Click here for
catalog of sonde drops.
Click here for zipped post-processed sonde data.
1. Synoptic situation
Tropical Storm Hermine developed from an easterly wave that had been tracked across the Atlantic since 5 September. This systems was not classified as a tropical depression until 1200 UTC 17 September, in the Gulf of Mexico. Since the system was not a tropical cyclone at the nominal time of the mission, track and intensity forecasts do not officially get verified. This report is simply to assess the possible impact of the synoptic surveillance mission, and results are not included in the summary information. By 16 September, the tropical wave was starting to develop in the central Gulf of Mexico. Due to the possibility of a landfalling tropical cyclone on the coast of the Gulf of Mexico, a synoptic surveillance mission was called for nominal time 0000 UTC 17 September. At that time (Fig. 1), a monsoon-like gyre was situated over the Gulf of Mexico, with a strong subtropical ridge extending to the north and east of this gyre.
2. General Assessment of dropwindsonde impact
A. GFDL model
Figure 2 shows the GFDL forecast tracks for Hermine, and Table 1 shows the errors and impact of the synoptic surveillance mission. The results are mainly negative, with improvement only at 12 h. However, the runs with the dropwindsonde data is qualitatively better than that with none of the dropwindsonde data. The upper-tropospheric data improved the forecasts only after 48 h. The GFAL predicted landfall near Franklin, Louisiana, and the GFP3 runs predicted landfall near Avery Island, Louisiana, both substantially closer to the actual landfall near Cocodrie, LA, than the GFNO landfall forecast near High Island, TX, 77 h into the forecast.
Figure 3 shows the VICBAR forecast tracks for Hermine, and Table 2 shows the errors and impact of the synoptic surveillance mission. The dropwindsonde data improved the forecasts at all times, and substantially improved the landfall forecast. The upper-tropospheric data improved the forecasts at all forecast times. The forecast of the landfall point was considerably improved. VBNO predicted landfall near Grand Isle, Lousiana, only 68 km from the actual landfall point near Cocodrie, Louisiana, 77 h into the forecast. However, VBAL forecast landfall at Dulac, Lousiana, and VBP3 forecast landfall near Golden Meadow, Lousiana, both substantially closer to the actual landfall point than the VBNO forecast.
Figure 4 shows the GSM forecast tracks for Hermine and Table 3 shows the errors and impact of the synoptic surveillance mission. The dropwindsonde data improved the GSM forecast track through 36h. The upper-tropospheric data improved the forecast at all forecast times except 48 h. The forecast of the landfall point was degraded, though all forecasts were good. AVNN forecast the landfall near Franklin, Louisiana, whereas the AVAL forecast landfall near Abbeville, Louisiana, and AVP3 forecast landfall near Grand Chenier, Louisiana.
Table 4 shows the GFDL intensity forecast errors and impact of the synoptic surveillance mission. The dropwindsonde data had a mixed impact, improving the forecasts at 24, 72, and 84 h. The upper-tropospheric data improved the forecasts until 72 h into the forecast. 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 tropical waves in the Eastern Pacific Ocean and east of the Lesser Antilles. Other areas of large perturbation include an upper-level cold low just east of the Lesser Antilles, another located north of Puerto Rico, an extension of the subtropical ridge into the Eastern Caribbean Sea, and axes of the subtropical ridge off the southeastern United States coast. Areas of large perturbation near to Hermine include an extension of the axis of the subtropical ridge over eastern Texas, a region of confluence over western Cuba, and a complex pattern in the Gulf of Mexico associated with the large monsoon gyre and associated troughs extending into Texas, Mexico, and the Gulf of Honduras. Of these, only the perturbation in eastern Texas was well-sampled, including regular rawinsonde observations over Texas and northeastern Mexico. The lack of rawinsonde data over central Mexico, and the relatively large area north of the Yucatan peninsula without dropwindsonde data preclude considering these other targets being well-sampled.
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 differences are very small, with the largest extending from the Bay of Campeche northeastward. Because of the lack of rawinsonde data over Yucatan, much of this maximum impact is in a region in which no sounding information is available. Therefore, the spread of the impact away from the dropwindsonde locations suggests that the data assimilation has allowed the dropwindsonde data from nearby areas to influence the initial conditions there. Since part of the relatively large difference is in an area of relatively large ensemble perturbation, this aliasing may amplify in time and negatively impact the forecasts.
Figure 7 shows that, by 24 h into the forecast, the largest difference between the forecasts with and without the dropwindsonde data has amplified and moved northward in the Gulf of Mexico. The origin of this maximum is the initial maximum located between dropwindsonde locations, and must therefore be suspect. The maximum over southern Mexico originated in the Bay of Campeche, in an ensemble perturbation minimum, and has not amplified. The maximum over southern Texas is in an area of relatively large ensemble perturbation. 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 Hermine at nominal time 17 September 1998 0000 UTC has provided mainly negative results. However, the forecasts of the landfall positions have mainly been substantially improved, with the forecast degradation due to a speed error in the forecasts with the dropwindsonde data. The MRF ensemble forecasting system suggested that data throughout the Gulf of Mexico would have the greatest impact on the Hermine forecast. Only one of these regions were effectively sampled, allowing for aliasing of data into unstable areas and a degradation of the forecast.
|GFAL Error (km)|
|GFP3 Error (km)|
|GFTG Error (km)|
|12||20.||15.||( 25%)||11.||( 45%)||15.||( 25%)|
|36||151.||217.||( -44%)||201.||( -33%)||213.||(-41%)|
|48||300.||389.||( -30%)||377.||( -26%)||378.||(-26%)|
|72||499.||722.||( -12%)||915.||( -83%)||875.||(-75%)|
|84||396.||552.||( -45%)||744.||( -88%)||704.||(-78%)|
|landfall||324.||98.||( 70%)||116.||( 64%)||120.||( 63%)|
|VBAL Error (km)|
|VBP3 Error (km)|
|VBTG Error (km)|
|12||142.||142.||( 0%)||142.||( 0%)||142.||( 0%)|
|24||363.||354.||( 2%)||358.||( 1%)||363.||( 0%)|
|36||510.||502.||( 2%)||505.||( 1%)||510.||( 0%)|
|48||642.||635.||( 1%)||636.||( 1%)||640.||( 0%)|
|72||661.||627.||( 5%)||634.||( 4%)||657.||( 1%)|
|84||811.||755.||( 7%)||770.||( 5%)||804.||( 1%)|
|landfall||68.||22.||( 68%)||49.||( 28%)||68.||( 0%)|
|GSAL Error (km)|
|GSP3 Error (km)|
|GSTG Error (km)|
|12||67.||22.||( 67%)||56.||( 16%)||45.||( 33%)|
|24||110.||80.||( 27%)||90.||( 18%)||99.||( 10%)|
|36||55.||11.||( 80%)||20.||( 64%)||40.||( 27%)|
|48||46.||93.||(-102%)||81.||( -76%)||30.||( 35%)|
|72||161.||298.||( -85%)||313.||( -94%)||177.||(-10%)|
|landfall||73.||138.||( -89%)||165.||(-126%)||45.||( 38%)|
|GFAL Error (kn)|
|GFP3 Error (kn)|
|GFTG Error (kn)|
|72||22||20||( 9%)||8||( 64%)||6||( 73%)|
|84||12||10||( 17%)||2||( 83%)||4||( 67%)|
Return to 1998 assessment page
Return to HRD home page