• Asano, J., and Coauthors, 2008: Analysis of tropical cyclones using microwave satellite imagery. RSMC Tokyo-Typhoon Center Tech. Rev., 10 , 3070.

    • Search Google Scholar
    • Export Citation
  • Barnes, S. L., 1964: A technique for maximizing details in numerical weather map analysis. J. Appl. Meteor., 3 , 396409.

  • Bessho, K., , M. DeMaria, , and J. A. Knaff, 2006: Tropical cyclone wind retrievals from the Advanced Microwave Sounding Unit (AMSU): Application to surface wind analysis. J. Appl. Meteor. Climatol., 45 , 399415.

    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., , and P. K. Smolarkiewicz, 1989: Gravity waves, compensating subsidence, and detrainment around cumulus clouds. J. Atmos. Sci., 46 , 740759.

    • Search Google Scholar
    • Export Citation
  • Briegel, L. M., , and W. M. Frank, 1997: Large-scale influences on tropical cyclogenesis in the western North Pacific. Mon. Wea. Rev., 125 , 13971413.

    • Search Google Scholar
    • Export Citation
  • Brueske, K. F., , and C. S. Velden, 2003: Satellite-based tropical cyclone intensity estimation using the NOAA-KLM series Advanced Microwave Sounding Unit (AMSU). Mon. Wea. Rev., 131 , 687697.

    • Search Google Scholar
    • Export Citation
  • Chen, S. S., , and R. A. Houze Jr., 1997: Diurnal variation and life-cycle of deep convective systems over the tropical Pacific warm pool. Quart. J. Roy. Meteor. Soc., 123 , 357388.

    • Search Google Scholar
    • Export Citation
  • Cheung, K. K. W., 2004: Large-scale environmental parameters associated with tropical cyclone formations in the western North Pacific. J. Climate, 17 , 466484.

    • Search Google Scholar
    • Export Citation
  • Demuth, J. L., , M. DeMaria, , J. A. Knaff, , and T. H. Vonder Haar, 2004: Evaluation of Advanced Microwave Sounding Unit tropical-cyclone intensity and size estimation algorithms. J. Appl. Meteor., 43 , 282296.

    • Search Google Scholar
    • Export Citation
  • Demuth, J. L., , M. DeMaria, , and J. A. Knaff, 2006: Improvement of Advanced Microwave Sounding Unit tropical cyclone intensity and size estimation algorithms. J. Appl. Meteor. Climatol., 45 , 15731581.

    • Search Google Scholar
    • Export Citation
  • Dickinson, M., , and J. Molinari, 2002: Mixed Rossby–gravity waves and western Pacific tropical cyclogenesis. Part I: Synoptic evolution. J. Atmos. Sci., 59 , 21832196.

    • Search Google Scholar
    • Export Citation
  • Dvorak, V. F., 1975: Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon. Wea. Rev., 103 , 420430.

  • Dvorak, V. F., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Rep. NESDIS 11, Washington, DC, 47 pp.

  • Fu, B., , T. Li, , M. S. Peng, , and F. Weng, 2007: Analysis of tropical cyclogenesis in the western North Pacific for 2000 and 2001. Wea. Forecasting, 22 , 763780.

    • Search Google Scholar
    • Export Citation
  • Gierach, M. M., , M. A. Bourassa, , P. Cunningham, , J. J. O’Brien, , and P. D. Reasor, 2007: Vorticity-based detection of tropical cyclogenesis. J. Appl. Meteor. Climatol., 46 , 12141229.

    • Search Google Scholar
    • Export Citation
  • Gray, W. M., , and R. W. Jacobson Jr., 1977: Diurnal variation of deep cumulus convection. Mon. Wea. Rev., 105 , 11711188.

  • Halverson, J. B., , J. Simpson, , G. Heymsfield, , H. Pierce, , T. Hock, , and L. Ritchie, 2006: Warm core structure of Hurricane Erin diagnosed from high altitude dropsondes during CAMEX-4. J. Atmos. Sci., 63 , 309324.

    • Search Google Scholar
    • Export Citation
  • Hawkins, H. F., , and D. T. Rubsame, 1968: Hurricane Hilda, 1964. II. Structure and budgets of the hurricane on October 1, 1964. Mon. Wea. Rev., 96 , 617636.

    • Search Google Scholar
    • Export Citation
  • Hawkins, H. F., , and S. M. Imbembo, 1976: The structure of a small, intense hurricane—Inez 1966. Mon. Wea. Rev., 104 , 418442.

  • Hawkins, J. D., , T. F. Lee, , J. Turk, , C. Sampson, , J. Kent, , and K. Richardson, 2001: Real-time Internet distribution of satellite products for tropical cyclone reconnaissance. Bull. Amer. Meteor. Soc., 82 , 567578.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, G. M., , J. B. Halverson, , J. Simpson, , L. Tian, , and T. P. Bui, 2001: ER-2 Doppler radar investigations of the eyewall of Hurricane Bonnie during the Convection and Moisture Experiment-3. J. Appl. Meteor., 40 , 13101330.

    • Search Google Scholar
    • Export Citation
  • Hoshino, S., , and T. Nakazawa, 2007: Estimation of tropical cyclone’s intensity using TRMM/TMI brightness temperature data. J. Meteor. Soc. Japan, 85 , 437454.

    • Search Google Scholar
    • Export Citation
  • Katsaros, K. B., , E. B. Forde, , P. Chang, , and W. T. Liu, 2001: QuikSCAT’s SeaWinds facilitates early identification of tropical depressions in 1999 hurricane season. Geophys. Res. Lett., 28 , 10431046.

    • Search Google Scholar
    • Export Citation
  • Kidder, S. Q., , M. D. Goldberg, , R. M. Zehr, , M. DeMaria, , J. F. W. Purdom, , C. S. Velden, , N. C. Grody, , and S. J. Kusselson, 2000: Satellite analysis of tropical cyclones using the Advanced Microwave Sounding Unit (AMSU). Bull. Amer. Meteor. Soc., 81 , 12411259.

    • Search Google Scholar
    • Export Citation
  • Kishimoto, K., , T. Nishigaki, , S. Nishimura, , and Y. Terasaka, 2007: Comparative study on organized convective cloud systems detected through early stage Dvorak analysis and tropical cyclones in early developing stage in the western North Pacific and the South China Sea. RSMC Tokyo-Typhoon Center Tech. Rev., 9 , 1932.

    • Search Google Scholar
    • Export Citation
  • Knaff, J. A., , R. M. Zehr, , M. D. Goldberg, , and S. Q. Kidder, 2000: An example of temperature structure differences in two cyclone systems derived from the Advanced Microwave Sounder Unit. Wea. Forecasting, 15 , 476483.

    • Search Google Scholar
    • Export Citation
  • Knaff, J. A., , S. A. Seseske, , M. DeMaria, , and J. L. Demuth, 2004: On the influences of vertical wind shear on symmetric tropical cyclone structure derived from AMSU. Mon. Wea. Rev., 132 , 25032510.

    • Search Google Scholar
    • Export Citation
  • Lee, C. S., 1989: Observational analysis of tropical cyclogenesis in the western North Pacific. Part I: Structural evolution of cloud clusters. J. Atmos. Sci., 46 , 25802598.

    • Search Google Scholar
    • Export Citation
  • Lee, T. F., , F. J. Turk, , J. Hawkins, , and K. Richardson, 2002: Interpretation of TRMM TMI images of tropical cyclones. Earth Interactions, 6 .[Available online at http://EarthInteractions.org].

    • Search Google Scholar
    • Export Citation
  • McBride, J. L., , and R. Zehr, 1981: Observational analysis of tropical cyclone formation. Part II: Comparison of non-developing versus developing systems. J. Atmos. Sci., 38 , 11321151.

    • Search Google Scholar
    • Export Citation
  • Muramatsu, T., 1983: Diurnal variations of satellite-measured TBB areal distribution and eye diameter of mature typhoons. J. Meteor. Soc. Japan, 61 , 7790.

    • Search Google Scholar
    • Export Citation
  • Nitta, T., , and S. Sekine, 1994: Diurnal variation of convective activity over the tropical western Pacific. J. Meteor. Soc. Japan, 72 , 627641.

    • Search Google Scholar
    • Export Citation
  • Olander, T. L., , and C. S. Velden, 2007: The Advanced Dvorak Technique: Continued development of an objective scheme to estimate tropical cyclone intensity using geostationary infrared satellite imagery. Wea. Forecasting, 22 , 287298.

    • Search Google Scholar
    • Export Citation
  • Ritchie, E. A., , and G. J. Holland, 1999: Large-scale patterns associated with tropical cyclogenesis in the western Pacific. Mon. Wea. Rev., 127 , 20272043.

    • Search Google Scholar
    • Export Citation
  • Schubert, W. H., , and J. J. Hack, 1982: Inertial stability and tropical cyclone development. J. Atmos. Sci., 39 , 16871697.

  • Sharp, R. J., , M. A. Bourassa, , and J. J. O’Brien, 2002: Early detection of tropical cyclones using Seawinds-derived vorticity. Bull. Amer. Meteor. Soc., 83 , 879889.

    • Search Google Scholar
    • Export Citation
  • Tsuchiya, A., , T. Mikawa, , and A. Kikuchi, 2001: Method of distinguishing between early stage cloud systems that develop into tropical storms and ones that do not. Geophys. Mag., 4 , 4959.

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., , T. L. Olander, , and R. M. Zehr, 1998: Development of an objective scheme to estimate tropical cyclone intensity from digital geostationary satellite infrared imagery. Wea. Forecasting, 13 , 172186.

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., and Coauthors, 2006: The Dvorak tropical cyclone intensity estimation technique: A satellite-based method that has endured for over 30 years. Bull. Amer. Meteor. Soc., 87 , 11951210.

    • Search Google Scholar
    • Export Citation
  • Zehr, R. M., 1992: Tropical cyclogenesis in the Western North Pacific. NOAA Tech. Rep. NESDIS 61, Washington, DC, 181 pp.

  • View in gallery

    The four cloud patterns (i–iv) that define CSCs (from Tsuchiya et al. 2001).

  • View in gallery

    A conceptual model of cloud clusters satisfying the T1 classification. The shaded areas are dense, cold (≤−31°C) overcast. The estimation accuracy of the CSC is expressed by the dotted circle with a diameter of 2.5° (condition 2). The dashed circle with a radius of 2.0° indicates the region that includes the dense, cold overcast (condition 4). The diameter of the dashed circle drawn in the overcast on the right side of the CSC is 1.5°. This circle shows the size of the overcast (condition 5; from Tsuchiya et al. 2001).

  • View in gallery

    Horizontal images of (right) AMSU-retrieved temperature anomalies (K) at 200 hPa and (left) GOES-9 infrared brightness temperatures for (a) EDA0428 at 2100 UTC 3 Jun and (b) EDA0453 at 2100 UTC 13 Aug. The cross hairs show the position of the CSC. Shading shows positive temperature anomalies, and a darker shade means a larger anomaly.

  • View in gallery

    Vertical cross sections showing AMSU-retrieved temperature anomalies (K) along an east–west line through the CSC at the same observation time as Fig. 3 for (a) EDA0428 and (b) EDA0453. Shading is as in Fig. 3.

  • View in gallery

    Pressure vs time cross sections of temperature anomalies (K) averaged within a 4° latitude by 4° longitude rectangle centered on the CSCs. Temperature anomalies are retrieved from AMSU observations in the cases of (a) EDA0428 and (b) EDA0453. Shading is the same as in Fig. 3.

  • View in gallery

    Positions of observed AMSU-based maximum temperature anomalies (K) at 200 hPa for OCCTS within a 10° latitude by 10° longitude rectangle centered on the CSC. Results are associated with T-number values of (a) 0.0, (b) 1.0, (c) 1.5, and (d) 2.0.

  • View in gallery

    Time series graphs of 200-hPa temperature anomalies (K) averaged within a 4° latitude by 4° longitude rectangle centered on the CSC. Values are shown for all cases of (a) OCCL, (b) OCCTD, and (c) OCCTS from the detection of CSCs to dissipation or to recognition as TSs. The red (black) lines show the cases with (without) anomalies larger than 0.9 K.

  • View in gallery

    Average lifetime of (a) the 25 OCCTS cases with a WCT and the first judgment of T1, (b) the 50 OCCL cases with no WCT, and (c) the 25 OCCTS cases with a WCT and the first identification of a WCT in an AMSU observation. In the diagram, CSC means the detection of the CSC, and TS refers to the evolution to TS status.

  • View in gallery

    Percentages of each of the five conditions (1–5) used in T1 diagnosis for all observational cases of OCCL (black bar) and OCCTS (white bar) with T numbers of 0.0.

  • View in gallery

    Diurnal variations of 200-hPa temperature anomalies (K) averaged at each AMSU local observational time for all OCCs classified by each T number from 0.0 to 2.0.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 33 33 5
PDF Downloads 48 48 7

Warm Core Structures in Organized Cloud Clusters Developing or Not Developing into Tropical Storms Observed by the Advanced Microwave Sounding Unit

View More View Less
  • 1 Meteorological Research Institute, Tsukuba, Japan
  • | 2 Japan Meteorological Agency/Regional Specialized Meteorological Center-Tokyo, Tokyo, Japan
  • | 3 Japan Meteorological Agency/Meteorological Satellite Center, Kiyose, Japan
© Get Permissions
Full access

Abstract

The temperature profiles of organized cloud clusters developing or not developing (nondeveloping) into tropical storms (TSs; maximum surface wind >34 kt) over the western North Pacific in 2004 were investigated using Advanced Microwave Sounding Unit (AMSU) observations in combination with the independently created early stage Dvorak analysis. Typical temperature profiles of the developing and nondeveloping cloud clusters were compared. From this comparison, positive upper-troposphere temperature anomalies were found in both cluster types; however, the spatial extent of the temperature anomalies for the developing cloud clusters was larger than those of the nondeveloping cloud clusters. Statistical analysis was performed on the temperature anomalies near the center of all clusters retrieved from AMSU observational data. Findings indicate that the area-average temperature anomalies increased along with the intensity of the clusters indicated by the Dvorak T-number classification. Using time series analysis of upper-level temperature anomalies associated with these cloud clusters, a definition of warm core structures showing the temperature anomaly greater than a threshold (WCT) was created. WCT exists when the area averaged temperature anomaly exceeds 0.9 K. Using this definition, almost 70% of the cloud clusters that had WCTs later became TSs, while 85% of those that did not have WCTs eventually dissipated without being classified as a TS. For the WCT clusters that developed into TSs, the lead time from the detection of their AMSU-based WCT to their classification as TSs was 27.7 h. These results indicate that there is a good possibility that the detection and forecasting of tropical cyclone formation, particularly those storms that later may become classified as TSs, will be improved using temperature anomalies derived from AMSU data.

Corresponding author address: Kotaro Bessho, Japan Meteorological Agency/Meteorological Research Institute, Tsukuba City, Ibaraki 305-0052, Japan. Email: kbessho@mri-jma.go.jp

Abstract

The temperature profiles of organized cloud clusters developing or not developing (nondeveloping) into tropical storms (TSs; maximum surface wind >34 kt) over the western North Pacific in 2004 were investigated using Advanced Microwave Sounding Unit (AMSU) observations in combination with the independently created early stage Dvorak analysis. Typical temperature profiles of the developing and nondeveloping cloud clusters were compared. From this comparison, positive upper-troposphere temperature anomalies were found in both cluster types; however, the spatial extent of the temperature anomalies for the developing cloud clusters was larger than those of the nondeveloping cloud clusters. Statistical analysis was performed on the temperature anomalies near the center of all clusters retrieved from AMSU observational data. Findings indicate that the area-average temperature anomalies increased along with the intensity of the clusters indicated by the Dvorak T-number classification. Using time series analysis of upper-level temperature anomalies associated with these cloud clusters, a definition of warm core structures showing the temperature anomaly greater than a threshold (WCT) was created. WCT exists when the area averaged temperature anomaly exceeds 0.9 K. Using this definition, almost 70% of the cloud clusters that had WCTs later became TSs, while 85% of those that did not have WCTs eventually dissipated without being classified as a TS. For the WCT clusters that developed into TSs, the lead time from the detection of their AMSU-based WCT to their classification as TSs was 27.7 h. These results indicate that there is a good possibility that the detection and forecasting of tropical cyclone formation, particularly those storms that later may become classified as TSs, will be improved using temperature anomalies derived from AMSU data.

Corresponding author address: Kotaro Bessho, Japan Meteorological Agency/Meteorological Research Institute, Tsukuba City, Ibaraki 305-0052, Japan. Email: kbessho@mri-jma.go.jp

1. Introduction

Numerous cloud clusters form over the western North Pacific. Some of these are well organized, rotate cyclonically, and eventually develop into tropical storms (TSs), although most dissipate. Despite the operational importance of distinguishing which cloud clusters will become TSs, it remains difficult to forecast tropical cyclone genesis using numerical weather prediction models. For this reason, tropical cyclone forecasters worldwide have relied on the Dvorak technique to estimate the potential of cloud clusters to develop into TSs (Dvorak 1975, 1984). This subjective analysis technique uses cloud patterns in visible and infrared satellite imagery to evaluate the intensity (central surface pressure and maximum wind speed) of cloud clusters, in the formative stages, as well as mature tropical cyclones. The intensity estimates are given in terms of the tropical number (or T number). The T-number index is equated to tropical cyclone intensity via a lookup table, where T numbers range from 1 to 8 and are counted in increments of 0.5. T number 1 (T1) corresponds to the minimum level of tropical cyclone intensity, while T8 describes the maximum intensity.

The Dvorak technique has endured for more than 30 years and has saved many lives (Velden et al. 2006). While this technique is considered to be a de facto standard of satellite analysis for tropical cyclones, it is not without its limitations. First, it provides a subjective intensity estimate, and considerable training and experience are needed to master the technique. Second, it is difficult to estimate the intensity of tropical cyclones correctly in their genesis stage, especially when the low-level cyclonic circulation is obscured by high clouds. The cyclonic circulations in many formative tropical cyclones are usually so small that they cannot be analyzed properly using the Dvorak method. Dvorak classification is also especially difficult when such circulations are embedded within the monsoon troughs—a common situation in the western North Pacific. While an objective version of the Dvorak technique has been developed to overcome the first limitation (i.e., subjectivity; Velden et al. 1998; Olander and Velden 2007), the other shortcomings survive. Thus, there remains a distinct need for a new approach to improve the detection and intensity estimates of tropical cyclones during the genesis stage.

Recently, microwave sensors (including microwave imagers, sounders, and scatterometers) on board low-earth-orbit satellites have been significantly improved and enhanced. Among these, microwave imagers are used to locate the center of tropical cyclones and ascertain their structure (Hawkins et al. 2001; Lee et al. 2002; Asano et al. 2008). They can detect microwave radiation from rain and ice particles through the dense clouds found in tropical cyclones. Using these characteristics of microwave imagers, Hoshino and Nakazawa (2007) presented an objective method of intensity estimation for tropical cyclones in the developing and mature stages. Microwave scatterometers can also estimate the sea surface wind distribution in and around tropical cyclones (Katsaros et al. 2001). By way of example, Sharp et al. (2002) and Gierach et al. (2007) inferred tropical cyclone genesis using the vorticity retrieved from observational data produced by the scatterometer of the Quick Scatterometer (QuikSCAT). Unfortunately, however, QuikSCAT observations were only available at most twice a day for any given tropical cyclone, and its mission ended in November 2009. There is still another microwave scatterometer of an Advanced Scatterometer instrument on the Meteorological Operation (MetOp; from the European Organisation for the Exploitation of Meteorological Satellites) satellite, which provides a partial mitigation for QuikSCAT.

One of the recent generations of microwave sounders, the Advanced Microwave Sounding Units (AMSUs) are now routinely used to observe air temperature profiles within dense clouds at a more detailed level of spatial resolution than former sounders (Kidder et al. 2000; Knaff et al. 2000) and routinely assimilated into global numerical weather prediction models. AMSUs are now employed on the National Oceanic and Atmospheric Administration satellites NOAA-15, -16, -18, and -19, MetOp, and the National Areonautic and Space Administration Earth Observing Satellite Aqua. Using the observations from six of the AMSUs, the air temperature structures of tropical cyclones can be sensed up to 12 times a day, effectively monitoring the upper-level warm core structures of tropical cyclones via, among other methods, AMSU-based temperature retrievals. A warm core is typically defined as the central region of a tropical cyclone where the air temperature is higher than that of the surrounding environment because of the latent heat released by active convection within the cyclone. Warm cores are generally observed in both the developing and mature stages of tropical cyclones (Hawkins and Rubsame 1968; Hawkins and Imbembo 1976; Heymsfield et al. 2001; Halverson et al. 2006). Some researches, who were mainly targeting the developing or mature stages of tropical cyclones, have statistically related variations of AMSU-based warm core signals to the tropical cyclone intensity (Brueske and Velden 2003; Demuth et al. 2004, 2006). However, the detailed conditions regarding warm core structures during the genesis stages of tropical cyclone development remains unclear. This leaves the possibility that AMSU-based observations of warm core structures may objectively complement the Dvorak technique to better anticipating tropical cyclone genesis.

In this study, the temperature structures of cloud clusters occurring in 2004, some of which developed into TSs while others did not, were investigated using the AMSUs on board NOAA-15 and -16. Particular attention was focused on the amplitude of positive temperature anomalies at 200–300 hPa associated with the clusters. This article will first show the early stage Dvorak analysis method and datasets used in this research in sections 2 and 3. Then we present the typical warm core structures associated with cloud clusters followed by a statistical investigation of the upper-level temperature anomaly amplitudes found in cloud clusters classified by their final developmental stage, which are described in sections 4 and 5. The warm core structure showing the temperature anomaly greater than a threshold (WCT) in the cloud cluster will then be defined in the next section. The duration from first detection of WCT to TSs is also investigated in section 7. Finally, from the statistical analysis results of the warm core structures in the clusters, we will discuss the possibility of distinguishing whether a particular cluster will or will not develop into a tropical cyclone as a function of the existence of WCTs in that cloud cluster.

2. Early stage Dvorak analysis

Since 1999 the Meteorological Satellite Center (MSC) of the Japan Meteorological Agency (JMA) has routinely monitored organized cloud clusters (OCCs) over the western North Pacific that had the potential to develop into TSs [defined in the study by Tsuchiya et al. (2001)] and logged their locations and T numbers at 6-hourly intervals. This process is called early stage Dvorak analysis (EDA). EDA is a part of JMA’s Dvorak analysis and also depends on subjective classification using observational data from geostationary satellites. We used the EDA dataset in this study to classify OCCs by their final developmental stage.

JMA classifies tropical cyclones over the western North Pacific Ocean into four grades based on their maximum wind speed (MWS; Table 1): typhoon (TY; MWS ≥ 64 kt), severe tropical storm (STS; MWS from 48 to <64 kt), tropical storm (TS; MWS from 34 to <48 kt), and tropical depression (TD; MWS < 34 kt). JMA also names tropical cyclones with an MWS > 34 kt and issues associated gale warnings to marine users. EDA, which is employed to distinguish TDs from TSs, includes the two steps outlined next.

As the first step, analysts investigate the cloud cluster’s cloud system center (CSC) to distinguish OCCs from other cloud clusters. In this paper, an OCC is defined as a cloud cluster with a CSC. According to Dvorak (1984), a CSC has at least one of the following four features (Fig. 1). To determine the CSC, analysts select the most suitable of the following patterns:

  • i) Dense, cold (−31°C or colder) overcast bands that show some curvature around a relatively warm area. They should curve at least one-fifth of the distance around a 10° logarithmic spiral. When visible observations are available, cirrus lines will indicate anticyclonic vertical wind shear across the expected CSC.
  • ii) Curved cirrus lines indicating a center of curvature within or near a dense, cold (−31°C or colder) overcast.
  • iii) Curved low cloud lines showing a center of curvature within two degrees of a cold (−31°C or colder) cloud mass.
  • iv) Cumulonimbus clusters rotating cyclonically in animated images.

After detecting the CSC, analysts give a T1 designation as the second step of EDA. In this diagnosis, the T number is determined as 1 when cloud systems have all five of the following conditions (Fig. 2):

  1. The cloud clusters have persisted for 12 h or more.
  2. The accuracy of estimation for the CSC in the clusters is 2.5° latitude or less.
  3. The CSC has persisted for 6 h or more.
  4. The clusters have dense, cold (−31°C or colder) overcast that appear less than 2° latitude from the CSC.
  5. The extent of the overcast is more than 1.5° latitude.

An OCC is usually classified as a TD (not yet a TS) when the T number becomes 1 or more and is classified as a TS when the MWS of the TD reaches 34 kt or more. Conversely, if an OCC has a T number smaller than 1, it is referred to as a low pressure area (L) in EDA.

Kishimoto et al. (2007) compared OCCs in EDA with tropical cyclones on weather charts that were detected by JMA using surface observations, geostationary meteorological satellite images, and observational results from microwave imagers and scatterometers in a comprehensive manner from 2002 to 2006. They found that 46% of OCCs with T number less than 1 became TDs, and 26% of them eventually developed to TSs. On the other hand, for OCCs with a T number of 1.0, 74% of them corresponded to TDs, and 61% of them evolved to TSs. And the percentage of OCCs with a T number of 2.0 that finally attained to TSs is 98%. From these results, it is understood that EDA has enough accuracy for analyses in this paper.

In this study, OCCs are classified by their final developmental stage as follows: OCCL are OCCs that stayed at the stage of L and eventually dissipated; OCCTD are OCCs that developed into TDs but did not develop into TSs; and OCCTS are OCCs that developed into TSs.

3. Data

AMSU brightness temperature data from NOAA-15 and -16 in 2004 came courtesy of the Cooperative Institute for Research in the Atmosphere (CIRA) of Colorado State University (CSU). AMSU is a successor of the previous sounder, the Microwave Sounding Unit, on NOAA satellites and has a finer horizontal and vertical resolution (Kidder et al. 2000). AMSU consist of two parts: AMSU-A and AMSU-B. The main mission of AMSU-A is temperature sounding and that of AMSU-B is to make moisture soundings. The brightness temperature data from AMSU-A is used for our analyses. AMSU-A has 15 channels and their frequency range is from 23.8 to 89.0 GHz. It takes a cross-track scan with a swath width of 2179 km, and its best horizontal resolution at nadir is 48 km. The Noise Equivalent Delta-T (NEDT) of each AMSU-A channel is from 0.25 to 1.20 K.

The air temperature profiles around the OCCs were retrieved by a Demuth–DeMaria–Knaff (DDK) algorithm developed by CIRA (Demuth et al. 2004; Bessho et al. 2006). The retrieved temperature data on the footprint of AMSU were interpolated into a grid of 24° latitude by 24° longitude with a resolution of 0.2° centered on the CSC of the OCCs using the method of Barnes (1964). The retrieved air temperature dataset had 30 pressure levels from 1000 to 10 hPa. Because of the attenuation by heavy precipitation, the low-level temperature structure near the center of OCCs could not be represented correctly. Combined the attenuation effect with spatial resolution and NEDT of AMSU-A mentioned above, it is difficult to catch the thermal structure of very small systems. In this paper, the size of OCCs detected by EDA is at least several hundred kilometers. So it is considered that AMSU can depict the thermal structure of OCCs. But we still need to pay attention to the effect of heavy rain to low-level temperature profiles of OCCs.

Temperature anomalies were used for analysis of the upper-level temperature structure in OCCs. To calculate the temperature anomalies at each pressure level, the temperature along a rectangular frame of 10° latitude by 10° longitude centered on the CSC of the OCCs was averaged as environmental temperature at first. This environmental temperature is not an area mean within the rectangular frame. Then, the environmental temperature was subtracted from the temperature at each pressure level to compute the temperature anomalies. When we took the rectangular frame larger (smaller), the environmental temperature became smaller (larger) and the temperature anomalies became larger (smaller) because OCCs usually has warm core structure near the CSC. As previously mentioned, the swath width of AMSU is 2179 km. To keep the AMSU observational number for analyses as large as possible, the cases in which AMSU observational footprint covers more than 50% of the analysis area were used in this study. The covered analysis region includes CSC, warm core structures, and surroundings adequately. From this reason, it is believed that the rectangular frame of 10° latitude by 10° longitude is suitable for the analyses in this paper. This method is almost similar to that used in Knaff et al. (2004), except that the environmental temperature was azimuthally and radially averaged from 500–600-km radius. The size of the rectangular frame of 10° latitude by 10° longitude depends on the latitude of the center. So the environmental temperature used in Knaff et al. (2004) is more reasonable than what is used in this study. But actually, the environmental temperature difference between two method of rectangular frame and radial belt is an average of −0.03 K for all AMSU observational 616 cases of all OCCs, which is a very small value compared with other temperatures handled later in this paper. Therefore, the averaged temperature along a rectangular frame of 10° latitude by 10° longitude centered on the CSC of the OCCs was used as an environmental temperature for convenience of the calculation in this paper.

The appendix provides a list of OCCs in the 2004 EDA file, which shows their numbers, final stages, locations of starting points, starting times, first T1 recognition times for OCCTD or OCCTS, dissipation times for OCCL, gale warning issued times for OCCTD or OCCTS, and the corresponding JMA typhoon numbers and names. This list also includes the first recognition times of WCT structures in OCCs from the AMSU observational data, which are defined in the later sections. If an OCC had no AMSU data observations of its genesis to dissipation as L or evolution as TD or TS, it is described as “no data” in the column of first recognition times of WCTs from AMSU data.

The EDA log file for 2004 lists 100 OCCs in the western North Pacific. OCCs numbered as 0423, 0448, 0494, and 0495 were unassigned in the list of the appendix after the postanalysis. Among the 100 OCCs, 29 were OCCTS, 14 were OCCTD, and the others were OCCL. Eliminating the 6 cases with no AMSU observational data, 28 cases of OCCTS, 13 cases of OCCTD, and 53 cases of OCCL were available for this study. As the number of OCCTD is too small to compare with those of OCCL or OCCTS, this study will concentrate on the results of comparison between OCCL and OCCTS cases.

4. Examples of warm core structures of OCCs estimated from AMSU

The warm core structures of two typical OCCs with a T number less than 1 are shown here before presenting the statistics of the air temperature anomalies retrieved from AMSU within OCCs. One OCC is numbered as 0428 (referred to below as EDA0428), and the other is as 0453 (EDA0453). EDA0428, which was an OCCL, was detected as an OCC at 1800 UTC 3 June 2004 at 8.2°N, 151.4°E and dissipated at 0600 UTC 4 June after a lifetime of just 12 h. On the other hand, EDA0453, which was an OCCTS, was first detected as an OCC at 0600 UTC 13 August 2004 at 13.2°N, 144.4°E. It reached T1 at 0000 UTC 14 August at 14.5°N, 140.5°E and became a TD. It then developed into a TS at 0600 UTC 16 August at 18.8°N, 130.8°E and eventually was named Typhoon Megi. JMA numbered this typhoon 0415.

Figure 3 shows Geostationary Operational Environmental Satellite-9 (GOES-9) infrared imagery and AMSU temperature anomalies at 200 hPa for EDA0428 at 2100 UTC 3 June and EDA0453 at 2100 UTC 13 August. At 2100 UTC 3 June, 3 h had passed since EDA0428’s genesis, whereas 15 h had passed at 2100 UTC 13 August since EDA0453’s detection as an OCC. It took 2.5 days to reach the stage of TS. While EDA0453 had very large and active convective clouds near its CSC, EDA0428 had scattered convective clouds near the center. Though both OCCs have positive temperature anomalies in the AMSU images, they look quite different. For EDA0453 of OCCTS, the area of positive anomalies spreads widely corresponding to the large convective areas around its CSC. In contrast, EDA0428 of OCCL shows small regions of positive temperature anomalies.

Figure 4 shows vertical cross sections of the temperature anomalies in OCCs retrieved from AMSU data along an east–west line through the CSC at the same observational time as Fig. 3. It is important to note that retrieved temperatures are not reliable near the center of OCCs below 300 hPa because of contamination (scattering and attenuation) from heavy rainfall. Both EDA0428 and EDA0453 have positive temperature anomalies near the center from 300 to 150 hPa.

Figure 5 shows the pressure–time cross sections of temperature anomalies of OCCs averaged within a rectangle of 4° latitude by 4° longitude centered on the CSC. Though EDA0428 has positive anomalies between 300 and 200 hPa, the values were less than 1 K around 2100 UTC 3 June when the horizontal and vertical snapshots in Figs. 3a and 4a were observed. On the other hand, the anomalies of EDA0453 are sometimes large and sometimes small. At 2100 UTC 13 August (the same time as Figs. 3b and 4b), the anomalies peaked at more than 1 K before becoming weaker and then later exceeding the 1-K level again. From 1200 UTC 15 August, EDA0453 had anomalies that were greater than 2 K.

Similar features were found from other cases of OCCL and OCCTS. In the cases with T numbers equal to 0.0, the horizontal distributions of positive temperature anomalies were usually wider in OCCTS when compared to OCCL. The temperature anomalies of the OCCTS cases were also positive at 300–200 hPa near the CSC, and their maximum values were increasing with T numbers. Based on these results and other cases not shown, the next section will focus on the statistics associated with temperature anomalies at 300–200 hPa with respect to the various types of OCCs.

5. Statistics of air temperature anomalies at upper levels for each case of OCCs

To investigate the spread of the positive temperature anomaly region associated with OCCs, the positions of maximum temperature anomalies retrieved from AMSU observational data at 200 hPa for OCCTS are first plotted in the rectangle of 10° latitude by 10° longitude centered on the CSC (Fig. 6). In Fig. 6a, the locations of maximum temperature anomalies of OCCTS with T-number values of 0.0 are scattered throughout the whole plot area. In Fig. 6b the points for OCCTS with T-number values of 1.0 gather near the center of the domain. When OCCTS reach to the stage of T-number values of 1.5 or 2.0 (Figs. 6c,d), the locations of maximum temperature anomalies become centered within a 4° latitude by 4° longitude rectangle. To improve the classification of positive 200–300-hPa temperature anomalies associated with OCCs with large T numbers, the temperature anomalies from AMSU data were averaged within a rectangle of 4° latitude by 4° longitude centered on the CSCs in this section.

The statistics of 200–300-hPa temperature anomalies associated with OCCs are presented as Table 2 and are classified by their final stage and T-number value. At 200 hPa, the temperature anomalies of OCCTS averaged within a rectangle of 4° latitude by 4° longitude centered on the CSCs increase along with their T numbers (Table 2). When the T number has a value of 0.0, the average temperature anomaly is 0.39 K. A temperature anomaly of 0.56 K corresponds with a T number of 1.0 (i.e., TD stage), and a temperature anomaly of 0.72 K also corresponds well with T numbers of 1.5. The average temperature anomalies of 0.93 K are associated with T numbers of 2.0 and larger (i.e., TS stage), which is more than twice the value at a T number of 0.0. This increasing tendency of average 200-hPa temperature anomalies near the center of OCCs with increasing T numbers is similar when temperature anomalies at 250 and 300 hPa are examined (Table 2).

On the other hand, averages of maxima (i.e., individual points) of temperature anomalies at 200 hPa of OCCTS within a rectangle of 10° latitude by 10° longitude centered on the CSC show nearly constant values of 2.2–2.5 K. A comparison with values found at 250 and 300 hPa shows that average maximum anomalies at those levels associated with OCCTS also have almost constant values ranging from 1.5 to 1.7 and from 1.3 to 1.5 K, respectively. It is notable that the lower-level values are smaller than the value at 200 hPa.

Another way to examine the temperature anomalies is to calculate the percentage of grid points where temperature anomalies exceed 1 K within the rectangular area bounded by 4° latitude by 4° longitude and centered on the CSC. For the OCCTS cases at 200 hPa, the percentages of the air temperatures increase from 20.4% with a T number of 0.0 to 42.9% with a T number of 2.0. This tendency is also found when this analysis is conducted at 250 and 300 hPa. Furthermore, the locations of the points of maximum temperature anomalies at 250 and 300 hPa are collocated with the locations of points of the 200-hPa maximum temperature anomalies OCCTS as shown in Fig. 6.

Table 2 shows the statistics associated with the sample average of the areal average temperature anomalies, averages of individual point maxima, and the percent area exceeding 1 K in the 4° × 4° box along with the 95% significance level. From these tables a general pattern emerges where the area-averaged temperature anomalies and the percent area exceeding 1 K near the center of the OCCs demonstrate an upward trend with respect to the observed T numbers, which indicates that most OCCs indeed have positive temperature anomalies at troposphere upper levels in these data. These results also imply that OCCs with larger T numbers typically have larger area-averaged temperature anomaly values than those with smaller T numbers. Also if OCCs have the same T number (e.g., 0.0), the average OCCTS temperature anomalies are larger than those of OCCL. In contrast, however, the averages of the individual maximum temperature anomalies associated with all types of OCCs have nearly constant values for the whole range of T numbers. To summarize, it appears that both the area-averaged (4° × 4°) temperature anomaly and the percent of this same area exceeding 1 K contain information relevant to the intensity evolution of OCCs.

6. Time series of air temperature anomalies at upper levels for each case of OCCs

Time series graphs of the upper-level temperature anomalies were examined for each OCC case. The objective was to understand the tendency of temperature anomalies calculated from AMSU data and to find the threshold value of the anomalies that distinguish developing OCCs (i.e., into TSs) from nondeveloping OCCs.

Figure 7a shows a time series of the 200-hPa temperature anomalies averaged within a rectangle of 4° latitude by 4° longitude centered on the CSC for all cases of OCCL. For all cases except one, these lines are located below 0.9 K. In the time series of 250 hPa, another one OCCL case had temperature anomalies exceeding 0.9 K during their lifetime (not shown in the figure). Similar case was also found in the time series of 300 hPa (not shown in the figure). In total, only three OCCL cases had temperature anomalies more than 0.9 K at each three levels from the detection of CSC to the dissipation, which means that the temperature anomalies did not reach 0.9 K in the other 50 OCCL cases. Figures 7b,c show similar time series graphs of the temperature anomalies at 200 hPa for all cases of OCCTD and OCCTS. In most OCCTS cases, temperature anomalies become larger than 0.9 K at least once in their lifetime. For the OCCTD cases, almost half have anomalies exceed 0.9 K at some time. From these results, we can consider that OCCs establish a distinct warm core structure at their upper levels when their 200-, 250-, or 300-hPa temperature anomalies (averaged within a rectangle of 4° latitude by 4° longitude centered on the CSC) exceed 0.9 K on one or more occasions in their lifetime. For this reason, WCT is used to refer to the situation when the area-averaged temperature anomaly exceeds 0.9 K for the remainder of this paper.

Table 3 shows the number of OCCs classified by this definition, indicating that a total of 25 out of 28 OCCTS have WCT structures. On the other hand, 50 out of 53 OCCL do not have a WCT. It is revealed that 89% of OCCTS have a WCT at least once in their lifetime, while 94% of OCCL do not. It is also clear that while 71% (25 out of 35) of OCCs with a WCT structure developed to TS level, 85% (50 out of 59) of those with no WCT stayed at the stage of L and dissipated. Therefore, it appears that the threshold-based definition of a WCT in the time series of upper-level temperature anomalies associated with OCCs is useful in distinguishing whether an OCC will develop into a TS.

7. Duration from the detection of CSC to each stage in OCCs

For the 25 OCCTS cases with a WCT, the average time from the detection of the CSC to evolution into a TS was 51.1 h in 2004 (Fig. 8a), which means that the OCCTS takes approximately 2 days to become a TS from the status of low pressure area, and an average of 19.4 h to reach the stage of T1 for the first time from its appearance as an OCC. Meanwhile, the durations of 50 cases of OCCL with no WCT from detection as an OCC to dissipation averaged only 14.0 h (Fig. 8b), which indicates that an OCCL with no WCT structure usually dissipates in approximately half a day.

On the other hand, for OCCTS cases with a WCT, the average time from OCC detection to the first observed WCT in AMSU observations is 23.4 h (Fig. 8c). This duration roughly corresponds to the period from the appearance of an OCC to the first classification of T1. Incidentally, for the two OCCL cases with WCT structures the average time from OCC detection to the WCT structure was only 3.7 h, and the average lifetime was 8 h.

8. Discussion

Most previous studies on the structure and environmental field of tropical cyclones in the genesis stage in the western North Pacific have only examined cases of cloud clusters that developed to TS status (Briegel and Frank 1997; Ritchie and Holland 1999; Dickinson and Molinari 2002; Cheung 2004; Fu et al. 2007), because there was no official record for those that did not develop into TSs. Although such studies found important characteristics in the formation of tropical cyclones, most of these were, by construction, necessary conditions. To reveal sufficient conditions, cloud clusters that did not develop into TSs must be analyzed. Exceptions to these previous studies include works by CSU researchers published in the 1980s and 1990s (McBride and Zehr 1981; Lee 1989; Zehr 1992). In these papers, the authors used data obtained from soundings (McBride and Zehr 1981; Lee 1989), objective analysis (Zehr 1992), and satellite infrared imagery (Zehr 1992) to investigate a large number of developing and nondeveloping cloud clusters in the western North Pacific. Our study also examines developing and nondeveloping cloud clusters by using the independent and operational (at MSC) EDA. The EDA classifications include not only clusters with intensity larger than that of TDs but also tiny clusters at the start of rotation, which is a distinctive feature of our study. The study also uses AMSU data representing the newest microwave sounding observations from low-earth-orbit satellites, which can detect warm core structures in cloud clusters explicitly. This is another advantage of the present paper compared with the earlier studies conducted at CSU that used conventional soundings, objective analyses, and infrared imagery.

From Table 2, it was found that the average upper-level temperature anomalies near the center of OCCTS increase along with their T numbers. In contrast, the composite average maximum temperature anomalies (i.e., point values) at the upper levels, within the 10° × 10° analysis domain, of OCCTS exhibit almost constant values. It is apparent that the average maximum temperature anomaly for all T numbers has an almost constant value because the total amount of latent heat released from each convective cloud system included in one grid point (i.e., 0.2° latitude by 0.2° longitude) has an upper limit. According to Bretherton and Smolarkiewicz (1989), it was confirmed from their numerical experiments that the gravity wave horizontally spreads out the latent heat in convective clouds, and the heating by precipitation in the cloud has an upper threshold. Meanwhile, the average temperature anomalies near the center of OCCs become larger with increased T numbers. This phenomenon depends on the enlargement of the convective area within OCCs accompanying the increase of T numbers and is likely also an indication of improved vortex efficiency (i.e., increasing inertial stability) with respect to latent heating as discussed in Schubert and Hack (1982). This phenomenon is also apparent from the observed variation of the percent of temperature anomalies exceeding 1 K near the CSC of the OCCs.

The percentage of grid points with 200-hPa temperature anomalies of more than 1 K in OCCL cases with T-number values of 0.0 is 10.6%. For OCCTS cases with T-number values of 0.0, the percentage at 200 hPa is 20.4%. It can be deduced that the percentage difference between OCCL and OCCTS with a T number of 0.0 derives from a balanced temperature in response to the accumulated latent heat in convective clouds over the past several hours and days. Reflecting on the difference between the percentages of grid points of anomalies more than 1 K in the two types of OCCs, the average temperature anomalies of OCCTS with a T number of 0.0 is 0.39 K, which is more than 3 times that of OCCL.

According to Fig. 8, the average lifetime of OCCL is 14.0 h, and the averaging includes their own genesis stage and weakening stage. On the other hand, the average time period from CSC to first T1 for OCCTS is 19.4 h. Thus, most OCCTS with a T number of 0.0 are also in their genesis phase and developing phase. To compare strictly the two types of OCCs in the same life stage such as their genesis and to find out the warm core structure of incipient OCCs, similar parameters in Table 2 were computed for OCCL and OCCTS at the first 6-h period after the first CSC. From these statistics it is found that the temperature anomalies at 200 hPa of OCCL (OCCTS) averaged within a rectangle of 4° latitude by 4° longitude centered on the CSCs is 0.06 K (0.31 K). The average of the maximum temperature anomalies at 200 hPa of OCCL (OCCTS) within a rectangle of 10° latitude by 10° longitude centered on the CSC is 2.08 K (2.16 K). The percentage of grid points where temperature anomalies exceed 1 K within the rectangular area bounded by 4° latitude by 4° longitude and centered on the CSC at 200 hPa for OCCL (OCCTS) is 10.0% (16.7%). These tendencies of the parameters are quite similar to the results for OCCL and OCCTS with a T number of 0.0 in all stages as described in Table 2. This fact means that there is a clear difference between the warm core structure of developing and nondeveloping OCCs right after their formation.

In EDA, OCCs need to satisfy the five conditions described in section 2 to be diagnosed as a T number of 1.0. Both OCCL and OCCTS with a T number of 0.0 did not meet the requirements, but there is a distinct difference between their convective cloud system areas especially in their inner thermal structure. Figure 9 shows the percentages of each of the five conditions (1–5) used in T1 diagnosis for all observed OCCL and OCCTS cases with T-number values of 0.0. As mentioned already, the T1 diagnosis focuses on the duration of the cloud cluster or CSC (condition 1 or 3), the accuracy of CSC estimation (condition 2), and the appearance of the clusters (condition 4 or 5). From this bar graph, it is found that there is no difference in T1 diagnosis results between OCCL and OCCTS with T numbers of 0.0. The same features are found in the percentages of each of the five conditions used in T1 diagnosis for OCCL and OCCTS at the first 6-h period after the first CSC (not shown). Figure 9 and Table 2 show the limit of visible and infrared technique to estimate the OCCs intensity and the importance of the observed inner thermal structures. The warm core structure is a result of other forcing mechanism that develops into OCCs and TSs. From the analyses for the incipient OCCs, it is inferred that OCCTS already have the mechanism to develop into a TS at their formation. Because the objectives of this paper are to describe the warm core structure of OCCs and to reveal the inner thermal difference between developing and nondeveloping OCCs using AMSU, the reason why OCCTS have the stronger warm core structure than OCCL at their genesis and what determines if an OCC will survive for a longer period as OCCTS are discussed continuously as a future issue.

The maximum temperature anomalies of OCCTS with a T number of 1.0, which are significant at the 95% level, increase along with the height. The maxima are 1.27 K at 300 hPa, 1.54 K at 250 hPa, and 2.19 K at 200 hPa, and the same tendencies are also found from all classified OCCTS cases with T numbers over the range of 0.0–2.0 (Table 2). In previous observational studies, the heights of peak warm core anomalies vary from paper to paper and differ from tropical cyclone to tropical cyclone. By way of example, the level of peak of the warm core in the intense Hurricane Hilda (1964) was estimated as 250 hPa (Hawkins and Rubsame 1968). On the other hand, the warming of Hurricane Erin (2001) was greatest at 500 hPa (Halverson et al. 2006). In Knaff et al. (2004), it was found statistically that the peak of the warm core was located at the upper levels and descended as the vertical wind shear increased for 186 tropical cyclone cases. They used air temperature information retrieved from AMSU brightness temperature data archived at CIRA, with the same retrieval method as in this paper. The heights of warm core anomaly maximums are consistent between Knaff et al. (2004) and this paper because the same kinds of data and retrieval method were used; however, there is a discrepancy between the height of the peak of 200 hPa in this paper (relatively high in the atmosphere) and that in other observational papers (lower in the atmosphere). The air temperatures at lower levels retrieved from AMSU data are likely cooler than in reality, especially near the center of OCCs because of the absorption and scattering of microwave radiation by liquid water and ice particles. This attenuation effect of microwaves could have resulted in lower maximum temperature anomalies at 300 hPa than those at 200 hPa. The exact reasons of these discrepancies, however, remain unknown.

Zehr (1992) found that developing cloud clusters (into TSs) experience at least a few distinct increases of convection or convective bursts, typically one event near the time of initial classification as a tropical cyclone and another event between genesis and designation as a TS. Convective bursts in Zehr (1992) were determined by a peak of cold infrared brightness temperature before appearance as a TS. Once a cluster experiences a convective burst, it typically takes 24 h before the first classification as a tropical cyclone. In our study, temperature anomalies near the center of clusters at upper levels sometimes reach the peak before developing into tropical cyclones, as described in Fig. 5b. This appearance of a WCT structure within clusters at upper levels agrees with the convective burst concepts presented in Zehr (1992). The convective burst is considered as the first occurrence of T-number 1 in EDA results and has an average timing of 19.4 h for the OCCTS with a WCT in the EDA files. The timing is almost the same as that of the first detection of a WCT in OCCTS at an average of 23.4 h. A possible scenario behind this is as follows: when the cluster has a convective burst in the infrared imagery, it is recognized as T number 1. After a few hours, the convective burst provides large amounts of latent heat to the upper troposphere near the center of the cluster, and the WCT structure is observed by AMSU. This hypothesis must be proven by further studies.

It is well known that tropical convections show significant diurnal variations (e.g., Gray and Jacobson 1977; Nitta and Sekine 1994; Chen and Houze 1997). Muramatsu (1983) found that areal coverage of deep cumulus convection in mature typhoon also has a diurnal variation with peak in the local morning. To investigate the diurnal variation of the warm core structure in OCCs, the temperature anomaly data of AMSU were divided into four categories according to their local observational time because of the usage of two NOAA satellites, which take almost same local observational time twice a day. The local observational time of AMSU from 0100 to 0400 LST is categorized as 0100 LST. In a similar way, the local time from 0500 (1300 or 1700) to 0800 (1600 or 2000) LST is categorized as 0500 (1300 or 1700) LST. Figure 10 shows diurnal variation of 200-hPa temperature anomalies averaged at each AMSU local observational time for all OCCs classified by each T number from 0.0 to 2.0. From this figure, it is found that the amplitude of the temperature anomalies becomes larger with T number. And for all T numbers, the maximum temperature anomalies are located at 0500 LST, which corresponds to the morning peak of deep convective activity in typhoons. Because the local observational time is divided into only four categories, it is difficult to do the further examination of diurnal variation of warm core structure in OCCs. But it can be estimated that there is a distinct diurnal variation of the warm core structure. In this paper, the diurnal cycle could not be excluded from the time series analysis shown in Fig. 7. It is considered that the time series were somewhat affected by the diurnal variation of the warm core structure. For example, the difference between the maximum and the minimum temperature anomalies is about 0.75 K for a T number of 2.0 and 0.5 K for a T number of 1.5. And among the 25 cases of OCCTS with WCTs, 8 cases have anomalies exceeding 0.9 K at the observational time of 0500 LST. It is recommended that the temperature anomalies more than 0.9 K observed in the morning are less emphasized.

As outlined in section 7, the average period from the detection of the CSC of an OCCTS to development into a TS is 51.1 h (Fig. 8a), and the period from the detection of the CSC in 25 cases of OCCTS to the first retrieval of the WCT is 23.4 h (Fig. 8c). If the WCT retrieved from AMSU data is regarded as a sign of TS genesis, the lead time to its genesis is 27.7 h, which indicates that the detection of WCTs in OCCs is a very useful tool for finding and forecasting TS genesis.

In contrast, there were only three cases of OCCTS in which no WCT structure was observed in the lifetime from detection of the CSC to evolution into TS, while only three cases in which OCCL did have WCT structures. Two of the three OCCTS cases with no WCT structure did in fact have large warm anomalies located near the rectangle of 10° latitude by 10° longitude centered on the CSC. For OCCTD, the total number of cases is small, but half of them had WCTs and half did not. These exceptions mean that 30% of OCCs with WCT structures did not develop into TSs, while only 5% of OCCs without a WCT did develop to TS status. To improve the performance of determining whether OCCs will develop into TSs, it is necessary to add another screening method. One possible solution is to use information from EDA concerning the lifetime of an OCC from CSC detection to dissipation as L or to evolution into a TS. The average lifetime of an OCCL with no WCT is 14.0 h, and the average time from the detection of CSC to the evolution into a TS of an OCCTS is 51.1 h. From the difference in these durations, it is expected that an OCCL can be distinguished from an OCCTS by combination of EDA and AMSU warm core detection. Moreover, it is expected that there is an environmental difference between OCCL and OCCTS such as low-level vorticity, low-level convergence, upper-level divergence, etc. As a future work, we would like to try to include this information related to the OCCs’ environment retrieved from other microwave sensors such as imagers and scatterometers, numerical model analyses, and forecast results in this screening algorithm.

From the previous discussion, it can be concluded that there is a distinct possibility of improving the detection and forecasting the formation of TSs using AMSU observations. A more detailed analysis, however, is needed to develop an objective method of detecting and forecasting TS genesis based upon the observation of warm cores from AMSU data and the monitoring of OCC durations. EDA operation was transferred from MSC to the Regional Specialized Meteorological Center (RSMC) Tokyo-Typhoon Center operated by JMA in 2007. Recently RSMC Tokyo has been running a follow-on project based upon the research outlined in this paper. Using the AMSU data available in near-real time at MSC, RSMC Tokyo will validate the method proposed in this paper as well as try to use other parameters retrieved from the raw AMSU channel-7 brightness temperatures with the cooperation of Meteorological Research Institute to better infer similar information. This operational test bed study is performed using operational EDA. The progress of that research will be presented in forthcoming reports.

9. Summary

Because of the subjective nature of the Dvorak analysis technique, it is necessary to develop new objective methods of estimating the intensity of OCCs and detecting the formation of TSs. To this end, we first applied AMSU observational data to the analysis of temperature profiles within developing and nondeveloping (into TSs) OCCs over the western North Pacific in 2004. The results of EDA, performed by JMA/MSC, were introduced to distinguish OCCs developing from those not developing into TSs in this process. Two typical cases of OCCTS and OCCL were shown to compare their inner temperature profiles at the upper levels. The OCCTS had a wider positive anomaly region than OCCL.

As the next step, the temperature anomalies at 200, 250, and 300 hPa near the center of each classified OCC were averaged and examined. For the OCCTS, the average positive anomalies increased along with the T number estimated using EDA. This tendency was also found in the relationship between the percentages of grid points with anomalies exceeding 1 K near the centers of the OCCTS. And the average anomalies and the percentage showed peak values at 200 hPa for each T number of 0.0–2.0. On the other hand, the average temperature anomaly maxima of the OCC anomalies showed almost constant values for each T number.

In time series analysis of the temperature anomalies, most OCCTS showed a value of larger than 0.9 K at least once, whereas the anomalies for most OCCL cases were below 0.9 K throughout their lifetime. From these analyses, we defined warm core structures showing the temperature anomaly greater than a threshold temperature (WCT) within OCCs using a temperature anomaly exceeding 0.9 K in this paper.

Using this definition, WCTs were found in most OCCTS, but not in most OCCL, and 70% of the OCCs with WCTs developed into TSs, while 85% of those with no WCT dissipated without such development. For OCCTS with a WCT, the time from the observation of the WCT by AMSU to classification as a TS was 27.7 h, meaning that there is a lead time of almost 1 day in detecting the genesis of TSs. From these results, we can conclude that there is a strong possibility of improving forecasting cyclogenesis through the use of AMSU observations.

Acknowledgments

First, the authors would like to thank an editor and anonymous reviewers for their fruitful comments on this manuscript. We are greatly indebted to our colleagues at the MRI, RSMC Tokyo-Typhoon Center, and MSC for their invaluable discussions, especially Masaaki Togashi, who was a director of the typhoon research department at MRI and passed away in August 2008. We also wish to thank Masanori Yoshizaki and his colleagues at the Institute of Observational Research for Global Change in Japan Agency for Marine-Earth Science and Technology for their helpful suggestions. Mark DeMaria of CSU/CIRA gave us permission to use the AMSU dataset and the DDK algorithm. His colleagues, John Knaff and Ray Zehr provided constructive comments. We extend special thanks to Mai Miyauchi for her dedicated efforts to arrange the dataset.

REFERENCES

  • Asano, J., and Coauthors, 2008: Analysis of tropical cyclones using microwave satellite imagery. RSMC Tokyo-Typhoon Center Tech. Rev., 10 , 3070.

    • Search Google Scholar
    • Export Citation
  • Barnes, S. L., 1964: A technique for maximizing details in numerical weather map analysis. J. Appl. Meteor., 3 , 396409.

  • Bessho, K., , M. DeMaria, , and J. A. Knaff, 2006: Tropical cyclone wind retrievals from the Advanced Microwave Sounding Unit (AMSU): Application to surface wind analysis. J. Appl. Meteor. Climatol., 45 , 399415.

    • Search Google Scholar
    • Export Citation
  • Bretherton, C. S., , and P. K. Smolarkiewicz, 1989: Gravity waves, compensating subsidence, and detrainment around cumulus clouds. J. Atmos. Sci., 46 , 740759.

    • Search Google Scholar
    • Export Citation
  • Briegel, L. M., , and W. M. Frank, 1997: Large-scale influences on tropical cyclogenesis in the western North Pacific. Mon. Wea. Rev., 125 , 13971413.

    • Search Google Scholar
    • Export Citation
  • Brueske, K. F., , and C. S. Velden, 2003: Satellite-based tropical cyclone intensity estimation using the NOAA-KLM series Advanced Microwave Sounding Unit (AMSU). Mon. Wea. Rev., 131 , 687697.

    • Search Google Scholar
    • Export Citation
  • Chen, S. S., , and R. A. Houze Jr., 1997: Diurnal variation and life-cycle of deep convective systems over the tropical Pacific warm pool. Quart. J. Roy. Meteor. Soc., 123 , 357388.

    • Search Google Scholar
    • Export Citation
  • Cheung, K. K. W., 2004: Large-scale environmental parameters associated with tropical cyclone formations in the western North Pacific. J. Climate, 17 , 466484.

    • Search Google Scholar
    • Export Citation
  • Demuth, J. L., , M. DeMaria, , J. A. Knaff, , and T. H. Vonder Haar, 2004: Evaluation of Advanced Microwave Sounding Unit tropical-cyclone intensity and size estimation algorithms. J. Appl. Meteor., 43 , 282296.

    • Search Google Scholar
    • Export Citation
  • Demuth, J. L., , M. DeMaria, , and J. A. Knaff, 2006: Improvement of Advanced Microwave Sounding Unit tropical cyclone intensity and size estimation algorithms. J. Appl. Meteor. Climatol., 45 , 15731581.

    • Search Google Scholar
    • Export Citation
  • Dickinson, M., , and J. Molinari, 2002: Mixed Rossby–gravity waves and western Pacific tropical cyclogenesis. Part I: Synoptic evolution. J. Atmos. Sci., 59 , 21832196.

    • Search Google Scholar
    • Export Citation
  • Dvorak, V. F., 1975: Tropical cyclone intensity analysis and forecasting from satellite imagery. Mon. Wea. Rev., 103 , 420430.

  • Dvorak, V. F., 1984: Tropical cyclone intensity analysis using satellite data. NOAA Tech. Rep. NESDIS 11, Washington, DC, 47 pp.

  • Fu, B., , T. Li, , M. S. Peng, , and F. Weng, 2007: Analysis of tropical cyclogenesis in the western North Pacific for 2000 and 2001. Wea. Forecasting, 22 , 763780.

    • Search Google Scholar
    • Export Citation
  • Gierach, M. M., , M. A. Bourassa, , P. Cunningham, , J. J. O’Brien, , and P. D. Reasor, 2007: Vorticity-based detection of tropical cyclogenesis. J. Appl. Meteor. Climatol., 46 , 12141229.

    • Search Google Scholar
    • Export Citation
  • Gray, W. M., , and R. W. Jacobson Jr., 1977: Diurnal variation of deep cumulus convection. Mon. Wea. Rev., 105 , 11711188.

  • Halverson, J. B., , J. Simpson, , G. Heymsfield, , H. Pierce, , T. Hock, , and L. Ritchie, 2006: Warm core structure of Hurricane Erin diagnosed from high altitude dropsondes during CAMEX-4. J. Atmos. Sci., 63 , 309324.

    • Search Google Scholar
    • Export Citation
  • Hawkins, H. F., , and D. T. Rubsame, 1968: Hurricane Hilda, 1964. II. Structure and budgets of the hurricane on October 1, 1964. Mon. Wea. Rev., 96 , 617636.

    • Search Google Scholar
    • Export Citation
  • Hawkins, H. F., , and S. M. Imbembo, 1976: The structure of a small, intense hurricane—Inez 1966. Mon. Wea. Rev., 104 , 418442.

  • Hawkins, J. D., , T. F. Lee, , J. Turk, , C. Sampson, , J. Kent, , and K. Richardson, 2001: Real-time Internet distribution of satellite products for tropical cyclone reconnaissance. Bull. Amer. Meteor. Soc., 82 , 567578.

    • Search Google Scholar
    • Export Citation
  • Heymsfield, G. M., , J. B. Halverson, , J. Simpson, , L. Tian, , and T. P. Bui, 2001: ER-2 Doppler radar investigations of the eyewall of Hurricane Bonnie during the Convection and Moisture Experiment-3. J. Appl. Meteor., 40 , 13101330.

    • Search Google Scholar
    • Export Citation
  • Hoshino, S., , and T. Nakazawa, 2007: Estimation of tropical cyclone’s intensity using TRMM/TMI brightness temperature data. J. Meteor. Soc. Japan, 85 , 437454.

    • Search Google Scholar
    • Export Citation
  • Katsaros, K. B., , E. B. Forde, , P. Chang, , and W. T. Liu, 2001: QuikSCAT’s SeaWinds facilitates early identification of tropical depressions in 1999 hurricane season. Geophys. Res. Lett., 28 , 10431046.

    • Search Google Scholar
    • Export Citation
  • Kidder, S. Q., , M. D. Goldberg, , R. M. Zehr, , M. DeMaria, , J. F. W. Purdom, , C. S. Velden, , N. C. Grody, , and S. J. Kusselson, 2000: Satellite analysis of tropical cyclones using the Advanced Microwave Sounding Unit (AMSU). Bull. Amer. Meteor. Soc., 81 , 12411259.

    • Search Google Scholar
    • Export Citation
  • Kishimoto, K., , T. Nishigaki, , S. Nishimura, , and Y. Terasaka, 2007: Comparative study on organized convective cloud systems detected through early stage Dvorak analysis and tropical cyclones in early developing stage in the western North Pacific and the South China Sea. RSMC Tokyo-Typhoon Center Tech. Rev., 9 , 1932.

    • Search Google Scholar
    • Export Citation
  • Knaff, J. A., , R. M. Zehr, , M. D. Goldberg, , and S. Q. Kidder, 2000: An example of temperature structure differences in two cyclone systems derived from the Advanced Microwave Sounder Unit. Wea. Forecasting, 15 , 476483.

    • Search Google Scholar
    • Export Citation
  • Knaff, J. A., , S. A. Seseske, , M. DeMaria, , and J. L. Demuth, 2004: On the influences of vertical wind shear on symmetric tropical cyclone structure derived from AMSU. Mon. Wea. Rev., 132 , 25032510.

    • Search Google Scholar
    • Export Citation
  • Lee, C. S., 1989: Observational analysis of tropical cyclogenesis in the western North Pacific. Part I: Structural evolution of cloud clusters. J. Atmos. Sci., 46 , 25802598.

    • Search Google Scholar
    • Export Citation
  • Lee, T. F., , F. J. Turk, , J. Hawkins, , and K. Richardson, 2002: Interpretation of TRMM TMI images of tropical cyclones. Earth Interactions, 6 .[Available online at http://EarthInteractions.org].

    • Search Google Scholar
    • Export Citation
  • McBride, J. L., , and R. Zehr, 1981: Observational analysis of tropical cyclone formation. Part II: Comparison of non-developing versus developing systems. J. Atmos. Sci., 38 , 11321151.

    • Search Google Scholar
    • Export Citation
  • Muramatsu, T., 1983: Diurnal variations of satellite-measured TBB areal distribution and eye diameter of mature typhoons. J. Meteor. Soc. Japan, 61 , 7790.

    • Search Google Scholar
    • Export Citation
  • Nitta, T., , and S. Sekine, 1994: Diurnal variation of convective activity over the tropical western Pacific. J. Meteor. Soc. Japan, 72 , 627641.

    • Search Google Scholar
    • Export Citation
  • Olander, T. L., , and C. S. Velden, 2007: The Advanced Dvorak Technique: Continued development of an objective scheme to estimate tropical cyclone intensity using geostationary infrared satellite imagery. Wea. Forecasting, 22 , 287298.

    • Search Google Scholar
    • Export Citation
  • Ritchie, E. A., , and G. J. Holland, 1999: Large-scale patterns associated with tropical cyclogenesis in the western Pacific. Mon. Wea. Rev., 127 , 20272043.

    • Search Google Scholar
    • Export Citation
  • Schubert, W. H., , and J. J. Hack, 1982: Inertial stability and tropical cyclone development. J. Atmos. Sci., 39 , 16871697.

  • Sharp, R. J., , M. A. Bourassa, , and J. J. O’Brien, 2002: Early detection of tropical cyclones using Seawinds-derived vorticity. Bull. Amer. Meteor. Soc., 83 , 879889.

    • Search Google Scholar
    • Export Citation
  • Tsuchiya, A., , T. Mikawa, , and A. Kikuchi, 2001: Method of distinguishing between early stage cloud systems that develop into tropical storms and ones that do not. Geophys. Mag., 4 , 4959.

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., , T. L. Olander, , and R. M. Zehr, 1998: Development of an objective scheme to estimate tropical cyclone intensity from digital geostationary satellite infrared imagery. Wea. Forecasting, 13 , 172186.

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., and Coauthors, 2006: The Dvorak tropical cyclone intensity estimation technique: A satellite-based method that has endured for over 30 years. Bull. Amer. Meteor. Soc., 87 , 11951210.

    • Search Google Scholar
    • Export Citation
  • Zehr, R. M., 1992: Tropical cyclogenesis in the Western North Pacific. NOAA Tech. Rep. NESDIS 61, Washington, DC, 181 pp.

APPENDIX

Characteristics of OCCs in 2004 EDA File

Table A1 is a list of OCCs in the 2004 EDA file. Each entry contains system number, final stage information, latitude and longitude of starting point, starting time, time of first T1 classification as OCCTD or OCCTS, dissipation time for OCCL, gale warning issuance time for OCCTD or OCCTS, time of first AMSU-observed WCT structure in OCCs, and corresponding JMA typhoon number and name for OCCTS. OCCs numbered as 0423, 0448, 0494, and 0495 were unassigned in this list after the postanalysis.

Fig. 1.
Fig. 1.

The four cloud patterns (i–iv) that define CSCs (from Tsuchiya et al. 2001).

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 2.
Fig. 2.

A conceptual model of cloud clusters satisfying the T1 classification. The shaded areas are dense, cold (≤−31°C) overcast. The estimation accuracy of the CSC is expressed by the dotted circle with a diameter of 2.5° (condition 2). The dashed circle with a radius of 2.0° indicates the region that includes the dense, cold overcast (condition 4). The diameter of the dashed circle drawn in the overcast on the right side of the CSC is 1.5°. This circle shows the size of the overcast (condition 5; from Tsuchiya et al. 2001).

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 3.
Fig. 3.

Horizontal images of (right) AMSU-retrieved temperature anomalies (K) at 200 hPa and (left) GOES-9 infrared brightness temperatures for (a) EDA0428 at 2100 UTC 3 Jun and (b) EDA0453 at 2100 UTC 13 Aug. The cross hairs show the position of the CSC. Shading shows positive temperature anomalies, and a darker shade means a larger anomaly.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 4.
Fig. 4.

Vertical cross sections showing AMSU-retrieved temperature anomalies (K) along an east–west line through the CSC at the same observation time as Fig. 3 for (a) EDA0428 and (b) EDA0453. Shading is as in Fig. 3.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 5.
Fig. 5.

Pressure vs time cross sections of temperature anomalies (K) averaged within a 4° latitude by 4° longitude rectangle centered on the CSCs. Temperature anomalies are retrieved from AMSU observations in the cases of (a) EDA0428 and (b) EDA0453. Shading is the same as in Fig. 3.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 6.
Fig. 6.

Positions of observed AMSU-based maximum temperature anomalies (K) at 200 hPa for OCCTS within a 10° latitude by 10° longitude rectangle centered on the CSC. Results are associated with T-number values of (a) 0.0, (b) 1.0, (c) 1.5, and (d) 2.0.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 7.
Fig. 7.

Time series graphs of 200-hPa temperature anomalies (K) averaged within a 4° latitude by 4° longitude rectangle centered on the CSC. Values are shown for all cases of (a) OCCL, (b) OCCTD, and (c) OCCTS from the detection of CSCs to dissipation or to recognition as TSs. The red (black) lines show the cases with (without) anomalies larger than 0.9 K.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 8.
Fig. 8.

Average lifetime of (a) the 25 OCCTS cases with a WCT and the first judgment of T1, (b) the 50 OCCL cases with no WCT, and (c) the 25 OCCTS cases with a WCT and the first identification of a WCT in an AMSU observation. In the diagram, CSC means the detection of the CSC, and TS refers to the evolution to TS status.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 9.
Fig. 9.

Percentages of each of the five conditions (1–5) used in T1 diagnosis for all observational cases of OCCL (black bar) and OCCTS (white bar) with T numbers of 0.0.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Fig. 10.
Fig. 10.

Diurnal variations of 200-hPa temperature anomalies (K) averaged at each AMSU local observational time for all OCCs classified by each T number from 0.0 to 2.0.

Citation: Monthly Weather Review 138, 7; 10.1175/2010MWR3073.1

Table 1.

Tropical cyclone classifications used at JMA. MWS assumed to be 10-min averages.

Table 1.
Table 2.

Statistics of air temperature anomalies retrieved from AMSU within OCCs according to final stage classification and the T number at 200, 250, and 300 hPa. Numbers in parentheses indicate the standard deviation. An asterisk denotes statistical significance at the 95% level.

Table 2.
Table 3.

A list of the kinds of OCCs along with the number of cases with or without WCT structures.

Table 3.
Table A1.

List of OCCs in the 2004 EDA file. Each entry contains system number, final stage information, latitude and longitude of starting point, starting time, time of first T1 classification as OCCTD or OCCTS, dissipation time for OCCL, gale warning issuance time for OCCTD or OCCTS, time of first AMSU-observed WCT structure in OCCs, and corresponding JMA typhoon number and name for OCCTS. OCCs numbered as 0423, 0448, 0494 and 0495 were unassigned in this list after the postanalysis.

Table A1.
Save