Interactions between Hurricane Catarina (2004) and

Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, C07002, doi:10.1029/2009JC005974, 2010

for

Full Article

Interactions between Hurricane Catarina (2004) and warm core rings in the South Atlantic Ocean M. L. Vianna,1 V. V. Menezes,2 A. B. Pezza,3 and I. Simmonds3 Received 9 November 2009; revised 25 January 2010; accepted 19 February 2010; published 3 July 2010.

[1] The unexpected evolution of the first recorded South Atlantic Hurricane Catarina over waters with homogeneous sea surface temperatures (SST) of 24°C in March 2004 was a challenge to the weather forecast community. This work concentrates on a thorough data‐driven comparative analysis to make reliable diagnostics of the role of the ocean in the genesis and evolution of Catarina. We used several high‐resolution multisatellite‐derived products, including three microwave‐based SST data sets, multisatellite collinear data of sea surface height (SSH) anomalies, significant wave heights and wind speeds, four QuikSCAT ocean surface wind vector products (including the 12.5 km resolution swath data), daily fields of absolute objectively analyzed SSH and corresponding geostrophic currents, and Argo floats. The synergic use of these data sets showed that Catarina interacted strongly with four warm core rings (WCRs), forcing upwelling of isotherms and mixed layer waters. These interactions minimized the known negative SST feedback, as attested by the SST differences being less than 1.2°C. Although the SST in the region was around 24°C, below the Palmén threshold, the surface air temperatures were 14°C which still furnished a large air‐sea temperature gradient capable of extracting large enthalpy fluxes from the WCRs influenced by Ekman pumping. It is shown here that Catarina achieved category 1 over the ocean on 26 March with its maximum intensity of 34 m/s seen in the 12.5 km swath winds. Citation: Vianna, M. L., V. V. Menezes, A. B. Pezza, and I. Simmonds (2010), Interactions between Hurricane Catarina (2004) and warm core rings in the South Atlantic Ocean, J. Geophys. Res., 115, C07002, doi:10.1029/2009JC005974.

1. Introduction [2] Recent advances in understanding the genesis and development of tropical cyclones and hurricanes/typhons indicated that the role of upper ocean structure is of much more fundamental importance than previously thought [Hong et al., 2000; Shay et al., 2000; Jacob et al., 2000; Goni and Trinanes, 2003; Kaplan and DeMaria, 2003; Emanuel et al., 2004; Lin et al., 2005, 2009; Scharroo et al., 2005, 2006; Pun et al., 2007; Wu et al., 2007; Oey et al., 2007; Mainelli et al., 2008; Jaimes and Shay, 2009]. [3] Studies of the interaction between Tropical Cyclones (TCs) and ocean features, such as warm core rings (WCRs), Cold Core Rings (CCRs) and currents, show abrupt cyclone intensification or weakening when they cross over such ocean features. These studies also show that dynamic topography, estimated by satellite altimetry, is a more reliable proxy for the upper ocean heat content than sea surface temperature (SST) [Shay et al., 2000; Goni and Trinanes, 2003; Lin et al., 2005; Scharroo et al., 2006].

1

VM Oceânica LTDA, San Jose dos Campos, Brazil. VM Oceânica LTDA, Arraial do Cabo, Brazil. School of Earth Sciences, University of Melbourne, Parkville, Victoria, Australia. 2 3

Copyright 2010 by the American Geophysical Union. 0148‐0227/10/2009JC005974

[4] Marks et al. [1998], Hong et al. [2000], and Shay et al. [2000] have analyzed especially this issue in the case of Hurricane Opal, the major hurricane that formed in the Gulf of Mexico in September 1995. They observed that the rapid intensification of Opal coincided with its encounter with a WCR. The wind field increased from 35 m/s to 65 m/s and the thermocline depth decreased by approximately 50 m (Pre‐Opal depth was between 175 to 200 m). The high resolution coupled numerical model study of Hong et al. [2000] was able to reproduce this phenomenon during the critical 72 hours of the hurricane intensification phase, indicating that the model physics is essentially correct. Other observational studies from Hurricanes Katrina, Mitch, and Bret in the North Atlantic, and supertyphoon Maemi and typhoon Imbudo in the North Pacific gave similar results [Goni and Trinanes, 2003; Scharroo et al., 2005, 2006; Wu et al., 2007; Lin et al., 2005]. Hurricane Katrina in 2005, one of the most devastating tropical cyclones ever to hit the United States, intensified from category 1 to 5 when it passed over the Loop Current and a WCR and weakened to category 3 after it passed over a CCR prior to making landfall [Jaimes and Shay, 2009]. These features, while clearly visible by altimetry, could not be detected in the SST fields, due to the warm homogeneous surface layer of 30°–31.5°C. The same occurred with supertyphoon Maemi, where the ocean surface exhibited uniformly warm SST between 30 and 31°C, while it encountered three WCRs in the southern eddy zone of the western North Pacific.

C07002

1 of 19

C07002

VIANNA ET AL.: HURRICANE CATARINA (2004)

[5] One known air‐WCR process that reduces the negative feedback of the ocean forced by the hurricane is the insulation of surface waters against cooling because of the deep thermocline in the warm eddy. It has been found that when air‐WCR interaction is incorporated into coupled cyclone‐ocean models such as CHIPS (Coupled Hurricane Intensity Prediction System), estimates of hurricane intensity can be improved markedly [Hong et al., 2000; Emanuel et al., 2004; Lin et al., 2005; Wu et al., 2007]. [6] The cyclone Catarina developed between 19 and 28 March 2004 near the Brazilian coast (25–29°S) and became the first documented hurricane in the South Atlantic Ocean [Silva Dias et al., 2004; Pezza and Simmonds, 2005; McTaggart‐Cowan et al., 2006; Bonatti et al., 2006; Pereira Filho et al., 2010]. In addition to the damage it caused on land, this cyclone has two unique aspects: it formed in a region previously thought to be hurricane free, and it took place over SSTs of 24°C–25°C, slightly cooler than the 26°C usually known to be the minimum SST required for the intensification of tropical cyclones into a hurricane status. Colder SSTs were previously observed also in hurricanes Ivan, Karl and Epsilon [Lawrence and Pelissier, 1981; Beven et al., 2008]. While the initial phases of Catarina seemed to be well accounted by atmospheric processes only, all forecast models grossly underestimated its maximum observed intensity (e.g. the eye’s low surface pressure was lower than predicted) and gave conflicting trajectories [Silva Dias et al., 2004]. [7] Almost all published studies of Catarina to date with the exception of Pereira Filho et al. [2010] rely only on purely atmospheric processes in trying to interpret the uncertainties in the genesis and evolution of Catarina. Different aspects of Catarina’s life cycle and the large scale reasons for its genesis have been investigated [Pezza and Simmonds, 2005, 2008; McTaggart‐Cowan et al., 2006; Veiga et al., 2008; Pezza et al., 2009; Pereira Filho et al., 2010], but many facets of its transition into a TC are puzzling. Pezza and Simmonds [2005] and Pezza et al. [2009] show that Catarina would not have formed without the key dynamic interplay triggered by a high latitude blocking and low vertical wind shear. Veiga et al. [2008] found a major energetic change in the large scale energetics associated with the point in time when Catarina began to make an anticyclonic loop back towards the continent. That behavior had been previously associated with the beginning of the Tropical Transition (TT) [Pezza and Simmonds, 2005]. McTaggart‐Cowan et al. [2006] explored the blocking’s dynamic participation via Potential Vorticity (PV) analysis, showing that repeated injections of transient troughs and ridges sustained the blocking pattern. The Catarina system clearly evolved under conditions that had deviated from the classical WISHE (Wind‐Induced Heat Exchange) and CISK (Conditional Instability of Second Kind) mechanisms typically observed in TC‐prone areas within the boundary layer of the tropical oceans. Without high resolution air‐sea flux data it is unclear to what extent WISHE participated during Catarina’s development. In the sensitivity experiment S4 posed by Nguyen et al. [2008], the wind speed dependence of the surface fluxes were capped at 10 m/s. Those results suggest that the air‐sea flux acts to replenish boundary layer theta‐e, which continues to fuel the deep convective activity that provides the low‐level convergence and spin‐up for the storm.

C07002

[8] In the works mentioned above the major role of the ocean is based only on SST and it is well known that Catarina developed under quasi‐isothermal surface waters of 24°C. On the other hand, Catarina developed in an area with a conspicuous subsurface ocean mesoscale eddy channel (EC) around 30°S, first discovered with GEOSAT (GEOdetic SATellite) altimetry data [Forbes et al., 1993]. These authors described the spectral characteristics of variability in this region and found dominance of wavelengths of 400 km and periods less than 100 days. The propagation was found to be west‐southwestward, with a slow phase speed between 0.8 to 1.6 km/d. The results suggest that the eddies are part of a planetary wave channel. The connection of this wave field with the Agulhas westward propagating retroflection eddies at 30°S [e.g., Gordon and Haxby, 1990] is still an open question. Ducet et al. [2000] describes some zonal and slender eddy kinetic energy distributions in the world ocean by multisatellite altimetry, and also records the presence of this EC in the South Atlantic (SA). Recently, Vianna et al. [2007] have found that this EC is well represented in the ocean Mean Dynamic Topography (MDT). The latter work also showed that in the EC region the daily multisatellite altimeter‐derived geostrophic fields correlates well with shipboard ADCP currents. The EC region was surveyed by the WOCE A10 line in 1992, which was repeated in 2003 with another cruise made by the Mirai (28–30°S, 47°W– 1°W, 7–26 November 2003). The latter cruise was conducted 4 months before Catarina. It traversed a few eddies seen by the altimeter‐derived Dynamic Height (DT) and Geostrophic Circulation, suggesting the EC presence. As in other eddies, this EC is not clearly seen in SST signatures, and still requires detailed studies about its nature. [9] Could the missing link which prevented a skillful forecasting of the anomalous intensification of Catarina be due to the lack of knowledge of the presence of this subsurface EC with which it could have interacted, in analogy to, e.g., supertyphoon Maemi or hurricane Opal and Katrina? [10] The present work documents the daily surface evolution of Catarina in parallel with the subsurface eddy field, showing that indeed daily intensity variations of Catarina occurred after its passage over WCRs, which were quiescent before and after its passage, but were suddenly deformed during interaction. To this end we use a synergy of several high‐level data sets obtained from orbital microwave cloud‐ penetrating sounders (three satellite altimeters and four QuikScat scatterometer products) and three high resolution SST products. We also use the GRACE‐based/altimeter‐ derived MDT for estimates of absolute geostrophic current, and vertical temperature/salinity profiles from Argo floats to support the subsurface ocean analysis. Section 3 presents the Catarina track, section 4 presents the surface‐subsurface ocean preconditions before the onset of Catarina, section 5 presents the evolution of these ocean structures in interaction processes, section 6 presents a analytical synthesis of the processes observed, and section 7 presents the main conclusions.

2. Data Sets 2.1. Sea Surface Height, Waves, and Wind Speeds [11] To retrieve as much information as possible we used sea surface height anomaly (SSHA), significant wave height

2 of 19

C07002


(SWH) and wind speed data sets aquired by satellite altimeters (ERS2/ENVISAT, JASON‐1, GFO), distributed by the Real Time Ocean Environment Project/U.S. Navy NRLSSC (Naval Research Laboratory at Stennis Space Center). These data sets span from January 2003 to December 2005. The ENVISAT and the JASON‐1 satellites are dual‐frequency altimeters, the former operating at the Ku band (13.6 GHz) and the S band (3.2 GHz) and the latter at the Ku band and at the C band (5.3 GHz). The GFO mission was ended as of October 2008 and operated only on the Ku band. [12] Data from each satellite has been processed by the NRLSSC team, with the standard altimetric corrections. The SSHA processing includes the inverse barometer correction, interpolation onto reference ground tracks at 1 second sampling (approximately 7 km), intercalibration among the various data sets, and referencing to the NRLSSC 1993– 2001 mean sea surface (MSS) [see Jacobs et al., 2002; Kara et al., 2008, and references therein]. The inverse barometer correction applied by the NRLSSC is the standard one used in the altimeter Geophysical Data Record (GDR). It is made to eliminate variations in the sea surface height due to long‐ wavelength atmospheric pressure variations (atmospheric loading) and is computed from the NCEP (National Centers for Environmental Prediction) or ECMWF (European Centre for Medium‐Range Weather Forecasts) weather prediction models. A 1 mbar atmospheric pressure change corresponds to a linear response of the sea level of about 1 cm. This correction is normally used for the study of ocean dynamics as the case here, and have to be changed for the study of mean sea level variations or when altimeter‐derived atmospheric pressure is sought for [Dorandeu and Le Traon, 1999; Carrère et al., 2009]. [13] The wind speeds are derived using the Gourrion empirical algorithm [Gourrion et al., 2002]. Although both ENVISAT and JASON‐1 are dual‐frequency altimeters the algorithm uses information only from the Ku band, besides the SWH estimates. It is evaluated for 10 m above the sea surface and is considered to be accurate in the absence of rain, since the Ku band is very sensitive to rain. Quilfen et al. [2006] proposed a new methodology for hurricanes rainy conditions that uses the dual‐frequency observations to estimate wind speeds more accurately. They showed that for hurricane Isabel the use of the C band to correct the Ku band improves the JASON‐1 wind speed estimates. However, to use this new algorithm it is necessary to extend it to ENVISAT data (S band), but this is outside the scope of present work. [14] Durrant et al. [2009] validated the JASON‐1 and ENVISAT SWH against in situ data from the National Data Buoy Center (NDBC) and the Marine Environmental Data Services (MEDS) networks. They found that the JASON‐1 estimates are consistent with in situ values even in high SWH as in hurricane force winds, and require no corrections. On the other hand, ENVISAT measurements tend to overestimate low SWH and underestimates high SWH. They proposed a linear correction to reduced this effect given by SWHC(m) = 1.093 * SWHR(m) − 0.233, where SWHC is the corrected SWH and SWHR is the raw one. We have applied this correction to the ENVISAT data. Despite the possibility of underestimation of hurricane force winds and SWH, the use of along‐track altimeter‐derived data is

C07002

becoming a standard for monitoring these parameters in the hurricane conditions [e.g., Scharroo et al., 2005; Quilfen et al., 2006; Oey et al., 2007]. [15] To support our study, an objective daily updated mapping of collinear SSHA data, usually of three concurrent satellites (ERS2 or ENVISAT and JASON‐1 and GFO) was made [Bretherton et al., 1976]. For the mapping, all SSHA collinear data passed by a quality control (QC) that includes outlier removal, interpolation of small gaps, and a low‐pass filtering by Singular Spectrum Analysis (SSA), mainly to remove unwanted internal wave signatures present in the collinear data. The SSA cutoff scale is 80 km, since we are interested only in the meso‐ and large scales. [16] The gridded fields have a spatial resolution of 1/8° × 1/8°. They were computed by use of a binned space‐time Gaussian covariance function with decorrelation scales of 150 km in latitude and longitude and 15 days in time. The grid limits are 55°W–29°W and 33°S–22°S. For mesoscale features away from the Equator, no meridional or zonal propagation speeds need to be used in the prescribed gaussian covariance function of our choice. These SSHA analyzed fields were then summed to the high‐resolution 0.1° × 0.1° EGM08‐based MDT (linearly interpolated to 1/8° × 1/8°) [Vianna and Menezes, 2010] to obtain the daily Absolute Dynamic Topography (here denoted as SSH), as a proxy for the upper thermocline depth field in high spatial resolution. It should also be noted that from the absolute daily SSH the daily absolute geostrophic velocity field was also derived. 2.2. Sea Surface Temperatures [17] Three different higher‐resolution daily gridded sea surface temperature (SST) products are used, all spanning the period between 1 and 31 March 2004. The first data set is the daily optimally interpolated microwave SST (MW‐OI) made available by Remote Sensing Systems, Inc (REMSS). It is a daily grid with spatial resolution of 0.25° × 0.25° based on the AMSR‐E (Advanced Microwave Scanning Radiometer) from the EOS‐NASA AQUA satellite and on the TMI/TRMM (Microwave Imager/Tropical Rainfall Measuring Mission) sensors. The second is the newest Reynolds blended Objectively Interpolated based on the AVHRR (Advanced Very High Resolution Radiometer) and AMSR‐E sensors. It has the same resolution as the REMSS product, uses in situ data from ships and buoys and include a large‐scale adjustment of satellite biases with respect to the in situ data [Reynolds et al., 2007]. In addition to SST fields, the Reynolds product includes the corresponding mapping errors and the sea surface temperature anomaly (SSTA) relative to 1971–2000 monthly climatology. The last data set is from the US Navy MODAS‐2D (Modular Ocean Data Assimilation System) with a higher spatial resolution of 0.125° × 0.125° [Barron and Kara, 2006]. It is available at the NRLSSC home page at http://www7320.nrlssc.navy. mil/modas/. We refer to the 2007 version of OI‐AVHRR‐ AMSR as Reynolds and MODAS‐2D as MODAS. 2.3. Temperature and Salinity Profiles From Argo Floats [18] We use the temperature and salinity profiles collected by Argo free drifting subsurface floats. Argo is an international project that deployed already 3000 floats around the

3 of 19

C07002


global ocean. The Argo profiling float typically measures the temperature and salinity in the upper 2000 m ocean at 10 day intervals when it rises to the surface. For more details see the Argo web site at http://www.argo.ucsd.edu. [19] From the Argo data base made available through the CORIOLIS web site, we selected the vertical profiles of temperature and salinity from the region limited by 33°S– 22°S and 55°W–29°W. In this region there are profiles available since June 2003, and we choose all profiles from this date to December 2005. A quality control against outliers, spikes, lack of consistency, etc, was subsequently performed. From the complete 269 profiles which passed the QC tests, selection of profile data in the depth range between 10 and 500 m was made, and then linearly interpolated to standard depths with 5 m of vertical resolution. [20] Various parameters were retrieved from these Argo profiles: the near sea surface temperature (MLT) and salinity (MLS), both at 20 m depth, the mixed layer depth (MLD), the depth of 17°C isotherm (D17), the heat content of the mixed layer (QML), and the total D17‐to‐surface (QD17) columns. The MLD was determined through the threshold method, for which the MLD is the depth at which temperature or potential density changes by a given threshold value relative to the one at a near‐surface reference depth [Montegut et al., 2004]. As shown by these authors the choice of the threshold value and the reference depth is rather arbitrary [see Montegut et al., 2004, Table 1]. Here the MLD was computed as the depth where the potential density changes by 0.125 kg/m3 relative to the one at 10 m depth. This threshold value is a usual value adopted, e.g., by the Levitus MLD climatology. The heat content was computed as Z Qh ¼

0

h

cp ð h Þ dz;

ð1Þ

where r and cp are the depth‐dependent sea water density and heat capacity, both computed from potential temperatures (), salinities and pressures. is referenced to the surface, and ‐h is the mixed layer or D17 depths. 2.4. Upper Layer Thickness [21] The general availability of Argo floats in the study area since 2003, surfacing every 10 days, and the daily high resolution SSH fields constructed for the period, makes feasible the construction of a data set of upper layer thickness through regressions between upper thermocline isotherm depths and SSH, since the varying SSH field largely reflects changes in the thermocline depth. Sudden variations of upper thermocline isotherm depths offer one important measure of the dynamic impact of hurricanes over the underlying vertical ocean thermal structure. [22] For this region, we chose the depth of the 17°C isotherm as the parameter representing the lower limit of the upper ocean thermally active layer. The regression between the D17 from Argo with satellite‐derived SSH at each Argo position and day was computed at a 95 percent confidence level and gave the equation (2), with a correlation of 0.45 b ðmÞ ¼ 42:66ð22:12Þ þ 2:55ð0:64Þ * SSH ðcmÞ: D17

ð2Þ

C07002

[23] It should be noted that before this computation, the time series of the daily SSH fields have been filtered to get rid of (noise) signals with statistically dominant periods of less than 15 days. For this, the SSH fields were analyzed with Empirical Orthogonal Functions (EOF) method and the time principal components (PC) were low‐pass filtered in the SSA sense [Vianna et al., 2007] and then reprojected into the daily fields. It is worth noting here that the SSA low‐pass filtering method used to get rid of these small periods has some differences to the traditional Fourier‐based method: it is based on SSA expansions, and the selection rule for the sequence of SSA data‐adaptive filters (RCs) is based on the identification of the spectral maximum period of each RC, obtained via the Maximum Entropy Method (MEM) of spectral analysis. The selected RCs here are those in which these periods are larger than 15 days. We remark that, as opposed to Fourier filtering, RC 15 day filtering does not cut away daily scale nonperiodic signals, it only cuts away statistically dominant (energy wise) oscillating patterns with periods less than 15 days, without losing boundary points. The final SSA‐filtered data is obtained by summing up all of the selected RCs. There are advantages in using such methods, especially for elimination of high amplitude noise, without excessive smoothing of the original signal amplitudes of the usually nonperiodic series [Vianna et al., 2007]. 2.5. Wind Velocity and Stress Fields From QuikSCAT [24] We used Seawinds ocean surface vector wind (OSVW) products distributed by two research groups, one from the REMSS Inc. and the other from the PODAAC/JPL/ NASA (Physical Oceanography Distributed Active Archive Center/Jet Propulsion Laboratory/National Aeronautics and Space Administration). It is important to note that these products have different processing algorithms. [25] Seawinds is an active microwave radar scatterometer operating at Ku band (13.4 GHz) on board the QuikSCAT satellite, launched in June 1999. The OSVW retrievals are referred to 10 m and are roughly equivalent to an 8–10 minute mean surface wind. [26] The REMSS products are made available in two processing levels: level 3 (version 3a) and level 2B. The level 3 consists of global gridded 0.25° × 0.25° OSVW in separated fields for the ascending and descending passes. In the Catarina region, the ascending pass is around 0900 UTC (morning) and the descending one at 1900–2000 UTC (evening). The level 2 product provides OSVW in swaths that are 1800 Km wide and have a nominal spatial resolution of 25 Km. These data sets are available since 4 October 2006, after they were reprocessed utilizing the new SSM/I (Special Sensor Microwave Imager) and TMI rain rates as inputs [Hilburn et al., 2006]. Both data sets are produced using the Ku‐2001 geophysical model function (GMF) that has the capability to retrieve winds up to 70 m/s [Wentz, 2001; Smith et al., 2003]. [27] The JPL products are also distributed in the same two processing levels. The level 3 data set has the same resolution as the REMSS and also is provided in separated maps for both the ascending and descending passes. On the other hand, the JPL level 2B is available in two resolutions: 25 Km and 12.5 Km. The 12.5 Km data set is obtained using a postprocessing technique and is available since 2003.

4 of 19

C07002


Bourassa et al. [2003] and Brennan et al. [2009] show that the swath data with 12.5 Km seems to be more adequate for resolving wind vectors closer to hurricane eyes, and also for near coastal winds, as compared to coarser resolution products. The JPL products use the QSCAT‐1/F13 GMF with the Direction Interval Retrieval with Threshold Nudging (DIRTH) wind vector solutions. The QSCAT‐1 has the capability to retrieve winds up to 50 m/s, and shows a tendency to underestimate high winds [Perry, 2001]. [28] It is important to note that altimeters and QuikSCAT Ku band radar‐returned data are sensitive to rain presence. For low wind speeds, heavy rain conditions can produce unrealistic high speeds. For high wind speeds, like in hurricanes, QuikSCAT may underestimate the real wind. Besides this, high rain rate contamination tends to produce erroneous cross‐track vectors. These effects are known to be stronger at category 3 plus hurricanes [Beven et al., 2008; Brennan et al., 2009]. [29] From OSVW fields we calculated the corresponding wind stress using the bulk formula t = raCdU210, where ra is the density of air, Cd is the drag coefficient and U10 is the wind speed at 10 m. The Cd adopted here is the same one used by Oey et al. [2007] Cd * 103 ¼ 1:2; U10 11m=s ¼ 0:49 þ 0:065 * U10 ; 11 < U10 19m=s 2 ¼ 1:364 þ 0:0234 * U10 0:00023158*U10 U10 > 19m=s:

[30] For the swath products, before computing wind stresses, the OSVW fields were mapped onto a regular grid with a grid resolution closer to original data. Note that we did not perform interpolation, indeed each swath data point was nudged to the nearest grid point. The data sets with 25 Km were mapped onto a 0.25° × 0.25° grid, and the 12 Km to 1/8° × 1/8°. We refer to the swath data sets from JPL as JPL‐2B (25 Km) and JPL‐2BHR (12 Km) respectively; from REMSS as RSS‐2B, and for level 3 products as JPL‐L3 and RSS‐L3.

3. Catarina Track and Meteorological Setting [31] The total lifetime of Catarina was approximately 8 days, between 19 and 28 March 2004 as shown in Figure 1a. Note that in the present work we have used the satellite‐derived track analysed by Roger Edson and presented by McTaggart‐Cowan et al. [2006]. [32] Catarina started as an extratropical cyclone weak system close to the coast at 25°S, drifting seawards to the southeast during 3 days over surface waters of less than 26°C. When it reached 28°S on 21 March, the SST was between 24.5°C and 25.5°C, while surface air temperature was always lower along the track [Pezza et al., 2009]. At 0000 UTC 23 March (31°S, 36.7°W) it underwent an abrupt anticyclonic track reversal, veering to the northwest, and then stabilizing its translation into an almost westward zonal track centered on 29°S. At 0600 UTC 26 March Catarina reached hurricane status (category 1 on the Saffir‐Simpson scale), with a well‐developed eye. This intensity was sustained up to 27 March [Pezza and Simmonds, 2005;

C07002

McTaggart‐Cowan et al., 2006]. On 28 March it weakened rapidly after landfall at 1800 UTC. The low pressure in the eye could not be directly measured, but was estimated by the Pennsylvania State University (PSU) fine mesh grid hurricane model as slightly above 974 hPa in its most intense stage (when the eye’s diameter was 60 km) just prior to landfall [McTaggart‐Cowan et al., 2006; Pezza and Simmonds, 2008; Pezza et al., 2009]. This pressure is consistent with the Mean Sea Level Pressure (MLSP) derived by R. Edson (972 hPa on March 2800 UTC). Most global forecast models, however, significantly underestimated the cyclone’s intensity. [33] Pezza and Simmonds [2005] described the Catarina rare loop trajectory obtained via an automatic tracking scheme based on high resolution MSLP, arguing that the beginning of the Tropical Transition (TT), around 23 March, coincided with the moment at which the trajectory veered back towards the continent following the upper level easterly flow embedded in a blocking system. The trajectory derived from high resolution MSLP is very similar to the satellite‐derived track analyzed by R. Edson. Veiga et al. [2008] confirmed the occurrence of the transition into a TC using a completely independent data set and methodology. They demonstrated that a robust change in the energetics of the environment facilitated the TT of Catarina. Note that in the present work we have used the satellite‐ derived track from R. Edson. [34] The positive difference between the SST and the air temperature facilitated the development and maturation of Catarina through the production of favorable heat and momentum fluxes over the ocean. Although previous studies could not show this feature in detail as oceanic data was previously unavailable, this study can now confirm a number of interactions related to subsurface ocean heat content that provided Catarina with additional heat sources during its maturation into a TC. A study of the change of the environmental energetics during the abrupt anticyclonic track loop of Catarina and the TT transition, as quantified by Veiga et al. [2008], lacked the inclusion of the important air‐sea interactions described here. [35] Figure 1b shows a satellite image of Catarina in its mature phase after the TT was completed, just before landfall. Figure 1b shows that Catarina held a very marked signature of a mature TC, being completely separated from the remaining of the cold front to the northeast as well as producing a defined band of cirrus around the outer core with very evident spiral bands of moderate convection. Satellite estimates suggest that the convection around the eye was relatively shallow by TC standards, but the estimated precipitation over the ocean was intense with at least 125 mm/d in a large sector of the inner cyclone sustained over several days until landfall. [36] The closest meteorological station with measurements relevant to Catarina is placed in the city of Sideropolis. The station is located at 28°36′S, 49°33′W, at an altitude of 135 m above the sea level, as indicated in Figure 1a (the black star). It gave good data on pressure, wind and precipitation. By applying a standard WMO correction, the minimum MSLP was estimated as 993 hPa. The maximum measured wind speed (sustained) was not corrected to the sea level due to the complex roughness and orographic effects. It gave 40.8 m/s at 0300 LT on

5 of 19

C07002


C07002

Figure 1. (a) Catarina’s track (gray line) as analyzed by R. Edson with data presented by McTaggart‐ Cowan et al. [2006]. Small gray dots over the track denote lack of MSLP data, and circles with color are for MSLP; circle size proportional to wind intensities. Catarina initiation on 19 March shown as a white square with the small gray circle, and landfall on 28 March shown as a white square with large gray circle. State transitions of Catarina recorded as a pink cross (extratropical hybrid), orange cross (TT), and red cross (tropical hurricane) and MET station at Sideropolis recorded as a black star. (b) Color‐enhanced image from GOES IR channel showing the mature phase of Catarina just before landfall on 27 March 1645 UTC. Date labels refer to 1800 UTC each day. Dotted line in Figure 1a is the 1000 m depth contour with topographic map using Smith‐Sandwell 2 min data and Catarina’s image in Figure 1b by courtesy of CPTEC/INPE. 28 March, when the eye had completely entered the land. The pressure and wind time series in the station are hourly sampled and are the only in situ measurements available in Brazil during Catarina’s passage. A factor that has to be

considered is that as observed in Figure 1a the meteorological station is located further inland and entirely missed the eye after it shrank in interaction with the topography. This also helps to explain the fact that the MSLP at Sideropolis

6 of 19

C07002


Table 1. Statistics of SST Data Comparisons in March 2004a Data Set

Correlation Coefficient

RMS

Argo MW‐OI MODAS Reynolds

– 0.87 0.86 0.90

– 0.31 0.32 0.29

Range [22.04 [22.20 [22.07 [21.53

24.17] 24.15] 24.32] 24.11]

a RMS centered difference between each data set at the grid cell of the Argo surfacing, and the Argo near surface temperature (20 m, inside the mixed layer) obtained from Argo profiles, and SST ranges. Range values and RMS are in °C.

was substantially higher than the minimum MSLP estimated over the ocean by the hurricane model. The precipitation inland was variable, with considerable interaction with the topographic features producing areas of shadow and areas of enhanced rainfall with totals of about 150 mm in a matter of hours. Those amounts agree well with the precipitation that was estimated over the ocean.

4. Ocean Conditions Before Catarina [37] It is now well known that the vertical ocean structure preconditioning is a strong determinant of the evolution of hurricane intensities and their wind velocity maxima [Mainelli et al., 2008]. Among the most important preconditions are the initial MLD and MLT fields and thermocline water stratification, and currents which may advect temperature fields adjacent to the hurricane track into the interaction region. These may determine the available energy stored in the ocean for maintaining favorable surface heat fluxes which drive the cyclone. In this sense, the presence of even relatively small WCRs along the surface track of the cyclones is potentially very important, since they store more heat with their deeper MLDs. The negative ocean air feedback is minimized because it is inversely related to MLD [see Mainelli et al., 2008, and references therein]. [38] In terms of the ocean preconditions for tropical cyclone development, the most striking one described during the Catarina storm was the sea surface temperature field. The preliminary description given by Pezza and Simmonds [2005] and McTaggart‐Cowan et al. [2006] showed that the onset and initial track of Catarina had unexpectedly developed over waters below or near the traditional Palmén SST threshold of 26.5°C for hurricane genesis. [39] However, these first studies used low‐resolution SST data sets of 1° × 1°. While Pezza and Simmonds [2005] used data obtained from the ECMWF operational model, McTaggart‐Cowan et al. [2006] used Reynolds version 2 weekly mean data set for the period 21–27 March 2004. Since that time, there has been some debate suggesting that at some point the storm could have crossed over waters warmer than 26°C, thereby significantly influencing somehow in the cyclone formation pattern. To better evaluate these matters, in our present study we used three different higher‐resolution daily gridded SST data, and further in situ SSTs derived from Argo floats, to describe in more detail the SST distribution before and during Catarina, especially along its track. [40] Prior to using any of the data sets for diagnosis, we evaluated the spatial consistency between these SST fields in March 2004 by expanding each daily ensemble (MW‐OI,

C07002

Reynolds, MODAS) into EOF modes to determine their mutual correlation and to estimate their difference fields. For this computation, the higher resolution MODAS was linearly interpolated into a 0.25° × 0.25° grid. The first EOF mode gives us a measure of the ensemble correlation by its maximum total variance, and the best correlated map, while the other EOFs give the residual maps associated with the uncertainties. The results show that the SST data from these different sources were quite consistent: the explained variance of the first EOF ranged from 97.16% to 97.52%, with an average value of 97.40%. The other EOFs explain only 2.6% of the variance (on average). However, we found that the Reynolds SST values seem to overestimate by almost 1° C relative to the other SST data sets at a few grid points. To complete the picture, we have computed the daily root mean square centered differences (RMS) of the three pairs of data sets (not shown). In general the agreement is quite good, with a maximum of 0.58°C between Reynolds or MW‐OI and MODAS. With respect to the daily spatial average, the data sets also show good consistency in March 2004: MW‐ OI presents regional (22°S–33°S; 55°W–30°W) means between 24.68°C and 25.48°C, MODAS between 24.63°C and 25.35° and Reynolds between 24.68°C and 25.30°C. The standard deviations of the time series formed by the averages are 0.23°C for MODAS and MW‐OI and 0.19°C for Reynolds. The Reynolds data is distributed with corresponding mapping errors, and in the off‐shelf open ocean Catarina region the maximum error is 0.25°C. [41] In March 2004, inside the Catarina region, there were only four Argo floats, all placed between 30.5°–32.5°S and 33°–48°W. These floats gave ten good vertical temperature profiles, eight being before Catarina (7, 8, 17, and 18 March), one during (27 March) and one just after the storm (28 March). The near surface temperatures measured by these floats show that SSTs in March 2004 were between 22.04 and 24.17°C. The SSTs obtained from the gridded data sets agree with the near‐surface measurements from these in situ Argo vertical profiles, as shown in Table 1. [42] Figure 2a shows the SST distribution in the region immediately before Catarina was formed on 19 March. It shows a SST pattern containing a north‐south tongue of about 200 km width centered at 44°W and southern limb at 28°S with 26°C. To make this point clear we overlayed the SST field with the future track of Catarina. Two important facts are observed: [43] 1. Except for the SST tongue centered at 44°W, all SSTs were below 26°C south of 26°S. This is true even when we look at the maximum SST during the 19 days before the storm. [44] 2. There is no definite signature of mesoscale ocean features in the vicinity of that track, except for the tongue centered at 44°W. [45] The SST anomaly field over this tongue (not shown) presents a structure with 1°C above the generally lower SSTs relative to the climatology. [46] The main conclusions that can be drawn from the SST analysis are the following. The first one is that the Palmén precondition for hurricane genesis fails in the Catarina case as noted by McTaggart‐Cowan et al. [2006], and we showed this as being independent of the data set chosen. The second one is that no other mesoscale feature besides the tongue centered at 44°W is observed in the SST fields,

7 of 19

C07002

C07002


Figure 2. SST, SSH, D17, and geostrophic currents on 19 March, 1 day before the onset of the Catarina storm, overlaid with the future centroid track of Catarina: (a) Reynolds SST exhibiting only one tongue with 26°C over a background of lower temperatures south of 27°S. (b) The SSH data showing the complex ocean topography featuring four main WCR signatures traversed by the track, with two nearby Argo floats at positions A1 (over WCR‐3) and A2 away from any WCR. (c) Same as Figure 2a but for the D17 subsurface isotherm. (d) The mesoscale absolute geostrophic currents where small vectors refer to less than 10 cm/s (see text for more details). Red squares denote daily storm center positions, and dotted line is the 1000 m depth isobath. even in those with appropriate mesoscale spatial resolution like MODAS. These partial results suggest that intensification cannot be ascribed to parameterizations based on SST alone. [47] When we look at the distribution of SSH the situation is radically different as can be observed in Figure 2b. Now we can see clear eddy structures present under the future Catarina track. The most notable are three WCRs in a ridge‐ like distribution located around 27–31°S and 40–45°W (WCR‐1, WCR‐3 and WCR‐4 in Figure 2b), and an eastern WCR centered at 30°S–36°W (WCR‐2). These 200–300 km features were present almost without change since the end of February 2004. As observed on time‐longitude plots (not shown) their westward propagation speed is very slow, with a maximum of 3.6 km/d. We will see in the next section that the only noticeable changes in these structures on daily time scales in March 2004 were those driven by the passage of Catarina. [48] Figure 2c shows the map of altimeter‐derived 17°C isotherm depth on 19 March, from which a measure of the relative heat content of the upper ocean may be derived. From the D17 map it is estimated that WCR‐1 to 4 have D17s deeper by more than 70 m than the background D17 of 100 m. We may notice also that D17 is much shallower near the shelf slope, where it may be less than 30 m. Table 2 shows the D17 and SST values for each WCR on 19 March.

[49] Although there were four Argo floats in the Catarina region, only two gave profiles immediately before, during and at the end of Catarina. We call these two surface positions of A1 (float Id 3900191) and A2 (float Id 3900190), A1 being located at 30.9°S–42°W and A2 at 31.5°S–38°W. According to Figures 2b and 2c the float A1 is on the rim of WCR‐3 on 19 March. It surfaced on 7, 17 and 27 March (1200 UTC). Float A2 was outside any eddy, having surfaced on 8, 18 and 28 March (1200 UTC). Table 3 shows various parameters retrieved from these Argo floats. We can see that before Catarina the D17 depth difference between A2 and A1 was then 40.8 m around 7 March, and 40.1 m around 17 March, showing that the D17 ocean structure change was only 70 cm in about 10 days. It should be noticed that A1 was off a warm eddy core, where the depth of the D17 isotherm and the MLD should be much deeper. Table 2. SST Data from Reynolds, MODAS, MW‐OI, and D17 During 19 March at the Positions of WCR‐1 to WCR‐4 SST (°C) Latitude Longitude Reynolds MODAS MW‐OI D17 (m) WCR‐1 WCR‐2 WCR‐3 WCR‐4

8 of 19

27.1°S 29.9°S 29.9°S 29.8°S

44.0°W 35.6°W 41.9°W 44.7°W

26.50 23.33 24.32 23.89

25.86 23.64 24.26 24.19

26.25 24.00 24.90 24.00

164.1 163.0 152.2 161.0

C07002

C07002


Table 3. Information Retrieved From the Argo Profiles at Positions A1 and A2 at Four Surfacing Datesa A1 (30.9°S, 42°W)

MLT (°C) MLS D17 (m) MLD (m) QML (kJ/cm2) QD17 (kJ/cm2)

A2 (31.5°S, 38°W)

7 Mar

17 Mar

27 Mar

6 Apr

8 Mar

18 Mar

28 Mar

7 Apr

24.16 36.32 158.48 43.07 9.88 226.76

23.59 36.33 157.35 56.16 6.57 232.96

22.83 36.00 127.53 57.76 21.57 184.90

23.01 35.70 116.67 33.65 2.52 158.18

23.24 35.63 117.68 37.11 2.29 169.33

23.24 35.80 117.35 41.51 0.05 163.80

23.33 35.92 112.97 34.84 3.69 147.40

23.75 35.85 118.41 34.93 2.79 171.91

a Position A1 is over WCR‐3 and position A2 is just off WCR‐2. Ranges of mixed layer temperatures (MLT), salinities (MLS), and depths (MLD) and D17 and estimated heat content in the mixed layer (QML) and in the total water column down to D17 (QD17).

[50] From the absolute sea surface height field showed in Figure 2b we have derived the absolute geostrophic current field for 19 March (Figure 2d). It can be seen that the mesoscale circulation shows a system of currents and eddies with 30 cm/s intensities in the open ocean, and southward flows of up to 40 cm/s well inside the continental shelf. Figure 2b is included here because the Catarina winds, as will be shown in the next section, sometimes are in opposition to these currents, which might increase the spray and possibly more efficient ocean‐air transfers. [51] The ocean structures just described may be interpreted as representing the stable initial ocean conditions in the time scale of at least 10 days before the onset of Catarina, with the vertical structure below the mixed layer well represented by the altimeter‐derived D17 topography. It is interesting to note that on the future track of the Catarina storm, the presence of the warm meridional 26°C tongue at 44°W in Figure 2a is above the deep WCR‐1 feature with an estimated depth of 164.1 m (Figure 2c), that could possibly be a trigger for the organization of the wind field into a cyclone.

5. SST, WCRs, Winds, and Waves During Catarina [52] The regional QuikSCAT OSVWs on the morning of 20 March exhibit a weak wind background of 6–8 m/s intensity, a southerly patch centered at 48°W between 26°S and 29°S, a stronger easterly wind patch around 27–28°S, 43–46°W which merges with the southerly wind, and an easterly jet centered at 29°S, 43–49°W (Figure 3a). These wind patches attain 12–14 m/s. The lower QuikSCAT wind intensity is confirmed by a coincident JASON‐1 wind speed data that also presents a 14 m/s wind (not shown). However, as can be seen in Figure 3a the OSVWs do not display any cyclone signature yet. On the evening of 20 March the surface wind field organizes into a cyclonic circulation, centered at 26.5°S, 44°W. Figure 3b shows that the south winds strengthened over the 200 km width warm 26°C ocean SST tongue and also over the deep WCR‐1, where the D17 maximum depth is 170.5 m. Higher wind intensities are found in the JPL‐2BHR product with Umax of 28.4 m/s, while in the other the winds are between 21 and 25 m/s. The area covered by Umax in JPL‐2BHR is very small and if we disregard this value, the JPL‐2BHR speeds agree with other QuikSCAT products (24–26 m/s). These QuikSCAT winds are larger than the estimates given by McTaggart‐Cowan et al. [2006], which shows speeds below 18 m/s between 23 and 20 March. In Figure 3 we use SSH(t − 1) and SST(t − 1), observed 1 day before the wind

velocity U(t), where t refers to the day of the QuikSCAT wind data, to illustrate the sea state just prior to the onset of the local winds. [53] One standard and useful oceanographic measure of the direct effect of wind stresses on upper layer ocean structure is the Ekman pumping (vertical) velocity We, which is forced by the spatially inhomogeneous wind stress field [Gill, 1982]. If the surface (Ekman) wind‐forced horizontal water transport is divergent, water from below must replace the displaced surface water, and a measure of the rate at which this water is upwelled or downwelled is given by We = r−1 ^z · r × (f ), where t is the wind stress, r is the water density, f the Coriolis parameter and ^z is the unit vector in the upward direction. If the water pumped up by the horizontally moving cyclonic wind has the same temperature as on the surface, which is the case if the MLD is large enough in the presence of a WCR, the surface cooling by subsurface water upwelling will be remain small. The moving cyclone with a speed that is usually larger than the baroclinic wave speed excites an inertial wave‐driven wake, with a negative SST anomaly signature, occurring to the left of the cyclone track in the southern hemisphere [Price, 1981; Gill, 1982]. If the cyclone displacement speed is smaller than the baroclinic wave speed, the ocean response is local, and dominated by Ekman pumping. [54] Based on the QuikSCAT wind stresses, we have computed the We fields. Figure 3c exhibits the RSS‐L3 We map for the evening of 20 March where positive values indicate upwelling and negative ones downwelling. A strong upward Ekman pumping velocity We greater than 30 m/d maxima over the 26°C SST tongue/WCR‐1 complex can be seen. Although the same spatial pattern is observed in the We maps derived from the others OSVWs, the intensities vary considerably between the products with 25–27 Km and 12.5 Km. In the JPL‐2BHR, the We attains 60 m/d. [55] On 20 March, we also observe a sequence of elongated SSTA patches of positive 1.0–1.5°C over a 1.0– 1.5 negative SSTA field, organized along a meandering axis to the southwest (Figure 3d). This is due to a observed open ocean meander of the Brazil Current, intensified by the directly wind‐forced Ekman transport by the 14 m/s easterlies of the morning of 20 March. The altimeter‐derived geotrophic circulation exhibits a patch of southward 60 cm/s current exactly at 27°S and 45°W, with the local intensification being due to the strong WCR‐1. [56] Therefore, based on these results from the high resolution profiles of ocean currents, QuikSCAT winds, Ekman pumping speed, SST and SSH, it appears that even the extratropical phase of Catarina received significant contribution

9 of 19

C07002


C07002

Figure 3. The 20 March QuikSCAT JPL‐2BHR wind vectors over SSH (color) in the (a) morning and (b) evening. Wind direction/intensity vectors are colored. (c) The evening RSS‐L3 wind vectors over the respective Ekman pumping speed We (in color) with the 26°C SST isotherm (black contour line). (d) The sea surface temperature anomalies (SSTA) (color), SST isotherms (thick black lines), the Catarina track (gray line), and the SSH contours (red lines) exhibiting the WCRs signatures. SST and SSTA are from Reynolds. Note that SSH and SST is from 1 day earlier. Blank wind areas due to no data (off swath). Dotted line for the 1000 m isobath. from the interaction with the ocean. This complements and further expands the earlier views of the classical baroclinic development that Catarina had apparently undergone on the 20 March, as well as the atmospheric blocking participation during the tropical transition [Pezza and Simmonds, 2005; McTaggart‐Cowan et al., 2006]. We speculate that the western limb (northward) winds in 20 March could have been organized by the presence of the opposed local

strong southward current caused by the combined action of the WCR current with the Brazil Current. This situation is favorable to intensified air‐sea enthalpy transfers by enhanced sea spray, since the surface air temperature at 2 m was at least 2°C below the SST, and the 925 mb air temperature was around 14°C. There is a great deal of heat stored in WCR‐1, and the curl of the cyclonic wind stress forces upwelling of subsurface water of the same temper-

10 of 19

C07002


ature as SST, which blocks a possible weakening of the wind by the negative wind‐SST feedback. Both the local SST and the available heat content from the deep WCR‐1 seem to have influenced the organization and maturation of Catarina into a hybrid storm, facilitating its marked tropical transition into a hurricane. [57] On 21 March the storm progresses to the southeast from 27.5°S, 42°W to 30°S, 39.5°W (not shown). Its center becomes better defined and the west limb wind attains 20– 24 m/s, except in the JPL‐2BHR (26 m/s). This displacement takes Catarina, which is then over a 25°C waters, to a region of lower SST (24°C). Near the western cyclone limb, it passes near WCR‐3, centered at 29.5°S, 42°W, which has the D17 depth of 160 m at its center. We note that the Ekman upwelling tongue, first with a N‐S orientation between 26 and 30°S during the morning pass, with 30– 35 m/d and higher in the JPL‐2BHR, rotates into a zonal feature spanning a region without WCRs, with a smaller We intensity of 15–17 m/d (not shown). [58] Catarina becomes a little disorganized while it progresses southeastward on the morning of 22 March, with its center away from any WCR. This situation changes on the afternoon pass of this day, just when it encounters the largest and equally deep WCR, WCR‐2, at 30°S, 36°W, with D17 being 160 m at its center, and more than 400 km in diameter. The cyclonic wind becomes reorganized into a circularly symmetric closed velocity contours, with speeds of 22 to 24 m/s (Figure 4a). The upwelling cell becomes concentrated at 31°S, 37°W, with an increased strength of 40–50 m/d in the RSS products (Figure 4c) and 70 m/d in JPL‐2B and 90 m/d in JPL‐2BHR (not shown). The values obtained in the JPL swath products are higher since they are not smoothed out as in the level L3 products. It is at this time (the evening of 22 March) that the cyclone track abruptly changes direction northwards, but still maintaining itself over WCR‐2. On 22 March there is an ENVISAT altimeter pass over the Catarina track around 1200 UTC, at 31°S, 39°W. The collinear data gives a wind speed of 12– 14 m/s on the lee of Catarina while it is still moving eastwards, and waves with a SWH of 4 m (Figure 4e). These waves are relatively high because Catarina had passed there several hours before this time, forcing this wave wake. [59] The upwelling cell gets a little attenuated on 23 March with We around 20 m/d (Figure 4d). This is when Pezza and Simmonds [2005] and McTaggart‐Cowan et al. [2006] record a transition of Catarina from the extratropical into a hybrid cyclone state. As shown in Figure 4b, the QuikSCAT JPL‐2BHR winds on the western forward flank of the cyclone is more intense (14 to 18 m/s) than its surroundings, and there is a noticeable spreading of the higher wind patch. Although in this day the infrared satellite image [McTaggart‐Cowan et al., 2006, Figure 4a] does not exhibit a Catarina eye, in JPL‐2BHR OSWV we can clearly see it centered at 29.6°, 37.5°W with low speeds of 4 m/s and diameter around 40 Km (Figure 4b). Catarina is now almost completely over WCR‐2, and the SST in this area is around 23.5°C. There are two passes of the JASON‐1 satellite (the crossover point is over the Catarina track), one around 0850 UTC and the other around 1850 UTC, and both record the same wind speed as seen from QuikSCAT products below the Catarina track, but with a intensified SWH of 4.5–5 m (Figure 4f).

C07002

[60] We skip the description of 24 March due to a QuikSCAT data gap for Catarina center, but its position at this time is not close to any WCR. No data for the Catarina center is available, but the data on the evening of 24 March suggests that the high wind patch mentioned above attenuated into 12 m/s as it crosses a low ocean dynamic topography region, and the leading edge of the cyclone presents the northward wind region between 41 and 43°W centered at 29°S (figure not shown). [61] On 25 March we have a complete surface view of Catarina based on QuikSCAT data sets (not shown). The high wind patch is now reduced in area but intensified into 20–24 m/s and it is more concentrated at 41°W. The speed maxima are in the south and western limbs of the cyclone, near the edge of WCR‐3. This situation does not change much in the afternoon pass, except that the eye is better defined. It is clearly seen in the JPL‐2BHR, centered at 28.75°S and 42.25°W, with speeds around 13–14 m/s surrounded by 23–29 m/s velocities. These QuikSCAT speeds are in agreement with McTaggart‐Cowan et al. [2006] estimates (20–28 m/s). The available collinear ENVISAT data at this date does not coincide in the timing and position of the highest winds of Catarina. It gives wind speed maximum of 15 m/s and SWH of 3.5 m. [62] On the morning of 26 March Catarina suddenly intensifies, with maximum surface winds forming a westward patch of 27–30 m/s on its southern limb at 29°S, 43.5°W, and reaching 30.7 m/s in JPL‐2BHR, (Figure 5a). The wind field symmetrically spans the large deep thermocline region of about 400 km width with D17 being more than 160 m depth, comprising the cores of WCR‐1, WCR‐3 and WCR‐4. In JPL‐2BHR, the eye is clearly seen at 28.7°S, 43.8°W, with core speeds of 12–15 m/s, and diameter around 41 Km. The circular upwelling cell has Ekman vertical velocity of more than 80 m/d on the eye (Figure 5c), and more than 140 m/d in JPL‐2BHR. In the afternoon pass, the maximum wind speed attains 34.42 m/s in JPL‐2BHR (Figure 5b), and lower in the other QuikSCAT OSVWs. According to the JPL‐2BHR OSVW, Catarina reaches category 1 on Saffir‐Simpson hurricane scale on the evening of 26 March. The We is more than 100 m/d in JPL‐2BHR and 65 m/d in the other products, and slightly displaced to the west at 44.5°W (Figure 5d). The eye is now at 29°S, 44.5°W. [63] It is considered that Catarina attains a category 1 hurricane status on 26 March and maintain this status during 27 March [Pezza and Simmonds, 2005; McTaggart‐Cowan et al., 2006]. Although there is no QuikSCAT pass on the morning of 27 March, on the evening of 27 March the QuikSCAT OSVWs show a small maximum westward wind patch (50 km width) of 30–32.5 m/s at 29.5–30°S and 48°W, already over the continental shelf (Figure 6e). In the RSS‐2B product, the intensity field exhibits a maximum wind speed of 34.67 m/s. Unfortunately, on the evening of 27 March, all QuikSCAT products show cross‐track vector winds that are a known sign of rain contamination in the Ku band. For that reason, since this work deals only with observations, computation of We is not presented. There is a coincident crossing of Catarina by the descending pass of the radar signals of the JASON‐1 altimeter on 0830 UTC (morning), where the wind speed is around 27 m/s, and the SWH is 5 m, while in the afternoon both winds and waves have decreased in amplitude (Figures 6a–6d). Therefore, the

11 of 19

C07002


Figure 4. (a) QuikSCAT JPL‐2BHR wind vectors over SSH (color) for the evening of 22 March. (b) Same as Figure 4a but for the morning of 23 March. (c) We (color) for the evening of 22 March computed from RSS‐L3, overlayed with wind vectors and SST from Reynolds (gray lines). (d) Same as Figure 4c but for the morning of 23 March. (e) Collinear SWH for the evening of 22 March and Catarina track (gray line). Blue dot marks the Catarina daily position, and SSH and SST from 1 day earlier. Dotted line is the 1000 m isobath. (f) Same as Figure 4e but for the morning of 23 March.

12 of 19

C07002

C07002


C07002

Figure 5. (a) QuikSCAT JPL‐2BHR wind vectors for the morning of 26 March overlayed to SSH and the 26°C SST contour from Reynolds, where vectors with U > 10 m/s are plotted. (b) Same as Figure 5a but for the evening of 26 March. (c) We (color) computed from RSS‐L3 and corresponding wind vectors for the morning of 26 March. The large sudden intensification of Catarina concentrated in the south limb due to its swift westward displacement, and corresponding intense We patch. SSH and SST from 1 day earlier, and dotted line for the 1000 m isobath. (d) Same as Figure 5c but for the evening of 26 March.

maximum wind intensity measured with the altimeter along‐ track data in the morning is smaller than that of the QuikSCAT on the afternoon, where it attains 30–32 m/s (see Figure 6e). It is important to mention here that SST is still not larger than 25°C in all regions south of 28°S. Figure 6f shows the absolute geostrophic currents observed on that day, and we call attention to the cold eddy centered at 31°S, 47°W, west of WCR‐4. These features cause eastward currents on 30°S around 47°W, while over the shelf there is a southward maximum shelf current (60 cm/s) on 29–30°S. [64] We can therefore suggest that throughout the day the surface wind maximum was around 30–34 m/s, and the waves attained a SWH of 5 m, with opposing Ekman and geostrophic currents. This situation might make the ocean‐to‐air fluxes be more efficient due to the fact that wind‐sea waves propagate against geostrophic currents, both reducing their phase speeds and increasing their amplitude [Holthuijsen and

Tolman, 1991], resulting (among other effects) in more efficient spray generation and spray‐mediated fluxes. The wind data analysis suggests that the high winds were on a small patch on the south limb of the cyclone, more influenced by its accelerated westward displacement speed than pertaining to its circularly symmetric cyclonic wind. [65] Figures 6a–6d show that on 27 March Catarina was traversed in the morning by a descending JASON‐1 track, and in the afternoon its lee was traversed by an ascending track, near a so‐called track crossover point. This crossover point (denoted here as P2) is important, because in the P2 position we can compare how the SSHA, winds and waves were in the morning, and 10 hours later in the afternoon. Table 4 gives the SSHA, the wind and the SWH at P2 for both passes. In Table 4 the track points 1 second prior to P2 (P1) and 1 second later than P2 (P3) are also included. One second difference in the JASON‐1 track corresponds to a

13 of 19

C07002


Figure 6. (a) Collinear passes of JASON‐1 at 0830 (ascending) 27 March, with SWH. (b) Same as Figure 6a but for 27 March 1830 (descending). (c) Same as Figure 6a but for wind speed. (d) Same as Figure 6b but for wind speed. (e) The wind speed from JPL‐2BHR for 27 March showing a clear eye. (f) The geostrophic currents for 27 March where small vectors refer to less than 10 cm/s. Figures 6a–6d and 6f overlayed with Catarina track (gray/white line). Blue dot marking the Catarina position. In Figures 6c and 6d, red lines for 1 day earlier 25°C SST contour from Reynolds. Dotted line for the 1000 m isobath, and time in UTC. 14 of 19

C07002

C07002

Table 4. Unfiltered Collinear Data Near the Crossover Point on 17 and 27 March, at 45.38°W, 29.66°S of JASON‐1a P1 Passes

P2 Passes

P3 Passes

Morning Evening Morning Evening Morning Evening 27 Mar SSHA (cm) SWH (m) Wind (m/s) 17 Mar SSHA (cm) SWH (m) Wind (m/s)

C07002


17.38 5.31 21.36

5.98 2.81 10.04

32.99 4.79 20.78

1.12 3.00 9.90

13.56 4.52 20.14

−1.45 2.87 9.82

4.09 3.86 10.05

4.64 4.12 8.95

1.47 4.00 9.69

4.00 4.18 8.83

0.44 4.13 9.73

4.68 4.07 8.75

a P2 is the crossover point. The passes of 17 March (27 March) were on 1030 (0833) UTC descending, 2031 (1829) UTC ascending. Data from point P1 1 s before (45.40°W 29.62°S descending, 45.34°W 29.71°S ascending) and P3 1 s after (45.26°W 29.90°S descending, 45.29°W 29.61°S ascending).

7 km distance. Table 4 also includes data taken on 17 March for comparison. [66] The impact of Catarina in P2 was evaluated computing the differences between the SSHAs observed in the evening and morning passes (dSSHA = SSHAPM − SSHAAM). From the dSSHA, the respective D17 differences (dD17) can be estimate with the help of the regression parameters given in equation (2), as dD17(m) = 2.55 * dSSHA(cm). From Table 4, on 27 March the P2 dSSHA is −31.87 cm which corresponds to a upward change in D17 of 81.27 m. Such jump does not occur on 17 March, 10 days before Catarina cross the P2, where dSSHA was +2.53 cm equivalent to a D17 downward change of 6.45 m. Table 4 also shows that on 27 March in 10 hours the wind in the region around P1, P2 and P3 weakened from 20–21 m/s to 9–10 m/s, and the SWH from 4.8–5.3 m to 2.8–3 m. On 17 March the winds were around 9 m/s in both passes and the SWH 4 m. We

Figure 7. Evolution of Catarina maximum wind speeds U max retrieved by different QuikSCAT products: JPL‐ 2BHR, JPL‐2B, JPL‐L3, RSS‐2B, and RSS‐L3. There were no data on 24 March and the morning of 27 March. The circles mark the data for each time series, with JPL‐2BHR swath 12.5 km data giving the daily maximum speeds. See text for further details.

Table 5. RMS Centered Differences of Catarina Maximum Wind Speeds Umax Between Different OSVW Productsa OSVW

RSS‐2B

RSS‐L3

JPL‐2B

JPL‐2BHR JPL‐2B JPL‐L3 RSS‐L3

3.56 2.82 1.95 1.94

2.99 2.31 2.17 –

1.93 – 0.82 2.31

a

See text for more details. All values are in m/s.

suggest that on 17 March these SWH refer to the influence of the winds related to the cold front prior to the genesis of Catarina [Pezza and Simmonds, 2005]. [67] The QuikSCAT data on the morning of 28 March (not shown) only exhibits a northeasterly wind maximum of 14 m/s adjacent to the coast, and no cyclonic wind signature, attesting that the Catarina winds have vanished from the ocean.

6. Analysis and Surface‐Subsurface Interactions [68] Figure 7 shows the time series of Catarina maximum wind speeds Umax recorded by different OSVW products. It is clear that JPL‐2BHR retrieves higher speeds, and two QuikSCAT products (JPL‐2BHR and RSS‐2B) show that Catarina have reached hurricane category 1 status. JPL‐L3 and JPL‐2B have the lowest speeds, and between 25 and 28 March its values are identical. The RMS between these OSVW series are recorded in the Table 5. The larger differences are found between JPL‐2BHR and RSS‐2B. Although JPL‐2BHR have retrieved the most intense winds, it is noisier, and if derivatives are needed to compute a diagnostic variable, e.g., We, some prior smoothing is necessary. [69] Analyzing the data on the Catarina winds and the wind curl forcing of the upwelling cell, we noticed that both the Umax and We amplitudes occur in patches of relatively small areas. To estimate the strength of the air‐sea interaction forced by the cyclonic winds, we computed the effective area of the upwelling cells (patches) by recording the positions of the maximum We and the area Sw enclosed by the isoline of 50% of the maximum amplitude. Note that this computation was performed only with the level 3 products because JPL‐2BHR is too noisy. [70] Figure 8a shows the time series of daily Umax, maximum We, and Sw during the Catarina. It can be seen that there is a slight anticorrelation between Umax or We and Sw which seems to suggest a patch area mean vorticity conservation, one example being the sudden increase of We on the evening of 22 March and the sharp decrease of its patch area. Figure 8b shows the evolution of D17 at the center of each of the four WCRs by a daily time series starting on 15 March, before the onset of Catarina, until 31 March. We can observe that on each of the four WCR centers the onset of sharp decreases in D17 coincide with the day of the Catarina passage over them, suggesting that the abrupt changes are due to the cyclone forcing. [71] A good basic description of the impact of a moving cyclonic storm over the ocean are given by Gill [1982, and references therein; see also Oey et al. 2007]. A key dimensionless parameter in this behavior is U/C1, where U is the progression speed of the cyclone and C1 is the first

15 of 19

C07002


C07002

Figure 8. Evolution and impacts of Catarina. (a) Daily maximum wind speeds Umax, area of We patch, and maximum daily vertical upwelling speed Wemax, based on the RSS‐L3 QuikSCAT product. (b) Evolution of D17 at the four WCRs, with the date of the Catarina pass over them marked by a pink square dot. (c) Map of SST differences between 28 and 19 March superimposed with the Catarina track (gray line) and 1000 m isobath (dotted line). (d) Same as Figure 8c but for D17 differences. In Figures 8c and 8d, the expected inertial wave wake signature can be observed on the surface and subsurface fields, respectively, but the leftward bias is better defined with the D17 data. Positive D17 differences means that D17 upwelled. The wake does not occur east of 40°W, when U/C1 < 1. baroclinic mode wave speed of the ocean. The region where Catarina occurred is characterized by C1 ≈ 2.6 m/s [Chelton et al., 1998], while it is easy to estimate from the daily positions and times of the storm center that 1.1 < U < 6.6 m/s. On most of the track U/C1 > 1, except near the region where Catarina makes the sharp turning loop, when U/C1 < 1, where the minimum values of U occur. In the former case, a negative SST perturbation due to the cyclone wake oscillations should be expected, with the well‐known leftward bias relative to the path of storms in the Southern Hemisphere [Price, 1981]. This should happen even over WCRs but in this case the negative SST perturbation amplitude is expected to be smaller. Correspondingly, SSH negative perturbations indicating upwelling of subsurface isopycnals, isotherms, and the mixed layer base should be also observable due to positive We. [72] One way to evaluate the rapid impacts of Catarina on the ocean is by computing differences from initial and final ocean state variables, as done in the detailed studies of the impacts of hurricanes over the ocean available in the literature [e.g., Oey et al., 2007, and references therein].

Figure 8c shows the dSST map, defined here as the change of SSTs between 19 and 28 March and in the Table 6 we can see the respective values over the WCRs. The time difference in Figure 8c exhibits the long‐term signature of the inertial wave wake, which has a lifetime of more than 2 weeks [Price, 1981]. We notice the general prevalence of negative dSST, the maximum amplitudes being around 1.5°C, with a leftward bias. In the vicinity of the Catarina’s loop the dSST is slightly positive (less than +0.5°C). On the continental slope, where the slope bottom cooler water is forced to upwell, the SST negative difference presents a blob with the maximum amplitude observed during the Catarina lifetime of 1.5°C at 30°S. Table 6 also shows that over the WCR‐1, WCR‐3 and WCR‐4, corresponding to the fast displacement speed of Catarina, the amplitude of the mean cooling is between −1.2°C and −0.2°C, while for the smaller speeds over WCR‐2 dSST is slightly positive if we look at Reynolds or MW‐OI data sets. [73] The reason for the dSST to be slightly positive in the loop region, where U/C1 < 1, is that the induced We is expected to be localized (not a stationary wave [see Gill,

16 of 19

C07002


Table 6. SST Differences From Reynolds, MODAS, MW‐OI, and D17 Differences, Between 28 and 19 March at the Positions of WCR‐1 to WCR‐4a dSST (°C) Latitude Longitude Reynolds MODAS MW‐OI dD17 (m) WCR‐1 WCR‐2 WCR‐3 WCR‐4

27.1°S 29.9°S 29.9°S 29.8°S

44.0°W 35.6°W 41.9°W 44.7°W

−0.66 +0.55 −0.18 −0.50

−0.75 −0.11 −0.86 −0.54

−0.36 +0.21 −0.26 −1.20

+17.0 +18.1 +22.1 +20.0

a Here dSST is SST differences and dD17 is D17 differences. Positive values of dD17 indicate isotherm uplift and negative ones of dSST a cooling.

1982]). This localized Ekman pumping perturbation should decay due to turbulence, and the local weak wind driving of less than 5 m/s which is subsequently sustained ceases to be important relative to other thermal forcings. The computation of dSST between different dates support this reasoning (not shown). On 23 March, when Catarina is closing the loop, the dSST (23, 19) in the region is around −1.5°C with a leftward bias. On 24 March dSST (24, 19) is −0.9°C and on 26 March, when Catarina is 400 Km away from the loop region, the dSST (26, 19) is close to zero. [74] By comparing this generally small observed negative SST feedback between the ocean and Catarina with the corresponding amplitude seen in almost all other tropical cyclones, we can interpret that such small values were due to the presence of the deep WCR‐dominated thermocline topography. [75] Before Catarina a small cooling trend of 1°C is seen on dSST between 13 and 19 March (not shown). However, no such trend was observed on dSST between 19 and 28 March in any of the data sets, but Reynolds SST on 29 March is slighly warmer than the other two data sets, thus having an impact on the difference dSST between 19 and 29 March, which would suggest that there was a slight warming trend in SST during the Catarina period. If the data from the other data sets are used to get dSST between 19 and 29 March, no warming trend is observed. [76] Figure 8d illustrates the changes in D17 in the same 8 day period as the dSST, with the predictable leftward biased mean upward D17 difference. Table 6 summarizes the results for dD17 between 19 and 28 March on the four WCRs, showing that on average the upward jump was 20 m. It is important to remember that the D17/SSH fields are highly attenuated as consequence of the data processing employed, mainly from the collinear filtering made prior to and from the objective mapping. Therefore, the magnitude of the upward jumps are probably underestimated in our analysis. This can be confirmed by the JASON‐1 collinear cross‐over point (P2) on 27 March, where the d D17 in 10 hours was about 81 m in the raw data and 44 m in the filtered. [77] The data from the two Argo floats nearby shows the same short‐term trends, consistent with the abrupt jumps and with Figure 8b. Table 3 shows data obtained in four different surfacing dates for floats A1 (A2), from 7 March (8 March) to 4 April (4 April) at 10 day intervals. They provide data on MLT, MLS, MLD, D17 and the heat contents between the surface and MLD, and surface and D17. At

C07002

A1 (rim of WCR3), the MLT shows only one noticeable small change on 27 March (around 1°C), which is consistent with the satellite‐derived data and MLS data shows a small freshening due to rainfall at the two last dates. It is interesting to note that in this float the D17 presents a large upwelling, around 30 m, on 27 March relative to the two previous dates and the MLD is almost unchanged. At A2, which is inside the Catarina loop region and outside WCR2, both D17 and MLD uplift only negligibly (5m). The reason for this is the same as explained above for the dSST case: the forced subsurface oscillation also dies out, and we speculate that the absence of Ekman pumping and subsequent diffusion of WCR2 after the passage of Catarina may be responsible for this effect. [78] At A1, although the MLD and MLT are comparable in both 27 and 17 March, the salinity vertical profiles are completely different in the first 100 m depth (not shown). The freshening of the water column at A1 goes down to at least 250 m depth, both due to precipitation, but also due to flow convergence of thermocline water due to Ekman suction below the mixed layer. This fact has a considerable impact in the heat content of the water column above the MLD since in the Q computation (see equation (1)) the cp is also salinity and depth dependent and it is inside the integral. At A1, the MLS falls from 36.33 to 36 causing an abrupt increase in cp and therefore a 15 kJ/cm2 increase in QML between the two dates. In spite of this small heat content increase in the mixed layer, the QD17 diminished of 48 kJ/cm2 in the same period.

7. Conclusions [79] As shown earlier, the Catarina genesis (20 March), its sharp changes in surface structure (23 March) and its sudden intensification (26–27 March) are clearly related with its higher speed winds crossing over the deeper D17 isotherm mesoscale structures related to the WCRs. Its vigorous interactions with the water column goes down to 170 m depth, which is a result consistent with present‐day knowledge of air‐sea interactions under tropical cyclones up to category 5 [e.g., Shay et al., 2000; Hong et al., 2000; Lin et al., 2005; Oey et al., 2007; Mainelli et al., 2008]. [80] According to McTaggart‐Cowan et al. [2006] and Pereira Filho et al. [2010], just a few hours before the landfall, Catarina reached on 28 March 0600 UTC the category 2 (43–49 m/s) status. The track position indicates that Catarina was almost over the continent when it possibly achieved this status. According to these authors, after that Catarina quickly lost intensity, from 45 m/s to 23 m/s. However, neither QuikSCAT nor altimeter wind speeds show that Catarina had reached category 2 over the ocean, and the Umax is 34.5 m/s in the JPL‐2BHR and 34.6 m/s in the RSS‐2B. Although it might happen that both the altimeter radar and the scatterometer pulses might be attenuated by rain, Catarina was characterized by generally low rainfall intensities [Pereira Filho et al., 2010] in relation to bad impacts on the scatterometer wind retrievals. We note however that in the 27 March evening QuikSCAT data, the rainfall contamination was detected by the loss of circular symmetry in the cyclonic wind field, which suggests the presence of a cross‐satellite track wind bias. Examination of the four possible wind direction ambiguity solutions in the

17 of 19

C07002


level 2 data did not solve the problem, but in any case such a discussion is beyond the scope of the present work. As to the winds measured on land by the Sideropolis station (hourly data), with a local maximum of 40 m/s, no comparison with the ocean wind data is possible, without reduction of the wind data to sea level, and taking into account a analysis of the turbulence fluctuations, necessary to estimate the land roughness length, and the orographic effects necessary to do the reduction [Powell et al., 1996; Beatty et al., 2004]. [81] Nevertheless, it is curious that Catarina having reached only category 1 over the ocean, it was able to cause abrupt impacts of the same order of magnitude as those observed in other known tropical cyclones of higher categories. For example, Oey et al. [2007] show in their Figure 3b that We of hurricane Wilma (category 5) attained 106.2 m/d in its main upwelling patch on 24 October 2005, with an isopycnal uplift of about 60 m, and a SSH drop of 25 cm. Catarina produced the surprising We of 100 m/d on the morning of 26 March, D17 uplift between 44 and 81 m, and SSH drops between 15 and 32 cm. The thorough study of hurricane Opal (category 4) by Hong et al. [2000] shows upwelling cells of subsurface isotherms of 70 m, while e.g., Shay et al. [2000] show that SSH drops were about 20 cm in the Opal WCR region. [82] The main goal of this work was to establish a rigorous oceanographic‐based data analysis of the Catarina hurricane. The results suggest that dynamical ocean processes need to be accounted for in traditional analysis and prediction models. It is believed that the Catarina transition was primarily initiated by the atmospheric blocking [Pezza and Simmonds, 2005; McTaggart‐Cowan et al., 2006], and the subsequent intensification was archieved via the fundamental ocean interactions here described, although in the present work we do not attempt any modeling of the physics of the air‐sea interactions. [83] Our results reinforce the comment of Hong et al. [2000] that perhaps hurricane forecast models with prescribed SSTs fail to predict the observed hurricane intensification, if such models are not coupled to a high resolution ocean circulation model in the simulation. This work may be considered as a prerequisite to a future analysis of the air‐ sea interaction physics during Catarina, and the formulation of a realistic storm prediction model for the southwest Atlantic ocean including both atmospheric and oceanic processes. [84] Acknowledgments. We are indebted to Manuel Gan and Noel Davidson for interesting discussions related to our first attempts to understand the influences of ocean processes on Catarina, and to several organizations for making available their data on the web, specifically: RSS, Inc. for SST and QuikSCAT data sets and NOAA‐CDC for Reynolds SST data and the US Navy NRLSSC for the MODAS SST; the French Coriolis Project for the Argo float global data set; JPL for wind scatterometer data, and US Navy NRLSSC for the along‐track altimeter data from the Real Time Ocean Environment archives. We also acknowledge the detailed critique of two anonymous reviewers which offered good corrections and suggestions to make the paper more easily readable.

References Barron, C. N., and A. B. Kara (2006), Satellite‐based daily SSTs over the global ocean, Geophys. Res. Lett., 33, L15603, doi:10.1029/ 2006GL026356.

C07002

Beatty, K., C. Miller, and A. Boissonnade (2004), Equivalent 10‐meter wind speeds in Hurricane Fabian, paper presented at 26th Conference on Hurricanes and Tropical Meteorology, Am. Meteorol. Soc., Miami, Fla., 3–7 May. Beven, J. L., L. A. Avila, E. S. Blake, D. P. Brown, J. L. Franklin, R. D. Knabb, R. J. Pasch, J. R. Rhome, and S. R. Stewart (2008), Atlantic hurricane season of 2005, Mon. Weather Rev., 136, 1109–1173, doi:10.1175/2007MWR2074.1. Bonatti, J. P., V. B. Rao, and P. L. Silva Dias (2006), On the westward propagation of the Catarina storm, paper presented at 8th International Conference on Southern Hemisphere Meteorology and Oceanography, Am. Meteorol. Soc., Foz do Iguaçu, Brazil, 24–28, Apr. Bourassa, M. A., D. M. Legler, J. J. O’Brien, and S. R. Smith (2003), SeaWinds validation with research vessels, J. Geophys. Res., 108(C2), 3019, doi:10.1029/2001JC001028. Brennan, M. J., C. C. Hennon, and R. D. Knabb (2009), The operational use of QuikSCAT ocean surface vector winds at the National Hurricane Center, Weather Forecasting, 24, 621–645, doi:10.1175/ 2008WAF2222188.1. Bretherton, F. P., R. E. Davis, and C. B. Fandry (1976), A technique for objective analysis and design of oceanographic experiments applied to MODE‐73, Deep Sea Res., 23, 559–582. Carrère, L., F. Mertz, J. Dorandeu, Y. Quilfen, and J. Patoux (2009), Observing and studying extreme low pressure events with altimetry, Sensors, 9, 1306–1329, doi:10.3390/s90301306. Chelton, D. B., R. A. deSzoeke, M. G. Schlax, K. E. Naggar, and N. Siwertz (1998), Geographical variability of the first baroclinic Rossby radius of deformation, J. Phys. Oceanogr., 28, 433–460. Dorandeu, J., and P. Y. Le Traon (1999), Effects of global mean atmospheric pressure variations on mean sea level changes from TOPEX/ Poseidon, J. Atmos. Oceanic Technol., 16, 1279–1283. Ducet, N., P. Y. Le Traon, and G. Reverdin (2000), Global high‐resolution mapping of ocean circulation from the combination of T/P and ERS‐1/2, J. Geophys. Res., 105, 19,477–19,498. Durrant, T. H., D. J. M. Greenslade, and I. Simmonds (2009), Validation of Jason‐1 and Envisat remotely sensed wave heights, J. Atmos. Oceanic Technol., 26, 123–134. Emanuel, K., C. DesAutels, C. Holloway, and R. Korty (2004), Environmental control of tropical cyclone intensity, J. Atmos. Sci., 61, 843–858. Forbes, C., K. D. Leaman, D. B. Olson, and O. B. Brown (1993), Eddy and wave dynamics in the South Atlantic as diagnosed from GEOSAT altimeter data, J. Geophys. Res., 98(C7), 12,297–12,314. Gill, A. E. (1982), Atmosphere‐Ocean Dynamics, 662 pp., Academic, New York. Goni, G., and J. A. Trinanes (2003), Ocean thermal structure monitoring could aid in the intensity forecast of tropical cyclones, Eos Trans. AGU, 84(51), 573, 577–578. Gordon, A. L., and W. F. Haxby (1990), Agulhas eddies invade the South Atlantic: Evidence from GEOSAT altimeter and shipboard conductivity‐ temperature‐depth survey, J. Geophys. Res., 95(C3), 3117–3125. Gourrion, J., D. Vandemark, S. Bailey, B. Chapron, G. P. Gommenginger, P. G. Challenor, and M. A. Srokosz (2002), A two‐parameter wind speed algorithm for Ku‐Band altimeters, J. Atmos. Oceanic Technol., 19, 2030–2048, doi:10.1175/1520-0426(2002)0192.0. CO;2. Hilburn, K. A., F. J. Wentz, D. K. Smith, and P. Ashcroft (2006), Correcting active scatterometer data for the effects of rain using passive radiometer data, J. Appl. Meteorol. Climatol., 45, 382–398. Holthuijsen, L. H., and H. L. Tolman (1991), Effects of the Gulf Stream on ocean waves, J. Geophys. Res., 96(C7), 12,755–12,771. Hong, X., S. W. Chang, S. Raman, L. K. Shay, and R. Hodur (2000), The interaction between hurricane Opal (1995) and a warm core ring in the Gulf of Mexico, Mon. Weather Rev., 128, 1347–1365. Jacob, S. D., L. Shay, and A. J. Mariano (2000), The 3D oceanic mixed layer response to hurricane Gilbert, J. Phys. Oceanogr., 30, 1407–1429. Jacobs, G., C. Barron, D. Fox, K. Whitmer, S. Klingenberger, D. May, and J. Blaha (2002), Operational altimeter sea level products, Oceanography, 15(1), 13–21. Jaimes, B., and L. K. Shay (2009), Mixed layer cooling in mesoscale oceanic eddies during hurricanes Katrina and Rita, Mon. Weather Rev., 137, 4188–4207, doi:10.1175/2009MWR2849.1. Kaplan, J., and M. DeMaria (2003), Large‐scale characteristics of rapidly intensifying tropical cyclones in the North Atlantic basin, Weather Forecasting, 18, 1093–1108, doi:10.1175/1520-0434(2003)0182.0.CO;2. Kara, A. B., C. N. Barron, J. Wallcraft, and T. Oguz (2008), Interannual variability of sea surface height over the Black Sea: Relation to climate patterns, Earth Interact., 12, 1–11, doi:10.1175/2008EI259.1.

18 of 19

C07002


Lawrence, M. B., and J. M. Pelissier (1981), Atlantic hurricane season of 1980, Mon. Weather Rev., 109, 1567–1582. Lin, I., C. Wu, K. A. Emanuel, I. Lee, C. Wu, and I. Pun (2005), The interaction of supertyphoon Maemi (2003) with a warm ocean eddy, Mon. Weather Rev., 133, 2635–2649. Lin, I., C. Chen, I. Pun, W. T. Liu, and C. Wu (2009), Warm ocean anomaly, air sea fluxes, and the rapid intesification of tropical cyclone Nargis (2008), Geophys. Res. Lett., 36, L03817, doi:10.1029/2008GL035815. Mainelli, M., M. DeMaria, L. K. Shay, and G. Goni (2008), Application of oceanic heat content estimation to operational forecasting of recent Atlantic category 5 hurricanes, Weather Forecasting, 23, 3–16, doi:10.1175/2007WAF2006111.1. Marks, F. D., L. K. Shay, and PDT‐5 (1998), Landfalling tropical cyclones: Forecast problems and associated research opportunities, Bull. Am. Meteorol. Soc., 79, 305–323. McTaggart‐Cowan, R., L. F. Bosart, C. A. Davis, E. H. Atallah, J. R. Gyakum, and K. A. Emanuel (2006), Analyses of hurricane Catarina, Mon. Weather Rev., 134, 3029–3053. Montegut, C. B., G. Madec, A. S. Fischer, A. Lazar, and D. Iudicone (2004), Mixed layer depth over the global ocean: An examination of profile data and profile‐based climatology, J. Geophys. Res., 109, C12003, doi:10.1029/2004JC002378. Nguyen, V., R. Smith, and M. Montgomery (2008), Tropical‐cyclone intensification and predictability in three dimensions, Q. J. R. Meteorol. Soc., 134, 563–582, doi:10.1002/qj.235. Oey, L., T. Ezer, D. Wang, X. Yin, and S. Fan (2007), Hurricane‐ induced motions and interaction with ocean curents, Cont. Shelf Res., 27, 1249–1263, doi:10.1016/j.csr.2007.01.008. Pereira Filho, A. J., A. B. Pezza, I. Simmonds, R. S. Lima, and M. L. Vianna (2010), New perspectives on the synoptic and mesoscale structure of hurricane Catarina, Atmos. Res., 95(2–3), 157–171, doi:10.1016/j. atmosres.2009.09.009. Perry, K. L. (2001), SeaWinds on QuikSCAT Level 3 Daily, Gridded Ocean Wind Vectors (JPL SeaWinds Project), Guide Doc. JPL D‐20335, Phys. Oceanogr. Distrib. Active Arch. Cent., NASA Jet Propul. Lab., Pasadena, Calif. Pezza, A. B., and I. Simmonds (2005), The first South Atlantic hurricane: Unprecedented blocking, low shear and climate change, Geophys. Res. Lett., 32, L15712, doi:10.1029/2005GL023390. Pezza, A. B., and I. Simmonds (2008), Large‐scale factors in tropical and extratropical cyclone transition and extreme weather events, Ann. N. Y. Acad. Sci., 1146, 189–211, doi:10.1196/annals.1446.005. Pezza, A. B., I. Simmonds, and A. Pereira Filho (2009), Climate perspective on the large scale associated with the transition of the first South Atlantic hurricane, Int. J. Climatol., 29, 1116–1130, doi:10.1002/ joc17.57. Powell, M. D., S. H. Houston, and T. A. Reinhold (1996), Hurricane Andrew’s landfall in South Florida. Part I: Standardizing measurements for documentation of surface wind fields, Weather Forecasting, 11, 304–328. Price, J. F. (1981), Upper ocean response to a hurricane, J. Phys. Oceanogr., 11, 153–175. Pun, I.‐F., I.‐I. Lin, C.‐R. Wu, D.‐S. Ko, and W. Liu (2007), Validation and application of altimetry‐derived upper ocean thermal structure in

C07002

the western north Pacific ocean for typhoon‐intensity forecast, IEEE Trans. Geosci. Remote Sens., 45(6), 1616–1630, doi:10.1109/ TGRS.2007. 895950. Quilfen, Y., J. Tournadre, and B. Chapron (2006), Altimeter dual‐frequency observations of surface wind, waves, and rain rate in tropical cyclone Isabel, J. Geophys. Res., 111, C01004, doi:10.1029/2005JC003068. Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax (2007), Daily high‐resolution‐blended analyses for sea surface temperature, J. Clim., 20, 5473–5496, doi:10.1175/ 2007JCLI1824.1. Scharroo, R., W. H. F. Smith, and J. L. Lillibridge (2005), Satellite altimetry and the intensification of hurricane Katrina, Eos Trans. AGU, 86(40), 366–367. Scharroo, R., W. H. F. Smith, and J. L. Lillibridge (2006), Reply to comment on “Satellite altimetry and the intensification of hurricane Katrina,” Eos Trans. AGU, 87(8), 89. Shay, L. K., G. J. Goni, and P. G. Black (2000), Effects of a warm oceanic feature on Hurricane Opal, Mon. Weather Rev., 128, 1366–1383. Silva Dias, P., M. A. F. Silva Dias, M. Seluchi, and F. A. Diniz (2004), Preliminary analysis of structure, dynamics and predictability, paper presented at 13th Brazilian Meteorological Conference, Soc. Bras. de Meteorol., Fortaleza, Brazil, 29 Aug. to 3 Sept. Smith, D. K., F. J. Wentz, and C. A. Mears (2003), Results of QuikSCAT high wind data validation, paper presented at 83rd Annual Meeting, Am. Meteorol. Soc., Long Beach, Calif., 11–12 Feb. Veiga, J. A. P., A. B. Pezza, I. Simmonds, and P. Silva Dias (2008), An analysis of the environmental energetics associated with the transition of the first South Atlantic hurricane, Geophys. Res. Lett., 35, L15806, doi:10.1029/2008GL034511. Vianna, M., and V. V. Menezes (2010), Mean mesoscale global ocean currents from geodetic pre‐GOCE MDTs with a synthesis of the North Pacific circulation, J. Geophys. Res., 115, C02016, doi:10.1029/ 2009JC005494. Vianna, M. L., V. V. Menezes, and D. P. Chambers (2007), A high resolution satellite‐only GRACE‐based mean dynamic topography of the South Atlantic Ocean, Geophys. Res. Lett., 34, L24604, doi:10.1029/ 2007GL031912. Wentz, F. J. (2001), Advanced algorithms for QuikSCAT and SeaWinds/ AMSR, paper presented at Internatioal Geoscience and Remote Sensing Symposium 2001, Univ. of N. S. W., Sydney, N. S. W., Australia, 9–13 Jul. Wu, C., C. Lee, and I. Lin (2007), The effect of the ocean eddy on tropical cyclone intensity, J. Atmos. Sci., 64, 3562–3578, doi:10.1175/ JAS4051.1. V. V. Menezes, VM Oceânica LTDA, Tr Santa Cruz 70/102, Arraial do Cabo, RJ 2893‐0000, Brazil. ([email protected]) A. B. Pezza and I. Simmonds, School of Earth Sciences, University of Melbourne, Victoria 3010, Australia. ([email protected]; [email protected]) M. L. Vianna, VM Oceânica LTDA, R Manoel Bandeira 210, San Jose dos Campos, SP 1224‐0710, Brazil. ([email protected])

19 of 19