The main goal of our project is to obtain a better understanding of the impact of tropical cyclones on the ocean and to learn how cold wake variability can be parameterized from upper ocean pre-storm conditions and tropical cyclone translation speed and strength variability. As part of this work, we developed the Tropical Cyclone Cold Wake Database which;includes a broad range of satellite and in situ data re-mapped to storm-centric grids.
The database contains tropical storms with cold wakes over the period from 1870 to 2014. The tropical storms exist in most global ocean regions. We use the IBTrACS-WMO v03r06 data set (Knapp et al., 2010) to construct the storm-centric database. The IBTrACS data set is endorsed by the World Meteorological Organization's Tropical Cyclone Programme as the official best track data.
Our Tropical Cyclone Cold Wake Database is freely available for download and can be used in further research. If used, please be sure to cite the database in your publications.
The following steps are taken to construct the Tropical Cyclone Cold Wake Database:
We use the IBTrACS (International Best Track Archive for Climate Stewardship) tropical cyclone data set to construct our database. Both the IBTrACS-WMO and the IBTrACS-all are used which contain 6-hourly storm center positions, winds, pressures, and translation velocites. Due to translational speed variability, the 6-hourly observations are not consistently spaced, so we use interpolation routines to create a smoothly varying track of latitudes, longitudes and times (red line) at a spacing of 12 km. The 12 km spatial scale was chosen to offset the need for a reasonable database size and to best represent the collocated data resolutions (1- 50 km). The IBTrACS storm track locations for storm Man-Yi are shown as blue asterisks in Figure 1. We use all IBTrACS storms from 1870 to 2014. For each interpolated track position, we construct a vector orthogonal to the storm track, shown as black lines in Figure 1. The evenly spaced track and orthogonals provide the structure of the database, as illustrated in the right panel of Figure 1.
Figure 1. Super Typhoon Man-Yi, 7-17 July 2007. The left panel shows the 6-hourly TC best track positions (blue stars), the interpolated position (red line), and the track orthogonals (black lines) drawn every 12 km. The right panel illustrates how the evenly-spaced orthogonals map onto a linear grid. The linear grid in the right panel is then populated with observations for each time relevant to that storm. Note the transformation into the storm-centered coordinates may distort the land.
This data array is then used to interpolate a large variety of SST products, satellite data, in situ measurements and other supporting data to the storm locational grid. The data are also organized temporally into a storm-centric frame of reference with t=0 indicating the day of storm passage, negative numbers representing the days before storm passage and positive numbers representing the days after storm passage. The database includes data for each orthogonal as the center of a square grid. The storm data are saved into a 3-dimensional grid (x,y,t) with the spatial dimensions (x and y) determined by the orthogonal line, and the temporal dimension (t) spanning 10 days prior to storm passage to 25 days after the storm passage as illustrated in Figure 2.
Figure 2. Illustration of data structure in storm-centered frame of reference. The storm moves up the center vertical line in each panel. The left panel illustrates how the orthogonal is used to center a grid of observations onto the storm spatial frame of reference, with stacks of data in time both before and after (right panel) the storm passage.
This storm-centric database includes over 90 individual data collections, including in situ and satellite data as well as model analyses. The varying spatial and temporal resolutions of the data are interpolated to the 12 km storm-centric grid of Figure 1. All tropical storms with more than 10 observations (~3 days) are included in our database. We exclude short-lived storms because it is too difficult to produce good quality orthogonals with limited track information.
The contents of the Tropical Cyclone Cold Wake Database are outlined in the table below. Obtaining the database contents and writing read routines for each product has involved significant time and effort. This is the greatest value of the database, that we provide a wide variety of important data for tropical cold wake study in a single format collection on a storm-centric grid. Ocean and Atmospheric data are selected based on availability and relevance to storm research. Profile data are also included in the database by using a variable called "depth-valid-at" in all files. This means ARGO profile in situ data files contain a 4th dimension: (x,y,t,d) where d is the depth of the measurement.
The track interpolation routines return a list of latitude, longitude and time for the entire storm track and the region perpendicular to the track. This data array is then used to interpolate the data products onto the storm locational grid. The database contains two temporal formats, one which is storm-centric (t=0 is equal to the storm overpassing the point for all locations) and one which is data accumulated by Julian day.
The database contains netCDF4 files with full metadata. Each file contains the data for one data type and a given julian day: "1998001S09195.bin_julian_time.nc" or a given storm "1998001S09195.bin_storm_time.nc". The file name contains the IBTrACS storm serial number followed by the storm or julian day designation. More information about the storm serial number is available from the IBTrACS web site.
The serial number contains YYYYJJJHTTNNN where
- YYYY is the corresponding year of the first recorded observation of the storm (important when storms cross calendar years - southern hemisphere storms)
- JJJ is the day of year of the first recorded observation of the storm
- H is the hemisphere of the storm: N=Northern, S=Southern
- TT is the absolute value of the rounded latitude of the first recorded observation of the storm (range 0-90, if basin=SA or SH, then TT in reality is negative)
- NNN is the rounded longitude of the first recorded observation of the storm (range 0-359)
To use the database in your research, download the full set of netCDF files. All files are stored in directories by data type. The subdirectory names are listed in far right column of the table above. One can download the entire collection or if you are studying one storm, just access the files for that storm only by using the serial number in the file name. Check the Serial Number to Storm Name document to determine which serial number applies to a storm of a given name, basin and year.
Data are available by ftp at ftp://ftp.remss.com/TC_wakes/
The database contains a collection of netCDF files complete with metadata. These self describing files are easily read using the many programming languages already available, including IDL, Matlab, and Python. For a quick look at file contents, use a tool like Panoply.
Knapp, K. R., M. C. Kruk, D. H. Levinson, H. J. Diamond, and C. J. Neumann, 2010: The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone best track data. ,Bulletin of the American Meteorolical Society, 91, 363-376.
Project Presentations and Posters
Gentemann, C. L. and J. Scott (2014), Variability in tropical cyclone induced upper ocean cooling, paper presented at Ocean Sciences Meeting, Honolulu, HI.
Scott, J. P., C. L. Gentemann, and D. K. Smith (2012), Storm-Centric View of Tropical Cyclone Cold Wakes, paper presented at Interdepartmental Hurricane Conference 2012, Charleston, SC.
Production of the Tropical Cyclone Cold Wake database has been made possible by the generous funding from the NASA Physical Oceanography Program under the direction of Eric Lindstrom. Data are available at www.remss.com/.
The distribution and web-interface for this database are supported by the NASA Earth Science MEaSUREs DISCOVER Project.