Click here for N42RF
catalog of sonde drops.
Click here for N42RF zipped post-processed sonde data.
Click here for N49RF catalog of sonde drops.
Click here for N49RF zipped post-processed sonde data.
1. Synoptic situation
Hurricane Georges developed from an easterly wave that moved off the coast of Africa 13 September, and developed into a tropical depression two days later. The system moved westward, strengthening rapidly into a hurricane by 17 September. Due to the potential threat to Puerto Rico and the Virgin Islands, a synoptic surveillance mission was tasked for nominal time 19 September 1998 0000 UTC, with a follow-on mission the next day. This first mission was called quickly due to the rapid movement of Georges, and occurred when the NOAA aircraft was due to ferry to the islands for missions the following day. Because of this, only a shortened two-plane mission was performed. At the nominal time, Georges was moving slowly west-northwestward about 1200 km east of Barbados, to the south a strong subtropical ridge (Fig. 1). A strong cold low was located in the Caribbean Sea to the south of Hispaniola. The cyclonic circulation in the Gulf of Mexico corresponded to Tropical Storm Hermine about to make landfall on the Central Gulf Coast.
2. General Assessment of dropwindsonde impact
A. GFDL model
Figure 2 shows the GFDL forecast tracks for Hurricane Georges, and Table 1 shows the errors and impact of the synoptic surveillance mission. A. GFDL model
The results are mainly positive, with large improvements at all times except 120 h. The upper-tropospheric data improved the forecasts at all forecast times through 36 h.
Georges made landfall five times during the forecast. The first was 52.5 h into the forecast at Mill Reef, Antigua. The GFAL forecast landfall near Plymouth, Montserrant, and the GFP3 forecast landfall near Anse-Bertrand, Grande-Terre, Martinique, both substantial improvements over the GFNO landfall forecast near St.-Francois, Grand-Terre, Martinique.
None of the three runs predicted the second landfall near Basseterre, St. Kitts, 56 h into the forecast, although all three forecasts were very good. The GFAL and GFP3 were qualitatively better than the GFNO forecast at the landfall time.
The GFNO forecast remained south of Puerto Rico, though Georges made landfall near Humacao 70 h into the forecast. Both the GFAL and GFP3 forecast landfall near Cabo Rojo, and both these forecasts were qualitatively better than the GFNO at the landfall time.
Georges made landfall on the eastern tip of the Dominican Republic 84.5 h into the forecast. All three versions of GFDL forecast landfall exceptionally close to the landfall point, although the dropwindsondes degraded the landfall forecasts.
The final landfall of Georges during the forecast period was 117.5 h into the forecast near Imias, Guantanamo, Cuba. None of the forecasts made landfall on Cuba. The GFNO forecast landfall near Matthew Town, Great Inagua Island, Bahamas, and the GFAL and GFP3 forecast landfall slightly further away over Little Inagua Island, Bahamas.
Figure 3 shows the VICBAR forecast tracks for Hurricane Georges, and Table 2 shows the errors and impact of the synoptic surveillance mission. The dropwindsonde data improved the forecasts at all times, and the upper-tropospheric data provided a better forecast than the run with all these data through 48 h.
Georges made landfall five times during the forecast. The first two were 52.5 h into the forecast at Mill Reef, Antigua, and 56 h into the forecast near Basseterre, St. Kitts. All three VBAR forecasts slipped between the islands and did not actually forecast landfall in the Lesser Antilles. However, all three forecasts were excellent, just missing both landfalls by just a few km. Further, the VBAR forecasts passed just south of Puerto Rico, missing the third landfall near Humacao, Puerto Rico, 70 h into the forecast, though again the forecasts were very good.
All three VBAR runs made landfall near Oviedo, Dominican Republic, far to the west of the actual landfall near La Romana. However, the forecast tracks were parallel to the coast, and all three forecasts came very close to the landfall point. And finally, all three forecasts made landfall near Portland Bight, Jamaica, to the southwest of the actual landfall point near Imias, Guantanamo, Cuba.
Figure 4 shows the GSM forecast tracks for Hurricane Georges, and Table 3 shows the errors and impact of the synoptic surveillance mission. The dropwindsonde data improved the GSM forecast track at all times except 120 h. The upper-tropospheric data improved the forecast at all forecast times except 36 h and after 96 h.
Georges made landfall five times during the forecast. The first was 52.5 h into the forecast at Mill Reef, Antigua. All GSM runs forecast landfall in Antigua, with the AVP3 forecast being slightly degraded compared to the other two. All three forecasts were exceptional.
Only AVAL predicted the second landfall near Basseterre, St. Kitts, 56 h into the forecast, although all three forecasts were exceptional.
Georges made landfall near Humacao 70 h into the forecast, and all three versions of the GSM made landfall in Eastern Puerto Rico, within 50 km of the actual landfall point. The AVAL was the best version, though all three versions provided exceptional forecasts of this landfall.
Georges made landfall on the eastern tip of the Dominican Republic 84.5 h into the forecast. The AVNN made landfall near Cabrera, and the AVAL forecast landfall near Cabo Cabron, in the Dominican Republic. The AVAL forecast was substantially better than the AVNN forecast. The AVP3 passed to the north of Hispaniola.
The final landfall of Georges during the forecast period was 117.5 h into the forecast near Imias, Guantanamo, Cuba. The AVNN forecast landfall near Punta Maisi, Guantanamo, Cuba, an exceptional forecast. The AVAL made a good forecast of landfall near Banes, Holguin, Cuba, though this forecast was substantially degraded compared to the AVNN forecast. The AVP3 did not forecast landfall in Cuba.
Table 4 shows the GFDL intensity forecast errors and impact of the synoptic surveillance mission. The dropwindsonde data had a positive impact through 36 h and at 84 and 96 h, and the upper-tropospheric data improved the forecast at all times except 48 h. Tuleya and Lord (1997) also showed modest improvements to GFDL model intensity forecasts in the HRD synoptic flow cases.
The atmosphere has long been recognized as a chaotic system (Lorenz 1963), e.g. very small perturbations to initial conditions result in increasingly large differences in the evolution of the atmosphere with time. Since the exact state of the atmosphere can never be measured, all analyses contain errors whose magnitudes can only be estimated. An indeterminate number of initial conditions consistent with the observational data can therefore be used in numerical weather prediction, and single model runs at any synoptic time only give one possible solution to the evolution of the atmosphere. Many operational forecast centers around the world, therefore, now employ ensemble forecasting as a means of quantifying the uncertainty in the evolution of the atmospheric system. Small perturbations from a best "control" state are calculated and added to and subtracted from this control to allow for different integrations starting from theoretically equally likely initial states. These perturbations are designed so as to mimic the fastest growing modes in the model and to create the largest envelope of possibilities in the forecast. Therefore, they generally correspond to locations where large analysis errors will most impact the forecast. Those features corresponding to perturbations which most impact the tracks of tropical cyclones must be found, properly sampled, and thoroughly tested, to prove the efficacy of targeting techniques.
Figure 5 shows the variance of the size of the perturbations in the National Centers for Environmental Prediction (NCEP) global model ensemble forecasting system (Toth and Kalnay 1993). The largest perturbations are associated with Hurricane Georges and Tropical Storm Hermine. The area of relatively large perturbation in the eastern Pacific Ocean is associated with a tropical wave. Only relatively small perturbations exist near Georges, with two maxima along 55W associated with areas of diffluent flow. Maxima off the U. S. East Coast along 32.5 N are associated with weak vorticity maxima in the mid-latitude westerly flow. A very weak maximum is associated with the upper-level low south of Hispaniola. None of these features were sampled during the synoptic surveillance mission.
Figure 6 shows the difference in the 850 - 200 hPa averaged winds between the runs in which all the dropwindsonde data and none of the dropwindsonde data are included. The largest differences is east of Barbados in the area of confluence west of Georges. This area extends more than 750 km away from the location of dropwindsonde observations, suggesting the the data assimilation spread the data from these locations into the data-void regions to the east. Other small maxima were in the axis of the subtropical ridge to the northeast of Puerto Rico, and in the axis of the trough north and south of Hispaniola.
Figure 7 shows that, by 24 h into the forecast, the differences between the forecasts with and without the dropwindsonde data have amplified in the area of diffluent flow to the west of Georges. Further, the area originally to the north of Hispaniola has moved northward into the midlatitude westerly flow and amplified rapidly as it converged with one of the abovementioned vorticity maxima. Both of these features were originally in areas of large ensemble spread, and the impacts are thus expected to amplify. The remaining to original impact maxima have decayed or stayed approximately the same amplitude. These results generally confirm that the largest (smallest) ensemble perturbations correspond to amplifying (decaying) modes in the model.
The dropwindsonde data obtained during the synoptic surveillance mission for Hurricane Georges at nominal time 19 September 1998 0000 UTC has provided mainly positive results throughout the five day forecast of Hurricane Georges in three models tested. All forecasts were very good, and provided excellent forecasts of the five landfalls during the forecast period. Even though the forecasts without the dropwindsonde data were very good, the data was able to improve these forecasts.
|GFAL Error (km)|
|GFP3 Error (km)|
|12||131.||118.||( 10%)||129.||( 2%)|
|24||192.||171.||( 11%)||172.||( 10%)|
|36||224.||160.||( 29%)||163.||( 27%)|
|48||173.||118.||( 32%)||101.||( 58%)|
|72||110.||61.||( 45%)||59.||( 46%)|
|84||114.||64.||( 44%)||35.||( 69%)|
|96||137.||100.||( 27%)||78.||( 43%)|
|120||246||258.||( -5%)||241.||( 2%)|
|Landfall#1||94.||54.||( 43%)||59.||( 37%)|
|Landfall#5||177.||220.||( -24%)||213.||( -20%)|
|VBAL Error (km)|
|VBP3 Error (km)|
|12||76.||65.||( 14%)||65.||( 14%)|
|24||128.||118.||( 8%)||118.||( 8%)|
|36||204.||182.||( 11%)||184.||( 10%)|
|48||259.||236.||( 9%)||238.||( 8%)|
|72||341.||320.||( 6%)||318.||( 7%)|
|84||456.||436.||( 4%)||435.||( 5%)|
|96||580.||562.||( 3%)||559.||( 4%)|
|108||681.||676.||( 1%)||673.||( 1%)|
|120||883.||879.||( 0%)||876.||( 1%)|
|Landfall#4||278.||291.||( -5%)||278.||( 0%)|
|Landfall#5||374.||382.||( -2%)||382.||( -2%)|
|GSAL Error (km)|
|GSP3 Error (km)|
|12||108.||86.||( 20%)||86.||( 20%)|
|24||141.||132.||( 6%)||132.||( 6%)|
|36||245.||162.||( 34%)||153.||( 38%)|
|48||292.||199.||( 32%)||211.||( 28%)|
|72||370.||299.||( 19%)||320.||( 14%)|
|84||503.||427.||( 15%)||438.||( 13%)|
|96||474.||459.||( 3%)||463.||( 2%)|
|108||515.||492.||( 4%)||487.||( 5%)|
|120||543.||566.||( -4%)||543.||( 0%)|
|Landfall#3||39.||0.||( 100%)||31.||( 21%)|
|GFAL Error (kn)|
|GFP3 Error (kn)|
|12||32||28||( 13%)||31||( 3%)|
|24||49||48||( 2%)||49||( 0%)|
|36||48||41||( 15%)||42||( 13%)|
|84||27||22||( 19%)||23||( 15%)|
Return to 1998 assessment page
Return to HRD home page