Click here for N42RF
HSA sonde data.
Click here for N42RF transmitted sonde data.
Click here for N43RF HSA sonde data.
Click here for N43RF transmitted sonde data.
Click here for N49RF HSA sonde data.
Click here for N49RF transmitted 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 Florida, a synoptic surveillance mission was tasked for nominal time 22 September 1998 0000 UTC, with a follow-on mission the next day. At the first nominal time, Georges was moving west-northwestward and was located over Puerto Rico, to the south a strong subtropical ridge (Fig. 1). A strong cold low was located in the Caribbean Sea to the south of Central Cuba. The cyclonic circulation in the Gulf of Mexico corresponded to a weakening monsoon gyre that had recently spawned Tropical Storm Hermine. The circulation in the central Tropical Atlantic was Tropical Storm Ivan, and the circulation to the northeast of Bermuda was a subtropical system which eventually developed into Tropical Storm Karl.
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. The results are positive, with improvements at all times. The upper-tropospheric data improved the forecasts at 12, 24, 96 and 120 h.
Georges made landfall three times during the forecast. The first was 12.5 h into the forecast at La Romana, Dominican Republic. The GFAL and GFP3 forecast landfall near San Rafael del Yuma, Dominican Republic, both substantially better than the run without the dropwindsondes, which predicted landfall near Cabo San Rafael, Dominican Republic.
The forecast without the dropwindsonde data failed to forecast the second landfall near Imias, Guantanamo, Cuba. Both the forecasts with the dropwindsonde data forecast landfall near Punta Maisi, Guantanamo, Cuba, only about 35 km to the east.
The forecast without the dropwindsonde data also failed to forecast the third landfall at Key West, Florida. Both forecasts with the dropwindsonde data forecast landfall near West Palm Beach, Florida, more than 300 km to the northeast.
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 also provided a better forecast than the run with all these data at all times.
Georges made landfall three times during the forecast. The first, 12.5 h into the forecast near La Romana, Dominican Republic, was well-forecast by all three VBAR runs. The run without the dropwindsonde data predicted landfall at El Macao, Dominican Republic, and the two runs with the dropwindsonde data predicted the landfall to be closer to the actual point at Cabo Engano, Dominican Republic. All three forecasts were very good.
The forecasts with the dropwindsonde data slipped between the Bahamas and Cuba, and so did not predict the second landfall near Imias, Guantanamo, Cuba, 45.5 h into the forecast. The forecast with no dropwindsonde data, while being qualitatively worse, did predict a landfall at Matthew Town, Great Inagua Island, Bahamas.
The runs with the dropwindsonde data performed substantially better than the run without the data at the third landfall point at Key West, Florida, 87.5 h into the forecast. The run without the dropwindsonde data forecast landfall at Lucaya, Grand Bahama Island, Bahamas. The run with all the dropwindsonde data predicted landfall at Hobe Sound, FL, and the run without the upper-tropospheric data predicted landfall slightly to the south at Jupiter, FL. Though the forecast without the dropwindsonde data had a slightly better landfall forecast than the run with all the data, the former was faster than the latter, so that the overall track forecast was degraded.
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 degraded the GSM forecast track at all forecast times. The upper-tropospheric data improved the forecast at all forecast times.
Georges made landfall three times during the forecast. The first was 12.5 h into the forecast at near La Romana, Dominican Republic. The forecast without the dropwindsondes made landfall near Cabo San Rafael, Dominican Republic, whereas the runs with the dropwindsonde data made landfall slightly further away near Cabo Cabron, Dominican Republic.
None of the runs predicted the second landfall to be near Imias, Guantanamo, Cuba, but instead forecast landfall in the Bahamas. The run without the dropwindsonde data forecast landfall on Little Inagua Island, Bahamas, whereas the runs with the dropwindsonde data predicted landfall much further away, at Snug Corner, Acklins Island, Bahamas.
Georges made its final landfall of the forecast period at Key West, Florida, 87.5 h into the forecast. Only the run without the dropwindsonde data forecast a landfall that late, at High Rock, Grand Bahama Island, Bahamas, more than 400 km from the actual landfall point.
Table 4 shows the GFDL intensity forecast errors and impact of the synoptic surveillance mission. The dropwindsonde data had a positive impact at 12 h and between 84 and 96 h, and the upper-tropospheric data improved the forecasts at all times except 48 through 72 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. Another large perturbation is centered over Wisconsin, and is associated with a vorticity maximum in the midlatitude westerly flow. Another maximum is associated with the cold low over central Cuba, and another with the ridging between Georges and this cold low. A set of three maxima surround the cyclonic vortex that eventually developed into Tropical Storm Ivan. None of these features is considered to be well-sampled, since none have regular coverage without large spacing surrounding the perturbation maxima.
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 form a nearly complete ring around Hurricane Georges, of radius 400 km. The difference maximum is located to the east, in a region where no dropwindsonde data were obtained. The second largest difference is located to the north of Georges in the axis of the subtropical ridge. Two other lesser maxima are noted further north in the westerly flow, and another is located in the western Caribbean near and to the south of the cold low. Since some of the impact extends up to 500 km away from the location of dropwindsonde observations, and since the largest difference is in an area where no dropwindsonde observations were made, the data assimilation may have spread the data from the observation locations into surrounding data-void regions.
Figure 7 shows that, by 24 h into the forecast, the differences between the forecasts with and without the dropwindsonde data have started amplifying after an initial decay in the area around Georges. The second large maximum, initially near an area of relatively large perturbation around Georges, has also started amplifying. The weak difference maximum near the cold low, another spread maximum, has moved southwestward to Costa Rica and amplified rapidly. The remaining two initial maxima in the westerly flow, initially in areas of small ensemble spread, have largely decayed. 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 22 September 1998 0000 UTC has provided mainly positive results throughout the five day forecast of Hurricane Georges in the VBAR and GFDL models. Even though the forecasts without the dropwindsonde data were good, the data was able to improve these forecasts.
|GFAL Error (km)|
|GFP3 Error (km)|
|12||91.||39.||( 57%)||46.||( 49%)|
|24||70.||57.||( 19%)||67.||( 4%)|
|36||84.||56.||( 33%)||33.||( 61%)|
|48||99.||47.||( 53%)||47.||( 53%)|
|72||150.||133.||( 11%)||125.||( 17%)|
|84||262.||230.||( 12%)||226.||( 14%)|
|96||426.||374.||( 12%)||380.||( 11%)|
|120||833||665.||( 20%)||761.||( 9%)|
|Landfall#1||91.||24.||( 74%)||31.||( 66%)|
|VBAL Error (km)|
|VBP3 Error (km)|
|12||70.||56.||( 20%)||56.||( 20%)|
|24||78.||70.||( 10%)||70.||( 10%)|
|36||146.||100.||( 32%)||100.||( 32%)|
|48||180.||142.||( 21%)||148.||( 18%)|
|72||349.||280.||( 20%)||295.||( 15%)|
|84||519.||375.||( 28%)||383.||( 26%)|
|96||797.||603.||( 24%)||620.||( 22%)|
|108||1151.||914.||( 21%)||942.||( 18%)|
|120||1517.||1242.||( 18%)||1285.||( 15%)|
|Landfall#1||64.||61.||( 5%)||61.||( 5%)|
|Landfall#3||405.||335.||( 17%)||316.||( 22%)|
|GSAL Error (km)|
|GSP3 Error (km) |
|12||98.||74.||( 24%)||84.||( 14%)|
|36||206.||217.||( -5%)||217.||( -5%)|
|48||232.||250.||( -8%)||250.||( -8%)|
|GFAL Error (kn)|
|GFP3 Error (kn)|
|12||27||22||( 19%)||25||( 7%)|
|36||3||4||( -33%)||5||( -67%)|
|48||12||15||( -25%)||12||( 0%)|
|84||6||5||( 17%)||5||( 17%)|
|96||9||8||( 11%)||14||( -56%)|
|120||16||19||( -19%)||20||( -25%)|
Return to 1998 assessment page
Return to HRD home page