an updated baseline for track forecast skill through five days for the atlantic and northeastern and...

20
An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane Research Division Miami, FL Interdepartmental Hurricane Conference Charleston, SC (And a little bit on surveillance during 2004)

Upload: garey-williams

Post on 18-Jan-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and

Northeastern and Northwestern Pacific Basins

Sim AbersonNOAA/AOML/Hurricane Research DivisionMiami, FL

Interdepartmental Hurricane ConferenceCharleston, SC

(And a little bit on surveillance during 2004)

Page 2: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

The National Hurricane Center and Joint Typhoon Warning Center now issue operational five day track and intensity forecasts for tropical cyclones within their regions of interest.

To assess the skill of these forecasts, five day climatology and persistence models (CLIPER for track and SHIFOR for intensity) are needed.

Page 3: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

At NHC the Neumann CLIPER program represents the standard for skill on which all operational verification is based from 1971 until the present. As such, it is a very important program for NHC.

The new CLIPER must be at least as skillful as the original Neumann CLIPER program for 3-days. i.e. the mean best track CLIPER errors should be comparable.

The mean best track errors between the 5-day and 3-day CLIPER programs must be well correlated.

Page 4: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

The initial goal is to be able to reproduce the original Neumann CLIPER and work from there.

The Neumann dependent dataset:

1. The 40-year period 1931-1970 (398 storms, 7433 cases).2. Only those storms with life histories of at least 5 days to ensure three 72-h forecasts and a previous 12-h position. This is to ensure a homogeneous dataset because the matrix calculations were so time-consuming (171 storms, 3156 cases).

This eliminates more than half the total number of cases and storms. Many early- and late-season storms and storms that formed in the Gulf of Mexico were eliminated. Storms eliminated include 1932 Texas (cat4), 1936 Florida (cat3), Audrey, Hattie, Hilda, Camille, Celia, and the entire 1946 and 1970 seasons.

Using the current version of HURDAT and the limits above, the sample contains 3515 data points. As a result, reproducing original CLIPER has not been accomplished.

Page 5: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Neumann used as predictors:

1. Initial longitude2. Initial latitude3. Initial zonal component of motion4. 12-h old zonal component of motion5. Initial meridional component of motion6. 12-h old meridional component of motion7. Initial intensity8. Day number...and all second- and third-order combinations of the above, for a totalof 165 predictors. Neumann acknowledges that there is no physical reason to use the third-order terms.

A step-wise regression was used, and the program was terminated when 9 predictors were chosen for each predictand.

If a predictor was chosen for one predictand, it was usedfor all predictands.

Page 6: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Proposal for new versions1. Use all available cases at each forecast time, not worrying abouthomogeneity. We now have computer power to accomplish this.

2. Limit predictors to only first- and second-order terms.

3. Use the most recent 40-year period (1964-2003).

4. Keep only those predictors which are statistically significant at the 95% or 99% level.

5. Use only those predictors that are chosen for each predictand to predict that predictand.

Page 7: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Using this technique, versions of CLIPER have been made available to NHC and JTWC, and have been undergoing testing.

Aberson, S.D., 1998: Five-day tropical cyclone track forecasts in the North Atlantic basin. Wea. Forecasting, 13, 1005-1015.

Aberson, S.D., and C.R. Sampson, 2003: On the predictability of tropical cyclone tracks in the northwest Pacific basin. Mon. Wea. Rev., 131, 1491-1497.

Page 8: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Differences between the original Neumann versions and the new versions have been seen in both basins and are being investigated.

Page 9: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Though two example show eastward bias, new CLIPER has more realistic tracks, and smaller overall biases, than old CLIPER

Page 10: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 11: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 12: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 13: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 14: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 15: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 16: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 17: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 18: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane
Page 19: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Non-SAL: 87 sondes SAL: 26 sondesNOAA G-IV GPS sondes: Hurricanes Fabian and Isabel

Page 20: An Updated Baseline for Track Forecast Skill Through Five Days for the Atlantic and Northeastern and Northwestern Pacific Basins Sim Aberson NOAA/AOML/Hurricane

Summary

Development of new versions of CLIPER model for three northern hemisphere basins is ongoing.

Consistent improvements to the GFS model are seen from surveillance missions. The impact of surveillance on GFDL and SHIPS forecasts is currently relatively small.