||Application of TMI SSTs and QuikScat Winds to Tropical Cyclone Research
Both the TRMM Microwave Imager (TMI) and the QuikScat SeaWinds Scatterometer have the potential to contribute greatly to the study of tropical cyclones in all ocean regions. Here, we show a few examples of how our scientists have used either or both of these instruments in data validation or research:
TMI SSTs near Tropical Cyclones
The TRMM Microwave Imager (TMI) has produced passive microwave observations at 10.7, 19.4, 21.3, 37.0, and 85.5 GHz since December 1997. Orbiting at 35° inclination from the equator, TRMM provides sampling throughout the diurnal cycle. Remote Sensing Systems has developed a Sea Surface Temperature (SST) algorithm, relying primarily on the 10.7 GHz channel. Physical parameters in the retrieval algorithm were adjusted to minimize cross-talk errors due to wind speed, wind direction, atmospheric water vapor, and cloud water. The performance of the algorithm has been assessed by comparison with the Reynolds and AVHRR SST products.
Comparison of 3 SST fields during the 1998 and 1999 Hurricane Seasons
The following animations consist of AVHRR Pathfinder interim version 4.1 "Best" SST fields on the left side (these are 3 day averages centered on the day indicated), a similar 3-day average of TRMM SST in the center, and the Reynolds Weekly 1deg SST product on the right. AVHRR SST is the main source of data for the Reynold's product, so when the IR data show significant cloud cover (causing missing data) or biasing due to high atmospheric vapor content this will show up in the Reynolds SSTs also. Two animations show the storms for each season using NHC best-track data:
1998 Season (Quicktime Movie)
1999 Season (Quicktime Movie)
Hurricanes Bonnie and Danielle
The following case study demonstrates how use of TMI SSTs would have improved intensity estimates for Hurricane Danielle during August 1998.
The Tropical Prediction Center (TPC)/National Hurricane Center (NHC) issues forecasts and warnings about all storms in the Atlantic and Eastern Pacific Oceans every 6 hours from May 15 November 30 each year. Reynolds weekly SST analysis is the chosen SST field for consideration when estimating maximum possible intensity of the storms. TMI SSTs are more recent and have a higher resolution (25 km). Figure 1 shows the microwave SSTs, IR SSTs and Reynold's SST product each overlaid with the best-tracks of Bonnie and Danielle. Evident in the microwave plot on the left is a large region of cold-water upwelling caused by Hurricane Bonnie several days earlier. The IR SSTs are blocked by Bonnie's cloud cover and the Reynold's weekly product shows SSTs from before Bonnie's passage through this region.
On August 25, 1998 Hurricane Bonnie was just off the Carolina coast and Hurricane Danielle was gathering strength in the center of the Atlantic, heading towards Florida. The Reynold's SST map shown in the above figure was used as input to the intensity models to forecast hurricane Danielle. Bonnie caused significant upwelling, that was completely missed in the IR retrievals of SST because of the clouds surrounding the storm. The TMI SST shown above could have been incorporated into the intensity models. As soon as Danielle hit the cold wake of Bonnie her intensity decreased sharply. This interaction can be seen in the following animation:
Hurricanes Bonnie and Danielle (Quicktime Movie)
These animations of hurricane Floyd contain TRMM SSTs overplotted with NHC Best Track position and wind speed data. The SST is a 3 day average centered on the day indicated. The second animation contains Hurricane Floyd's track plotted on SST anomaly maps. The anomaly maps are created by subtracting the Reynolds SST product from the TMI SSTs.
Floyd (Quicktime Movie)
Floyd Anomaly (Quicktime Movie)
Water Vapor and SST for the 1998 Hurricane Season
The following animation demonstrates the effect of atmospheric water vapor on IR SSTs. The water vapor map on the left comes from the TMI instrument and is simultaneously retrieved with the TMI SSTs. In the center, a 3 day average map of AVHRR Pathfinder "best" SST is shown. On the right are TMI SST 3-day averaged maps. Infrared SST products are biased slightly cool when the columnar water vapor content is greater than 40 mm.
Water Vapor (Quicktime Movie) - 32,603 kb
QuikScat Winds within Tropical Cyclones
Scatterometers are used to derive the wind speed and direction of winds over the ocean surface. These microwave radars send and receive pulses, and winds are calculated from the amount of returned signal. QuikScat wind plots for worldwide tropical cyclones have been archived since 1999. Validation of QuikScat data has been performed with particular focus placed on better understanding the winds within tropical cyclones. Two hurricanes in particular were used since these storms stalled for a short time, allowing us to better collocate data from several instruments.
Hurricane Dennis raveged the Carolina coast during late August and early September 1999. This plot of Hurricane Dennis on Aug 31st 1999 consists of an SSM/I precipitation plot on the left and a QuikScat wind plot on the right. The time difference between these two observations is approximately 1.5 hours. Five NDBC buoys were located within this 10-degree square region. Buoy wind observations matching the time of the QuikScat map are listed. Four of these buoys are in low or no rain regions and agree very well with the mapped winds (within 6 knots or 3 m/s), especially buoy 41001 which may have been within Dennis' eyewall. Buoy 44014 located just offshore in the heaviest rain region has a 49 kt (25 m/s) wind, where QuikScat winds were approximately 34 m/s. In this case, we believe both the buoy and QuikScat may be in error. The buoy is likely to be in very rough seas, therefore unable to accurately measure the winds, while QuikScat may be affected by the presence of rain within the observation area.
Keith was a rapidly-intensifying tropical cyclone over the northwestern Caribbean Sea, reaching Category 4 on the Saffir-Simpson Hurricane Scale as it stalled just off the coast of Belize. It was in this stalled location that we obtained good winds from QuikScat as shown in the figure below. Overplotted on the winds are two dotted curves representing the 35kt and 50kt radii for the storm. These winds and radii are in excellent agreement with model results produced by the Hurricane Research Division of NOAA shown on the left side of the plot. QuikScat input was not used in the running of this model output.