FIFTH INTERNATIONAL WORKSHOP ON TROPICAL CYCLONES


Topic 0.1e: Satellite Applications at the Joint Typhoon Warning Center

Presenter: G. T. Engel
Joint Typhoon Warning Center
425 Luapele Road
Pearl Harbor, HI 96860-3103

E-mail: engelg@npmoc.navy.mil
Fax: 808-471-4581

Co-Chairs: Chris Velden, John LeMarshall

0.1e.1 Introduction

The Joint Typhoon Warning Center (JTWC), Pearl Harbor Hawaii, is responsible for providing tropical cyclone forecasts to the United States Department of Defense, and Department of State for the Pacific and Indian Oceans. Prior to 1987, JTWC used reconnaissance aircraft and meteorological satellite data to fulfill this mission. After 1 October 1987, due to budgetary considerations, aircraft reconnaissance flights in the western North Pacific ceased and the JTWC has since relied primarily on satellites for tropical cyclone reconnaissance.

Data from geostationary satellites owned by the United States and other countries, as well as data from United States polar and equatorial orbiting satellites are used to assess synoptic features and to ascertain cyclone position, intensity, and size. This past year has seen an increased use of microwave imagery by JTWC in an effort to improve the temporal and spatial continuity of the analysis of the tropical cyclone prior to issuing a warning on the storm and during the warning process. The increased emphasis on the use of microwave imagery and scatterometer data combined with the time limitations of a tropical cyclone warning process, and difficulty in interpreting the data has periodically contributed to information overload for the satellite analyst. A systematic approach needs to be developed that allows the satellite analyst to apply a Dvorak style technique to microwave imagery and scatterometer data that will enhance initial analysis of storm structure, intensity and position given the time constraints of the forecast process.

0.1e.2. Tropical Cyclone Analysis Process Overview


a) Satellite Display and Analysis Capabilities

JTWC has four forecast teams each of which has a dedicated satellite analyst who works closely with a typhoon forecaster. The satellite analyst maintains a continuous watch of the tropical Pacific and Indian Oceans using two key pieces of equipment.

The Defense Meteorological Satellite Program’s (DMSP) Mark IV-B is an Air Force satellite data acquisition and display system built by Lockheed Martin. This equipment provides the satellite analysts timely access to imagery via direct-feed from dedicated ground stations on Guam, Kadena, Japan, Hickam, Hawaii, Elmendorf, Alaska, and Lajes, Portugal. The Mark IV-B provides JTWC the capability to view geostationary and polar orbiting imagery and is the only system JTWC has that is capable of displaying Fen Yung 2 (FY-2) and moon light visible imagery.

The second system is the Navy’s FMQ-17 or Naval Satellite Display System- Enhanced (NSDS-E), built by the Sea Space Corporation. The NSDS-E provides access to Tropical Rainfall Measuring Mission (TRMM) Microwave Imager and DMSP Special Sensor Microwave/Imager (SSM/I) data. Both systems also provide the JTWC satellite analyst with data from the GMS, GOES, MET-5/7, NOAA, and DMSP satellites.

The Internet is also used to receive microwave imagery and products (scatterometer and AMSU microwave data). Near-real-time access to imagery through a web site developed by the Naval Research Laboratory has allowed the satellite analyst to better exploit microwave imagery in the detection and analysis of tropical cyclones (Hawkins et al. 2000).

b) Synoptic or large area analysis

Geostationary satellite water vapor (WV) imagery is used in time-lapsed loops to ascertain the past and current state of the atmosphere. Water vapor imagery provides an analysis of synoptic features such as long-wave patterns, tropical upper-tropospheric troughs, and regions of divergence/convergence that can both help or hinder tropical cyclone development.

The satellite analyst uses this and other satellite data along with surface observations, upper air observations, satellite derived winds (scatterometer and upper air), radar observations, aircraft observations and model forecast output from DoD, NWS and foreign meteorological agencies to maintain a continuous meteorological watch over the Pacific and Indian Oceans. Imagery from the Cooperative Institute for Meteorological Satellite Studies (CIMSS) at the University of Wisconsin overlays winds onto imagery enhancing our ability to analyze the atmosphere around a storm. Subsequently, the satellite analyst and typhoon forecaster increase scrutiny on the area(s) where tropical cyclone development is most likely and monitors the satellite data for the development of a circulation and persistent convection.

Once an area is identified as a tropical cyclone development region (suspect area), the satellite analyst contacts the Fleet Numerical Meteorology and Oceanography Center, and the Navy Research Laboratory, Monterey and establishes invest areas or satellite windows over the area of interest.

c) Tropical Cyclone Location and Intensity Estimation

After a suspect area has been identified and developed enough to determine position and intensity, the satellite analyst is required to provide location and intensity estimates or “a fix” for the cyclone. The analyst provides a fix of position at least every 3 hours and intensity at least every 6 hours. An intensity fix is primarily produced through the application of the Dvorak Technique, which utilizes visible and infrared (IR) imagery. Fixes are also made using scatterometer, TRMM, multispectral, and Special Sensor Microwave Imager (SSM/I) data.

Visual (VIS) and IR imagery are the basis for nearly 75% of the position and intensity fixes produced by JTWC. IR imagery is the data most used for determining tropical cyclone location and intensity because of its 24-hour a day, seven days a week availability. IR imagery can also be enhanced to identify features in the various atmospheric levels. VIS imagery provides the highest resolution and is the best satellite data available for detection of surface features that may not be seen in the IR or WV imagery. Multi-spectral imagery which highlights both upper- and low-level features is also used to determine tropical cyclone position and intensity.

Fixes produced by the JTWC satellite analyst using VIS, IR, microwave, and scatterometer data are added to the fix database along with fixes from other meteorological agencies. This collection of fixes along with observational data is used by the typhoon forecaster to evaluate trends in the tropical cyclone track and intensity to develop a “working best track” and for input of a tropical cyclone bogus into numerical models.

Over the past year, the fusing of all available imagery with observational and model data has allowed the satellite analyst to provide the typhoon forecaster with greater storm continuity prior to issuing a warning on the storm. The time between when a cyclone or developing cyclone is first detected by the JTWC satellite analyst to the first tropical cyclone warning has increased from 21.7 hours in 2001 to 32.1 hours in 2002.

0.1e.3 Satellite tropical cyclone analysis using polar orbiting platforms

VIS and IR data are the primary tools for determining tropical cyclone position and intensity using the Dvorak Technique (Dvorak, 1975) because of the availability of the imagery. This past year has seen an increased emphasis on the use of microwave imagery and scatterometer data (Table 1.5.1) as we strive to obtain additional accuracy and detail that is not available using the traditional Dvorak Technique.



Geostationary Imagery

Microwave Imagery

Scatterometer Data

1999 (71)

2972

373

11

2000 (65)

3219

391

186

2001 (59)

2666

372

130

2002 (56)

2696

729

116


Table 1.5.1. Number of fixes by imagery/data type. Number of fixes for the years 2002 are through October. Numbers in parenthesis are number of storms.


a) Scatterometer

Scatterometer data provides the satellite analysts with a view of the low-level circulation and seems to be a very good tool for determining outer wind structure during the weak tropical cyclone stage. Studies have shown that scatterometer data is unreliable above 50 kts however, the data is also used for stronger tropical cyclones to determine the radius of 35 kt winds (Edson et al. 2002). The relatively broad swath width and orbital characteristics provides the JTWC with at least one pass over any given tropical cyclone during a 24-hour period. Rain flagged winds appear to provide some value when they are consistent and cover a quadrant of the storm. Experience has shown that when this happens, the rain-flagged winds represent the low end of wind speeds in the area.

JTWC retrieves scatterometer data from three different websites (NOAA, FNMOC, and NRL) which provide different coverage, resolution, and in some instances, different (apparent) wind speed / center positions for the same tropical cyclone. Furthermore, none of the sites are able to deliver the products in a timely manner (less than 2-3 hours after the pass). This lack of timeliness and disparity in data values has on occasion prevented the analyst from using scatterometer data for assessing intensity or storm structure and in making fixes that impact the next tropical cyclone forecast.

Once the scatterometer data is available, a fix is provided and incorporated into the tropical cyclone analysis and forecast process. While scatterometer data has not been used as frequently as in previous years (Table 1.5.1) to provide a position fix during 2002, the typhoon duty officers have used the data to assess storm structure.

Use of the scatterometer data over the past three years has also shown problems with depicting false centers in the scatterometer wind field. At times, these differences have been up to 120 nm from centers depicted in the TRMM, SSM/I or VIS data. False center events have even been noted in tropical cyclones of typhoon intensity with an eye present in VIS. Using the scatterometer ambiguity product (raw data) from the NOAA site to resolve the false center problem has shown some promise however, the technique of using the ambiguity data is difficult to train upon and there appears to be a large degree of subjectivity present in the process.

b) SSM/I - TRMM

85 GHz and 37 GHz DMSP SSM/I and NOAA TRMM data are used in conjunction with the VIS and IR satellite data. The near equatorial orbit of these satellites allows for up to three passes per day over a given tropical cyclone. When high-level cloud cover is present, the data has been shown to be useful in detecting a low-level cyclonic circulation when none were evident in the VIS or IR data. SSM/I and TRMM data has been especially useful at the Tropical Storm stage (35kt-64kt), when the low level circulation center is well defined, but an eye has not yet become apparent in enhanced IR or VIS imagery. While no rigorous validation has been done, it appears that an eye becomes evident in the microwave data when a Dvorak T number of 3.5 was assigned to the cyclone based upon VIS or IR data. Microwave data is also used to determine the tropical cyclone 35kt wind radius however, scatterometer data has become the primary source of data for this application. JTWC has used techniques developed by Cox et al. (1999) to categorize microwave images into intensity bins based on the degree of wrap by a developing eye and correlate that to an intensity estimate (Figure 1.5.1). Comparative analysis of this technique with the traditional Dvorak technique has shown good agreement, however there have been times when a fix derived using good quality VIS imagery was not validated by the microwave imagery. JTWC is working with the Navy Research Lab and Roger Edson to develop better techniques to interpret microwave imagery and scatterometer data within the operational tropical cyclone analysis process.




Fig 1.5.1 Example showing 85 GHz SSM/I imagery depicting a broken ring pattern correlating to 50 to 85 kt maximum wind.


c) Advanced Microwave Sounding Unit (AMSU)

Two new microwave-sounding units on NOAA-15 and NOAA-16 polar orbiting satellites have the ability to see through the upper-level cirrus and provide insight into the environment around and within the tropical cyclone. Techniques are being developed by CIMSS (Velden et al. 1998) and the Cooperative Institute for Research in the Atmosphere (CIRA) at Colorado State University (Demuth et al. 2000) to assess wind fields and improve numerical specification of a tropical cyclone vortex and intensity. This capability has helped the JTWC to identify eyes within tropical cyclones that would not otherwise be detected in VIS and IR imagery and to evaluate TC intensity. Despite some signs showing the utility of AMSU data, it has been used sparingly at JTWC.





d) Objective Dvorak Technique (ODT):

The ODT is being evaluated at the JTWC for intense (> 64 kts) tropical cyclones. The technique developed by the University of Wisconsin (Velden et al. 1998) in cooperation with the Navy Research Lab uses SSM/I and TRMM images to evaluate cloud patterns and temperatures to compute an ODT intensity or T#. Use of the technique at the JTWC over the past two years suggests that ODT final T# estimates were typically within 0.5 T# of JTWC Best Track estimates (Figure 1.5.2).




Fig 1.5.2: Comparison of ODT verses Best Track for TY 25W.


Very cold cloud temperatures led to high bias in Final T# relative to JTWC Best Track estimates (cloud temperatures < upper -70°C). The Rapid intensification flag correctly identified events, but led to a large Final T# high bias when coupled with very cold cloud top temperatures.


Given the increased use of microwave imagery, scatterometer data, and the incorporation of model data and observational data in the metwatch process, the satellite analysts have been dealing with the problem of information overload. This is compounded by the limited time schedule associated with the tropical cyclone forecast process, the subjective nature of the tropical cyclone analysis process, and delays in data receipt. This points towards the need for a system that fuses additional imagery and data to exploit automation, and provide a more systematic approach to conduct tropical cyclone reconnaissance.

0.1e.4 Cooperative Efforts

Many organizations have worked closely with the JTWC in recent years to aid in the satellite data application development effort. All of these entities have provided excellent assistance, guidance and counsel to the JTWC Satellite Operations. The National Environmental Satellite, Data, and Information service (NESDIS) has been instrumental in the development of scatterometer wind speed and direction algorithms as well as display techniques. The University of Wisconsin CIMSS has had a long professional association with the JTWC and continues to be one of the major developers of satellite data applications for the tropical cyclone mission. The Naval Research Laboratory, Monterey, CA, has been in the forefront of satellite data display techniques especially with the use of multiple sensors or data sets centered over a tropical cyclone and the effort to develop tropical cyclone analysis and forecast techniques using scatterometer data. The Fleet Numerical Meteorology and Oceanography Center has been a mainstay data provider for over 20 years. JTWC is engaged in ongoing collaborative efforts with NASA National Space Science and Technology Center and the Colorado State University CIRA on use of AMSU data for tropical cyclone analysis and forecasting.

0.1e.5 Future Capabilities

JTWC is currently working to acquire the capability to ingest MODIS data. MODIS's 2,330km swath width and resolution of between .25km and 1km will provide exceptional tropical cyclone analysis capabilities however, the data is encoded in NASA Hierarchical Data File (HDF) format, which we cannot currently utilize. A total of four MODIS instruments are planned for launch prior to 2006. Further coordination with NASA, AFWA, FNMOC, and NOAA is required to gain access to near real-time MODIS data. The Mark IVB must be modified to interpret/display HDF file format to exploit MODIS capabilities.

0.1e.6 Conclusions

Microwave imagery has helped to improve our ability to fix on tropical cyclones earlier and obtain a better understanding of the position and structure. The imagery has contributed to greater lead times from first analysis of position and intensity to warning. Microwave and scatterometer data has at times allowed the satellite analyst to adjust position and intensity estimates made using VIS and IR imagery, and have allowed the typhoon forecaster to adjust their tropical cyclone analysis and forecast process. Much work still needs to be done to fully exploit the operational benefits of this data.

The most significant issue with utilizing TRMM, SSMI, AMSU, and scatterometer data to analyze tropical cyclones for intensity and position is the lack of a unified and operationally feasible technique that has been scientifically validated against ground truth. JTWC also must overcome the challenges of personnel turnover associated with military duty that increase the training requirement for a new satellite analyst. A Dvorak style technique that utilizes microwave imagery and scatterometer data in a systematic approach which results in an estimate of tropical cyclone position, intensity and structure is a critical need. This process would aid in improved analysis and forecasts of tropical cyclones, help alleviate some of the information overload, enable the analysis of a tropical cyclone within the time constraints of the forecast process, and provide a structured program for training new satellite analysts.




Acknowledgments

Mr Ed Fukada, Capt Rob Mazany, Capt Cheryl Kendall, MSgt Jim Herron, and SSgt Erik Waugaman for taking part in the collaborative effort in the writing of this paper. LCDR Rich Jeffries and Capt Steve Vilpors for reviewing this paper and the helpful suggestions they provided.


Bibliography
Cocks, S.B., I. Johnson, R. Edson, M.A. Lander, and C.P. Guard, 1999. Techniques for incorporating SSM/I imagery into Dvorak tropical cyclone intensity estimates. Proceedings of the 23rd Conference on Hurricanes and tropical Meteorology, Dallas, TX.

Demuth, J. L., M. DeMaria, J. A. Knaff, and T. H. Vonder Haar, 2000. An objective method for estimating tropical cyclone intensity and structure from NOAA-15 Advanced Microwave Sounding Unit (AMSU) data. 24th Conference on Hurricanes and Tropical Meteorology, American Meteorological Society, Ft. Lauderdale, 29 May - 2 June 2000

Dvorak, Vernon F., 1975: A workbook on tropical clouds and cloud systems in satellite meteorology, Volumes I and II. Naval Oceanographic Office.

Edson, R., M. A. Lander, C. E. Cantrell, J. L. Franklin, P. S. Chang, and J. Hawkins, 2002: Operational Use of QuikScat Over Tropical cyclones, 25th Conference on Hurricanes and Tropical Meteorology.

Hawkins, D.J., Thomas F. Lee, Joseph Turk, Charles Sampson, John Kent, and Kim Richardson 2000: Real-Time Internet Distribution of Satellite Products for Tropical cyclone Reconnaissance, Bull. Amer.Meteor. Soc.

Joint Typhoon Warning Center, 2001: Annual Tropical cyclone Report, Naval Pacific Meteorology and Oceanography Center / Joint Typhoon Warning Center, Pearl Harbor, Hawaii.

National Oceanic & Atmospheric Administration, 2000: KLM User’s Guide, September 2000 revision.

Velden, C.S., T.L.Olander, and R.M. Zehr, 1998: Development of an Objective Scheme to Estimate Tropical cyclone Intensity from Digital Geostationary Satellite Infrared Imagery. Wea. Forecasting, March 1998.