Issues in merging storm data
Data were obtained from numerous sources. Upon receipt of the electronic files, data were converted to common format (netCDF). Thereafter, numerous merging procedures were implemented, including an assessment of storm position, the identification of unique storms, and an accounting for a variety of storm tracks, storm splits, and storm mergers.
Combining storm position and intensity data
When more than one center provided information on a storm, the IBTrACS position, wind and pressure values result from an average of the reported values. Information is also retained in IBTrACS on the range of the values.
To average or not to average
... that is the question.
The following is a summary why data should and should not be averaged when including it in various formats.
Averaging tropical cyclone maximum sustained wind
Pros
- Provides users with a final number for comparison
- Allows for numerous output formats
- We've already done it, so this would be the status quo option.
Cons
- Averaging is against the consensus of the Best Track community.
- Providing the average may give the appearance of acceptance of the value.
- Averaging is not the answer.
- During the satellite era, it is possible to normalize back to CI
- Pre satellite era, inter agency differences might be documented.
- Wind estimates from various agencies are not necessarily independent, thus invalidating central tendency theorem.
Not averaging tropical cyclone maximum sustained wind
Pros
- Avoids performing a questionable task.
- Requires each user to fully consider the errors and uncertainties of the data sources.
- Diversify WFS/WMS by supporting (serving) each data centers data separately.
Cons
- Many users would scream
- Maps cannot be made which include shading by intensity
- Would require each user to become an expert on the use of the data
- Could not provide one source of data in the WMS
