WMO/CAS/WWW

FIFTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES


Key Note Topic 0.2a: Imminent uses and data assimilation

Title: TRMM and NSCAT/QuikSCAT Applications for Tropical Cyclones Studies

Presenter: T. Nakazawa
Typhoon Research Department
Meteorological Research Institute/JMA
1-1 Nagamine, Tsukuba, Ibaraki 305-0052 JAPAN

E-mail: nakazawa@mri-jma.go.jp
Fax: +81.298.53.8735


0.2.1. Introduction

Over the western North Pacific we have not had reconnaissance flights for tropical cyclones since 1987. Thus we do not have “true” observational data and operationally the Dvorak technique (Dvorak, 1975) is applied to the GMS visible(VIS)/infrared(IR) data to estimate the intensity of tropical cyclones.
In addition, other sensors are available for tropical cyclone study, e.g., TRMM (Tropical Rainfall Measurement Mission) Precipitation Radar(PR) and microwave sensors, such as SSM/I(Special Sensor Microwave/Imager), TMI(TRMM Microwave Imager) and just recently AMSR-E(Advanced Microwave Special Radiometer) as radiometers, and NSCAT(NASA Scatterometer) and QuikSCAT as scatterometers. These sensors provide us useful information that we cannot get from VIS/IR data only (Velden, 1993).
This presentation will focus on the application of each of these other sensors for tropical cyclone studies.



0.2.2. TRMM

a) Precipitation Radar(PR)


TRMM was developed under the US-Japan collaborative project. There are five sensors on TRMM: PR, TMI, VIRS(Visible Infrared Scanner), LIS(Lightning Imaging Sensor) and CERES(Clouds and the Earth’s Radiation Energy System). The most prominent feature is that TRMM has a 14 GHz radar, PR, which measures rainfall from the space similar to a ground radar. Although the swath of the PR is only about 200 km, the PR provides the vertical profile of rain with 250 m resolution over the globe. Almost 5 years after launch TRMM is still in good condition to give us the global rainfall information. In combination with TRMM/PR data, the microwave(TMI) algorithm has been greatly improved. One of the advantages of the TMI is the wide swath (700 km), which is more than three times wider than the PR swath. Fig. 1 shows an example of a TRMM rainfall image. The VIRS image is embedded in the PR image (Fig. 1, left) and the TMI image is on the right side of the figure. We still have slight differences among the rainfall products. The TMI rainfall tends to be larger than the PR rainfall. We are seeking the source of the difference and to understand the differences. We have now the TRMM Version 5 products and will update the products to Version 6 in May 2003 with many improvements.



Fig. 1 Rainfall distribution of Hurricane Floyd(1999) from TRMM
PR with VIRS(left) and TMI(right) at 09:30 UTC in 13 Sep.1999.

An accurate rainfall estimation is the key for tropical cyclone study. As is well known, the tropical cyclone is maintained by the latent heating, which comes from the phase transition from vapor to rain. Thus rainfall amount is an index of the latent heating of tropical cyclone. We notice that the VIS/IR image is frequently different from the PR/TMI rainfall image, mainly because the VIS/IR image includes the anvil clouds with no rain underneath. In the CISK mechanism, the positive feedback between the latent heating and the large-scale circulation develops tropical cyclones.

  1. SST decrease after the passage of a slow-moving TC
The TMI is also a unique sensor onboard TRMM, because of the 10 GHz channel for rainfall and SST. The TMI can estimate the SST under the cloudy condition. This is important since the tropical cyclone intensity is influenced by the SST, especially for a slow-moving cyclone. Fig. 2 shows an example of an extensive cold wake after the passage of a tropical cyclone.



Fig. 2 Cold wake after the passage of Typhoon 9804. (Left) Path of
Typhoon. (Middle) TMI SST before the passage. (Right) TMI SST after the passage.


0.2.3 Scatterometer

  1. Validation of QuikSCAT Wind
It is difficult to find a validation wind dataset for speeds greater than 25 m/s. We chose a small island in the western Pacific, Okinotorishima (20°25’N, 136°04’E), to check the QuikSCAT high-wind data. There are two datasets: one is 30 min. averaged wind, and the second is the 4-sec averaged maximum wind during each 30 min. This means we have no completely?compatible match-up dataset. However, we think it is worthwhile to compare these data with the QuikSCAT data. The QuikSCAT data we used in this study were obtained from the Remote Sensing Systems(RSS), http://www.ssmi.com.
Fig. 3 The surface wind observation (OBS) at Okinotorishima (abscissa) vs. the difference QuikSCAT wind minus OBS. (Left) OBS: 30 min. averaged wind. (Right) OBS: 4-second averaged maximum wind in 30 min.

Compared with the standard QuikSCAT data provided by the NASA/JPL(Jet Propulsion Laboratory), the RSS wind is stronger in the higher wind regime (greater than 15 m/s).
Fig. 3 shows the scatter diagram of “collocated” QuikSCAT wind speed with the 30 min. averaged one(left) and the instantaneous one(right). The data period is from July 1999 through December 2000. In the weak wind regime the QuikSCAT tends to be stronger than the truth, especially during rain. In the strong wind regime, it tends to be weaker than the truth. As we do not have enough samples in high wind conditions, it is too early to conclude that this tendency is generally true.

  1. Maximum Wind Speed of TC from QuikSCAT and Best Track
Now we would like to compare the maximum wind near the TCs observed by QuikSCAT with the Best Track data (JMA, JTWC and NHC). The total number of TC is 75 and the sample size is 394. Fig. 4 is the temporal variation of the maximum wind speed of Hurricane Alberto(2000) (+:NHC, x:QuikSCAT).

Fig. 4 The temporal variation of the maximum wind speed of Hurricane Alberto(2000) (+:NHC, x:QuikSCAT).

c) Areas with 15 m/s and 25 m/s winds
Operationally the JMA reports the radius with wind speeds greater than 15 m/s and 25 m/s from the center. However, the QuikSCAT data show that the strong wind area is sometimes observed far away from the center position. For disaster mitigation it is really important to tell fishery ships the information about the wind distribution. Fig. 5 shows one example of the wind speed distribution for Typhoon 0014 at 21:36 UTC 12 September 2000. The surface pressure was 950 hPa and the maximum wind by the JMA was 39 m/s and 39.2 m/s by the QuikSCAT. Even though the maximum wind speeds are similar, the wind distributions are quite different.



Fig. 5 Wind speed distribution of Typhoon 0014 at 21:36 UTC 12 September 2000. Light (heavy) shading denotes the wind speeds greater than 15 (25) m/s. The inner (outer) circle is the area with greater than 25 (15) m/s, as reported by the JMA.


0.2.4 Summary

In this presentation, we show several applications for TC study by TRMM/PR and microwave data. These are


These results are encouraging. At this moment, most of the results are for analysis purposes, not for forecasting. However, we hope these data will be used for forecasting by utilization in the numerical modeling.


Bibliography

Dvorak, Vernon F., 1975: Tropical Cyclone Intensity Analysis and Forecasting from Satellite Imagery. Monthly Weather Review: Vol. 103, No. 5, 420-464.

Evans, Jenni L., 1993: Sensitivity of Tropical Cyclone Intensity to Sea Surface Temperature. Journal of Climate: Vol. 6, No. 6, 1133-1140.

Velden, Christopher S., Brian M. Goodman, Robert T. Merrill, 1991: Western North Pacific Tropical Cyclone Intensity Estimation from NOAA Polar-Orbiting Satellite Microwave Data. Monthly Weather Review: Vol. 119, No. 1, 159-168.