G O D A E S P E CIA L ISSU E FEATURE Applications of Satellite-Derived Ocean Measurements to Tropical Cyclone Intensity Forecasting B Y G U S TAVO GONI , M A R K D E M A RIA , JO H N KNAFF, C H A RLES SAMP SON , ISAA C G I N I S , F R ANCIS B R I NGAS , ALBERTO MAVU ME, C HRI S LAU E R , I . I . L I N , M.M. ALI, PAU L S ANDE RY, S I LVANA RAMOS BUA RQU E , K I RY ONG KANG , AV ICH A L ME HRA , E R I C C H ASS I GNE T, AND GEOR GE HALLIW E LL 176 Oceanography Vol.22, No.3 asgjkfhklgankjhgads ABSTRACT. Sudden tropical cyclone (TC) intensi!cation has been linked with tropical cyclone heat potential (TCHP), high values of upper ocean heat content contained in mesoscale features, particularly has been shown to play a more impor- warm ocean eddies, provided that atmospheric conditions are also favorable. tant role in TC intensity changes (Shay Although understanding of air-sea interaction for TCs is evolving, this manuscript et al., 2000). TCHP shows high spatial summarizes some of the current work being carried out to investigate the role that and temporal variability associated with the upper ocean plays in TC intensi!cation and the use of ocean parameters in oceanic mesoscale features. TC intensi!- forecasting TC intensity. cation has been linked with high values of TCHP contained in these mesoscale INTRODUCTION problem’s complexity and because many features, particularly warm ocean eddies, Tropical cyclones (TCs) occur in of the errors introduced in the track provided that atmospheric conditions are seven ocean basins: Tropical Atlantic, forecast are translated into the intensity also favorable. Because sustained, in situ Northeast Paci!c, Northwest Paci!c, forecast (DeMaria et al., 2005). ocean observations alone cannot resolve Southwest Indian, North Indian, Leipper and Volgenau (1972) !rst global mesoscale features and their Southeast Indian, and South Paci!c recognized the importance of ocean vertical thermal structures, di"erent (Figure 1). TC intensi!cation involves thermal structure in TC intensi!ca- indirect approaches and techniques are several mechanisms, including TC tion. Although sea surface temperature used to estimate TCHP. Most of these dynamics, upper ocean interaction, and (SST) plays a role in TC genesis, the techniques use sea surface height obser- atmosphere circulation. In general, accu- ocean heat content contained between vations derived from satellite altimetry, racy of TC intensity forecasts has lagged the sea surface and the depth of the a parameter that provides information behind TC tracking because of the 26°C isotherm (D26), also referred to as on upper ocean dynamics and vertical Figure 1. Global map showing the tracks of tropical cyclones (category 1 and above) during the period 2000–2008, with green circles indicating where they formed. "e background color is the satellite-derived mean sea surface temperature during the same years, for June through November in the Northern Hemisphere, and November through April in the Southern Hemisphere. Oceanography September 2009 177 thermal structure. #is article highlights the 20°C isotherm in tropical regions accurate than, those from much more the importance of collecting a variety (Goni et al., 1996). Climatological rela- general, dynamical models. For recent of data, particularly satellite-derived tionships are used to determine D26 category 5 hurricanes, TCHP input observations, for tropical cyclone inten- from the depth of the 20°C isotherm. improved SHIPS forecasts by about 5%, si!cation studies. NHC forecasters use these TCHP !elds with larger improvements for indi- qualitatively for their subjective TC vidual storms (Mainelli et al., 2008). A NORTH ATLANTIC OCEAN intensity forecasts and quantitatively validation performed on 685 Atlantic An operational satellite-altimetry-based in the Statistical Hurricane Intensity SHIPS forecasts from 2004–2007 TCHP analysis was implemented at Prediction Scheme (SHIPS; DeMaria shows that the average improve- the National Oceanic and Atmospheric and Kaplan, 1994). SHIPS is an empirical ment of SHIPS due to the inclusion of Administration (NOAA) National model that uses a multiple regression TCHP and Geostationary Operational Hurricane Center (NHC) in 2004 method to forecast intensity changes Environmental Satellite (GOES) SST data (Mainelli et al., 2008). #is approach out to 120 h. #e 2008 version of is as much as 3% for the 96-h forecast uses sea surface height anomaly !elds SHIPS includes 21 predictors, mostly (Figure 2, le$). Nearly all improvements derived from altimetry and historical related to atmospheric conditions. #e at the longer forecast intervals are due hydrographic observations in a statistical ocean predictors are SST and TCHP. to TCHP because that input is averaged regression analysis to determine the Despite its simplicity, SHIPS forecasts along the storm track. Although not as depth of the main thermocline, usually are currently comparable to, or more large as the sample of just the category 5 hurricanes, this result indicates that Gustavo Goni ([email protected]) is Oceanographer, National Oceanic and TCHP input improved the operational Atmospheric Administration (NOAA) Atlantic Oceanographic and Meteorological SHIPS forecasts, especially at the longer Laboratory (AOML), Miami, FL, USA. Mark DeMaria is Chief, NOAA National forecast intervals. Environmental Satellite, Data, and Information Service (NESDIS), Regional and Mesoscale Altimetry observations are also used Meteorology Branch, Fort Collins, CO, USA. John Knaff is Meteorologist, NOAA NESDIS to initialize the ocean component of a Regional and Mesoscale Meteorology Branch, Fort Collins, CO, USA. Charles Sampson is coupled hurricane prediction model with Meteorologist, Marine Meteorology Division, Naval Research Laboratory, Monterey, CA, !elds extracted from data-assimilative USA. Isaac Ginis is Professor of Oceanography, University of Rhode Island, Graduate School ocean hindcasts generated as part of of Oceanography, RI, USA. Francis Bringas is Research Associate, Cooperative Institute for the Global Ocean Data Assimilation Marine and Atmospheric Sciences, University of Miami, Miami, FL, USA. Alberto Mavume Experiment (GODAE). #ese hindcasts is Chair, Marine Science and Oceanography Group, Eduardo Mondlane University, Maputo, rely heavily on altimetry to properly Mozambique. Chris Lauer provides computer programming and technical support to the locate mesoscale features, such as ocean NOAA National Hurricane Center, Tropical Prediction Center, Miami, FL, USA. I.-I. Lin is currents and eddies. Halliwell et al. Associate Professor, Department of Atmospheric Sciences, National Taiwan University, (2008) examined this initialization Taipei, Taiwan. M.M. Ali is Scientist and Head, Oceanography Division, National Remote approach in ocean model simulations of Sensing Centre, Hyderabad, India. Paul Sandery is a member of the ocean forecast- the response to hurricane Ivan (2004) ing team, Center for Australian Weather and Climate Research, Melbourne, Australia. in the Northwest Caribbean and Gulf Silvana Ramos-Buarque is Meteorologist, Météo-France/Mercator Ocean, Ramonville St. of Mexico. #is simulation was driven Agne, France. KiRyong Kang is an oceanographer at the National Typhoon Center, Korea by quasi-realistic forcing generated by Meteorological Administration, Jeju, South Korea. Avichal Mehra is Physical Scientist, NOAA blending !elds extracted from the Navy National Centers for Environmental Prediction, Environmental Modeling Center, Camp Coupled Ocean/Atmospheric Mesoscale Springs, MD, USA. Eric Chassignet is Professor and Director, Center for Ocean-Atmospheric Prediction System atmospheric model Prediction Studies, Florida State University, Tallahassee, FL, USA. George Halliwell is with higher-resolution !elds obtained Research Scientist, NOAA AOML, Miami, FL, USA, and Professor, Rosenstiel School of Marine from the NOAA/Atlantic Oceanographic and Atmospheric Science, University of Miami, Miami, FL, USA. and Meteorological Laboratory- 178 Oceanography Vol.22, No.3 135 3 SHIPS 3 STIPS t t 2 2 393 234 544 738 1 1 640 Percent Inprovemen Percent Inprovemen 890 822 0 0 01224364860728496 108 120 01224364860728496 108 120 Forecast Interval [h] Forecast Interval [h] Figure 2. (left) Percent improvement of the 2004–2007 operational Statistical Hurricane Intensity Prediction Scheme (SHIPS) forecasts for the Atlantic sample of over-water cases west of 50°W due to the inclusion of input from altimetry-derived tropical cyclone heat potential (TCHP) and GOES- derived sea surface temperature (SST) fields. (right) Percent improvement resulting from the use of TCHP information in the Statistical Typhoon Intensity Prediction Scheme (STIPS). "is homogeneous comparison between STIPS with TCHP and STIPS without TCHP is based on forecasts of 63 western North Pacific tropical cyclones. "e number of cases used at each forecast time is given at the top of each bar. Hurricane Research Division H*WIND climatological ocean temperature !eld more intense (Figure 3, right panel). product (Powell et al., 1998) to resolve prior to the passage of a hurricane. For #is assimilation improved the actual the inner-core structure of the storm. the 2008 Atlantic hurricane season, storm’s intensity forecast with respect to Halliwell et al. (2008) concluded that for the full version of this procedure was that obtained without assimilating the the ocean component of the Hurricane implemented in the NOAA Geophysical altimetry !elds. Weather Research and Forecast System Fluid