Geo Time Series Modeling
So by now, you should have the IBTrACS sample data accessible on the Warp 10 Sandbox instance. Before going into the analysis, you should understand the model of the data, based on Geo Time Series, or GTS for short.
There are two GTSs per tropical cyclone:
- Wind time series, in knots, with name
gov.noaa.storm.wind
- Pressure time series, in millibars, with name
gov.noaa.storm.pres
Each tropical cyclone is defined by a code which is set in the serial
label of each GTS. So by choosing a GTS name and labels you are effectively selecting a time series of measurements for a uniquely identified tropical cyclone.
As storm names are not unique, they are stored in the name
attributes. Contrary to labels, attributes are only additional information and do not serve to identify a GTS.
Finally, for each timestamp where data is present in the IBTrACS sample dataset, the latitude, longitude and value, which is either wind speed or pressure, is stored. The data given by the NOAA is well presented as each timestamp is mostly aligned on either 00:00, 06:00, 12:00 or 18:00. This simplifies the analysis but Warp 10 can also handle irregular timestamps.
Here is an example of GTSs for a single tropical storm with serial "2009251N12336":