September 2004 0000 UTC from the Ensemble Transform Kalman Filter run from the previous day NCEP ensemble forecast. The green circles represent the dropwindsonde locations.

Figure 6. Initial condition differences in the 850 - 200 hPa mean wind between the no and all dropwindsonde cases for 16 September 2004 0000 UTC. The circles represent the dropwindsonde locations.

Figure 7. 24-h forecast differences in the 850 - 200 hPa mean wind between the no and all dropwindsonde cases for 16 September 2004 0000 UTC.