**7. Multi‐sensor‐based TC monitoring and tracking**

Observation from the LEO satellite PMW sensors is the best way for detection and monitoring of the global TC activities. **Figure 14** is an example of total orbits of GPM GMI observations at 89 GHz horizontal polarization on March 31, 2014. It is obvious that there are good coverages at high latitudes at daily scale, but there are large gaps in tropical and mid‐latitudes. Therefore, multi‐PMW sensors are required to have a reasonable coverage to monitor and track the global TC activities. Six PMW sensors could generally provide at least one measurement over a location every 3 hr. However, the high frequency channel of the PMW sensors onboard

**Figure 14.** Example of one day satellite orbital measurements of GPM GMI 89 GHz at horizontal polarization on March 31, 2014.

different satellites is different. For example, TMI and SSM/I are at 85 GHz while SSMIS is at 91 GHz; AMSR‐E, AMSR2, and GMI are at 89 GHz. These frequency differences could lead to TB discrepancies up to 13K, which could mislead none‐expert analysts in monitoring the TC's intensity, structure, and development. In order to have consistent TBs from different PMW sensors for improved monitoring and tracking of global TC activities, a physically based calibration scheme to calibrate TBs from 85 or 91 GHz TB into 89 GHz has been developed by utilizing outputs of the cloud‐resolving model simulations for convective cloud systems and the associated radiative transfer model simulated TBs [105]. Thus, the resultant unified TBs at 89 GHz will be consistent among all PMW sensors.

because of the abundant polar‐satellite measurements, while the errors of maximum wind speed (*V*max) and minimum center pressure (*Pc*) are significantly reduced. Although the standard deviation of track forecast is slightly large mainly due to the deteriorated Debby track

The COAMPS‐TC system developed at NRL‐MRY has been transitioned to operations for real‐ time TC forecasts for several hurricane seasons at a spatial resolution of 5 km and systematically evaluated for large samples of TC forecasts over Atlantic and West Pacific basins [104]. Results demonstrate the accurate predictions of TC track and intensity, as well as the sea surface temperature cooling response to the storm, indicating the capability of the COAMPS‐TC system to realistically capture characteristics of the ocean surface waves and their interactions with boundary layers above and below the ocean surface. There are more satellite measurements than what are actually assimilated into the models. Proper utilization of satellite data with positive impacts on forecast skills still requires more investigations and

Observation from the LEO satellite PMW sensors is the best way for detection and monitoring of the global TC activities. **Figure 14** is an example of total orbits of GPM GMI observations at 89 GHz horizontal polarization on March 31, 2014. It is obvious that there are good coverages at high latitudes at daily scale, but there are large gaps in tropical and mid‐latitudes. Therefore, multi‐PMW sensors are required to have a reasonable coverage to monitor and track the global TC activities. Six PMW sensors could generally provide at least one measurement over a location every 3 hr. However, the high frequency channel of the PMW sensors onboard

**Figure 14.** Example of one day satellite orbital measurements of GPM GMI 89 GHz at horizontal polarization on March

forecasts, the overall *V*max and *Pc* errors are obviously reduced.

156 Recent Developments in Tropical Cyclone Dynamics, Prediction, and Detection

**7. Multi‐sensor‐based TC monitoring and tracking**

validations.

31, 2014.

**Figure 15** presents TB differences at horizontal polarization for TMI 85 and SSMIS 91 GHz against AMSR‐E 89 GHz under four classified clouds: large rain, light rain, cloudy, and clear sky. The associated fitting curves are components of the unified calibration scheme. It is evident that the TB differences between TMI 85 and AMSR‐E 89 GHz‐H at heavy rain situations could be as large as 13K. The difference is decreased to 1.5K after applying this calibration scheme. By same token, the TB differences between SSMIS 91 GHz and AMSR‐E 89 GHz are also decreased from 3K to 0.5K. The impacts of this calibration scheme are significant in improving monitoring and tracking of the global TC intensity, structure, and development because of the unified TBs from different PMW sensors. **Figure 16** shows an example of comparison of the observed Hurricane Igor TBs from TMI 85 GHz‐H and AMSR‐E 89 GHz‐H before and after the calibration. Without the calibration (top left panel), the TB patterns from TMI 85 GHz‐H indicate a false TC intensification in four minutes because of the enhanced eyewall in red color from AMSR‐E 89 GHz‐H (bottom left panel). This misleading is caused by the TB differences due to their frequency shift. With the calibration (bottom right panel), the TB distribution patterns are very close to those observed by AMSR‐E 89 GHz. The top right panel is TB corrections due to the frequency shift. In addition, this unified calibration scheme has been applied to create a self‐consistent TB database for TCs observed by all PMW sensors, including a TC center fixing algorithm and high quality interpolation scheme. The new database can be utilized for climatology studies of TC structure, intensity, and life cycles [68, 69]

**Figure 15.** (a) Comparison of TB differences between the simulated TMI 85 and AMSR‐E 89 GHz H pol for Hurricane Bonnie and squall line. The black, yellow, blue, and green color points are for the classified cloud conditions of rain, light rain, non‐rain, and cloudy, respectively. The heavy dash lines are their related polynomial fitting lines. (b) Same as (a) except for SSMIS 91 and AMSR‐E 89 GHz (from Yang et al. (2014). Reproduced by permission of Remote Sensing).

**Figure 16.** Impact of the newly developed physically based calibration scheme on hurricane Igor: (upper‐left panel) original TMI 85 GHz‐H pol; (bottom‐left panel) original AMSR‐E at 89 GHz‐H pol; (upper‐right panel) TB correction distribution; and (bottom‐right panel) calibrated TMI 89 GHz‐H pol (from Yang et al. (2014). Reproduced by permis‐ sion of Remote Sensing).

The GEO IR/VIS sensors are also important in monitoring the global TC activities because the LEO PMW sensors are limited. The IR/VIS sensors can fill the gaps missed by PMW sensors. The precise TC center position is the most important index not only in monitoring and tracking of TC lifecycles, but also in improving TC forecasts. The Automated Rotational Center Hurricane Eye Retrieval (ARCHER) algorithm has been developed to automate‐objectively determine the TC center from 85–92 GHz channels of PMW imagers [106]. This algorithm has been applied at NRL‐MRY (http://www.nrlmry.navy.mil/TC.html) and Cooperative Institute for Meteorological Satellite Studies (CIMSS) at University of Wisconsin‐Madison (http:// tropic.ssec.wisc.edu ) for operational TC monitoring and tracking. Its updated version (ARCHER‐2) is now available for including LEO 37 GHz PMW imagers, GEO IR/VIS imagery, and scatterometers [107]. It also produces a quantitative expected error estimate used for evaluation of the suitability of the estimated TC centers. The multi‐satellite and multi‐sensor‐ based TC track called the ARCHER‐track is able to provide a fast access to additional TC center positions for operational forecasting processes. An example of the ARCHER‐track for Hurri‐ cane Michael (**Figure 17**) presents the TC centers during its lifecycle from different sensors compared with the best official TC track. It demonstrates the TC centers from PMW sensors have smaller errors than from IR/VIS sensors, although these TC centers from IR/VIS are still accurate. In addition, the IR/VIS‐based TC positions provide important information to fill the gaps missed by the limited PMW sensors.

**Figure 16.** Impact of the newly developed physically based calibration scheme on hurricane Igor: (upper‐left panel) original TMI 85 GHz‐H pol; (bottom‐left panel) original AMSR‐E at 89 GHz‐H pol; (upper‐right panel) TB correction distribution; and (bottom‐right panel) calibrated TMI 89 GHz‐H pol (from Yang et al. (2014). Reproduced by permis‐

158 Recent Developments in Tropical Cyclone Dynamics, Prediction, and Detection

The GEO IR/VIS sensors are also important in monitoring the global TC activities because the LEO PMW sensors are limited. The IR/VIS sensors can fill the gaps missed by PMW sensors. The precise TC center position is the most important index not only in monitoring and tracking of TC lifecycles, but also in improving TC forecasts. The Automated Rotational Center Hurricane Eye Retrieval (ARCHER) algorithm has been developed to automate‐objectively determine the TC center from 85–92 GHz channels of PMW imagers [106]. This algorithm has been applied at NRL‐MRY (http://www.nrlmry.navy.mil/TC.html) and Cooperative Institute for Meteorological Satellite Studies (CIMSS) at University of Wisconsin‐Madison (http:// tropic.ssec.wisc.edu ) for operational TC monitoring and tracking. Its updated version (ARCHER‐2) is now available for including LEO 37 GHz PMW imagers, GEO IR/VIS imagery, and scatterometers [107]. It also produces a quantitative expected error estimate used for evaluation of the suitability of the estimated TC centers. The multi‐satellite and multi‐sensor‐ based TC track called the ARCHER‐track is able to provide a fast access to additional TC center

sion of Remote Sensing).

**Figure 17.** Example of the ARCHER‐Track product for Hurricane Michael (2012). Components of the graphic are ex‐ plained in the top‐right legend. Inside the white circles, D = tropical depression, S = tropical storm, 1 = category‐1 hurri‐ cane, etc. (from Wimmers and Velden (2016). ©American Meteorological Society. Used with permission).

**Figure 18** presents another example of a multi‐satellite PMW sensor‐based TC imagery at 89 GHz‐H for a 2014 typhoon Rammasun to display its structure, intensity, development, and tracking. ARCHER is used to find the TC centers while the multi‐sensor calibration scheme is applied to generate the unified TBs at 89 GHz. In addition, the Backus‐Gilbert interpolation scheme is utilized in providing high spatial resolution PMW TB images [108]. The 4° × 4° boxes centered at the TC eye positions from all PMW sensors are extracted and overplayed into one image to exhibit a summary overview of the TC structure, intensity, and tracking during its lifecycle. It clearly shows evaluations of TC's key characteristics for purposes of a global TC monitoring and tracking. This new live TC tracking imagery will be added into the NRL‐MRY TC web page in the near future.

**Figure 18.** Example of the multi‐satellite PMW sensor‐based TC track for 2014 West Pacific Typhoon Rammasun (09W). The light white line is the JTWC near real‐time TC track.

## **8. Summary**

TC is one of the most destructive weather phenomena. It is initiated in tropical oceans and has a lifecycle mostly over water surface with unique horizontal characteristics of eyewall, spiral convective zones, and a vertical warm core. Satellite remote sensing is the only way to provide complete observation and monitoring of the global TC activities. The GEO IR/VIS is very useful in monitoring TC activities but not in providing accurate estimates of the TC center locations and intensity. The LEO PMW sensors are better suited for detecting TC genesis, development, and structures because of their ability to measure the atmospheric profiles. TC structure and intensity can be estimated from the PMW measurements.

Heavy precipitation is another important feature of TC activities. The abundance of TC rainfall is crucial to the drought‐impacted regions because even one TC precipitation process could lead to significant relief to the severe drought situation. However, the large amount of rainfall from TC activities is also one of TC's impacts for loss of human lives and property damages. The asymmetric property of TC rainfall makes it hard to predict TC rainfall distribution. Although accurate rainfall retrievals from PMV sensors and the modern TC rainfall prediction schemes have led to reasonable TC rain forecasts, a more consistent TC rainfall from various PMW sensors and the TC diurnal characteristics are required to make further advances in TC rainfall forecasts.

Satellite remote sensing is very important in improving TC forecasts with the data assimilation process. The near real‐time measurements of accurate atmospheric conditions from LEO and GEO sensors are used to improve the NWP model's initial conditions and to minimize the innovation error for better forecasts. The LEO PMW sounding sensors are especially critical in improving weather forecasts because of their ability to provide accurate atmospheric temper‐ ature and humidity profiles. The TC track forecast errors have been gradually and substantially reduced in past decades with the improved NWP models and the data assimilation schemes. Although deduction of the TC intensity forecast errors is also statistically significant, the amplitude of its improvements is much smaller than that for the TC track forecast errors. Future efforts on optimum selections of the combined satellite sensor channels which have positive impacts and better data assimilation methods are necessary in addition to improvements in the next generation NWP models and developments as well as future advanced satellite sensors onboard adequate satellites for better spatial and temporal global coverage.

Data fusion from multi‐satellite sensors is the only way to provide a global coverage of TC activities. The LEO PMW sensors have advantages in high spatial resolution for TC structures, accurate TC positions, intensity analysis, and precipitation distributions, but they lack in temporal observations because each polar‐orbital satellite could provide measurements only twice over a location per day. The LEO IR/VIS sensors have advantages in frequent observa‐ tions of TC activities, but they lack in accurate TC eye positions, intensity analysis, and horizontal structures. ARCHER is an advanced algorithm in fixing the TC center positions from both PMW and IR/VIS sensors in near real‐time with high confidence. The ARCHER track provides excellent TC positions for monitoring of TC activities and initialization in model TC data assimilation processes. The TC live track from PMW sensors will display evolutions of TC structures and intensity for purposes of better monitoring and forecasts.
