P1.53 An objective method to predict near real time rapid intensification of tropical cyclones using satellite passive microwave observations

Tuesday, 11 May 2010
Arizona Ballroom 7 (JW MArriott Starr Pass Resort)
Christopher S. Velden, CIMSS/Univ. of Wisconsin, Madison, WI; and C. Rozoff, A. Wimmers, M. Sitkowski, M. E. Kieper, J. Kossin, J. Hawkins, and J. Knaff

The physical processes associated with rapid intensification (RI, we define as an increase in the analyzed maximum sustained surface wind of at least 30 kts in 24 hrs) of tropical cyclones (TCs) for the most part remain a mystery. Predicting these events is one of the most challenging aspects for a TC forecaster. Recent research efforts have found some predictive skill in analyzing simple environmental characteristics, such as the vertical wind shear or sea surface temperatures (e.g., Kaplan and DeMaria 2003; Wea. Forecasting), but a relatively low probability of RI detection still remains in statistical models relying only on environmental predictors. The inclusion of infrared satellite imagery captured from Geostationary Operational Environmental Satellites (GOES) has added fractional skill (Kaplan and DeMaria 2006; 27th AMS Conf. on Hurr.). However, infrared imagery can have serious limitations in unambiguously depicting the distribution of important quantities related to internal storm dynamics. On the other hand, passive microwave imagery (PMW) provided by low Earth orbiting (LEO) satellites can peer through the non-raining overlying cirrus cloud canopy to more effectively monitor evolution of TC convective structures and organization. While the temporal coverage of PMW is rather sparse in comparison with geostationary satellite imagery, the information that is periodically captured may prove invaluable in the prediction of RI.

Therefore, our research venture involves the search for an objectively-determined indicator of RI in multiple channel PMW (i.e., 19, 37, 85-92 GHz), and evaluating its potential utility in statistical RI forecasting. Using data from past TCs, we are developing algorithms to objectively seek physically meaningful patterns found in the PMW imagery, and will evaluate the statistical skill of such features using state-of-the-art RI forecasting schemes. For example, it has been suggested through primarily subjective analysis that symmetric ring patterns in developing storms, as detected in the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and the Advanced Microwave Scanning Radiometer – Earth Observing System (AMSR-E) at 37 GHz, most often occur just before or as storms undergo RI, and that the patterns appear particularly skillful when the operational Statistical Hurricane Intensity Prediction Scheme (SHIPS) suggests a reasonable probability of RI (Kieper 2008; 28th AMS Conf. on Hurr.). Our initial pursuit is to develop an objective algorithm to validate the potential of this particular predictor. Beyond this will be a more general search for a wide variety of physically sensible predictors in all of the available PMW channels, including those in the operational PMW sensors onboard the DMSP Special Sensor Microwave/Imagers (SSM/I and SSMI/S).

Our PWV data analysis, primarily being conducted at the Cooperative Institute for Meteorological Satellite Studies (CIMSS), should contribute toward a collaborative effort with the Cooperative Institute for Research in the Atmosphere (CIRA) and the National Oceanic and Atmospheric Administration's Hurricane Research Division (NOAA/HRD) involving multi-sensor product development to improve statistical RI forecast schemes. The potential skill of predictors being developed using the PMW imagery are evaluated through two statistical rapid intensification schemes: an operational linear discriminant index based on past SHIPS and GOES data (Kaplan and DeMaria 2006), and a Bayesian probabilistic model developed at CIMSS (Kossin and Sitkowski 2009; Mon. Wea. Rev.) which is also based on SHIPS and GOES data. We are particularly interested in finding indicators that can contribute to predicting RI at forecast lead times of 6-12 hours. While the primary aim of this work is to improve operational forecasting, increased theoretical understanding of RI may result from the systematic study of a large dataset of PMW imagery as well. Progress on this research effort will be presented at the meeting.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner