2.1. Typhoon, Ocean and Atmospheric Datasets
Hato’s track data are based on the best warning track from the collaboration site of the JTWC (
https://pzal.ndbc.noaa.gov/collab/login.php) that provides the position, intensity and radius of maximum wind (RMW). More importantly, unlike the regular 6-hourly best-track archive, it provides higher temporal (i.e., 3-hourly) information during the critical landfalling period. Based on the JTWC track data, Hato suddenly further intensified during the last 3 h (00Z–03Z 23 August) before making landfall in Macau, although it had been undergoing a rapid intensification after entering the South China Sea (SCS). More strikingly, the intensification rate in the last 3 h prior to landfall doubled, as compared to the previous 24 h (
Figure 2). In this study, the intensification rate and translation speed of Hato are computed based on two adjacent track points.
High-quality SST observations along the coastal area of Macau and over the SCS are required to examine the ocean environment for the intensification of Hato. In this study, a fine-resolution (4 km) daily SST dataset from the Moderate Resolution Imaging Spectroradiometer (MODIS;
https://oceandata.sci.gsfc.nasa.gov/MODIS-Aqua/Mapped/) was used to characterize the ocean surface conditions prior to Hato’s passage. The advantage of using infrared SST data is that it is less affected by land, which is more suitable for the purpose of the present study. Since the infrared SST is often obscured by clouds, an 8-day composite product was used to reduce the missing data rate (
Figure 3a).
Due to the lack of ocean vertical temperature observation around Hato’s passage, 0.25° × 0.25° monthly temperature climatology from South China Sea Physical Oceanographic Dataset (SCSPOD14) [
32] was used to reconstruct the temperature profiles. Produced by the South China Sea Institute of Oceanology (SCSIO), SCSPOD14 is a new climatology which was specifically built for the SCS. It makes use of all available in situ temperature profiles during the past century (i.e., 1919-2014), including not only data from the comprehensive US World Ocean Database (WOD) and international Argo project, but also exclusive measurements from SCSIO [
32]. Therefore, SCSPOD14 is deemed to be one of the best climatological datasets for the SCS.
To reconstruct the ocean subsurface thermal structure, the mixed layer depth (MLD) and underneath vertical temperature gradient (i.e., dT/dz; hereafter called temperature gradient) are extracted from the SCSPOD14 climatology. It should be noted that MLD here refers to the isothermal layer depth with 0.3 °C threshold criterion [
33]. The along typhoon track temperature profiles prior to Hato’s passage were generated by a combination of the SCSPOD14 and the corresponding MODIS observed SST (
Figure 3a). Such temperature profiles are herein referred to as synthetic profiles, and will be input into an ocean mixing scheme (introduced in the next section) to investigate the effect of shallow water on Hato’s rapid intensification.
For air-sea heat flux calculations, atmospheric conditions from the daily global reanalysis of ERA-interim [
34] from the European Centre for Medium-Range Weather Forecasts (ECMWF) was employed. The ERA-interim dataset used is on a 6-hourly basis with 0.25° spatial resolution. Near-surface air and dew-point temperatures were extracted for heat flux calculations.
In addition, to represent the most realistic atmospheric condition for Hato’s landfalling period, the station observations of air and dew-point temperatures at the Macau Meteorological and Geophysical Bureau were used instead of the ERA-interim at the landfalling point.
2.2. Numerical Models
To investigate the effect of suppressed SST cooling in shallow water on Hato’s rapid intensification, we will examine SST cooling response and associated heat fluxes at four selected track points, i.e., P0, P1, P2 and P3, as shown in
Figure 1. These points are chosen to represent different water depths and typhoon intensities as Hato approached the city of Macau. As listed in
Table 1, P3 is the location of Hato just before landfall at 03Z 23 August, with a peak intensity of 100 kt and shallowest water depth of 27 m; P2 is at 00Z 23 August with an intensity of 90 kt and water depth of 52 m; P1 is at 12Z 22 August with an intensity of 70 kt and water depth of 625 m; and P0 is at 06Z 22 August with an intensity of 60 kt and water depth of 2845 m. The water depths along Hato’s track over the SCS are depicted in
Figure 1b.
In this study, the critical SST values under Hato will be estimated from mixing depth by the ocean mixing scheme proposed by Price [
28], which was developed based on the three-dimensional Price-Weller-Pinkel (3DPWP) [
35] ocean model:
where
is the acceleration due to gravity,
the density difference at the base of the mixed layer,
the mixed layer thickness,
the density of sea water taken as 1024 kg m
−3,
and
are the maximum wind radius and translation speed of the typhoon, and
is the scale parameter taken as 1.3, as proposed by Price [
28] to account for the earth rotation effect; finally
is the wind stress induced by the typhoon, computed via:
where
is the density of air taken as 1.2 kg m
−3,
the drag coefficient, which is based on Powell et al. [
36] accounting for high-wind condition, and
is the maximum surface wind, which directly adopts the surface wind speed from JTWC. Hereafter, this scheme is referred to as the Price 2009 model.
Basically, this model is intended to find the thickness (or depth,
) from the sea surface until the left-hand side of the Equation (1) is equal or greater than the bulk Richardson number of 0.65 [
37]. The Price 2009 model is an efficient way to estimate typhoon mixing depth, i.e.,
, by a given typhoon’s general characteristics (i.e.,
V,
R and
Uh) and ocean stratification [
19,
38]. The SST cooling generated by the typhoon was obtained by vertically averaging the temperatures within the mixing depth
. In this study, the corresponding ocean temperature profiles input to the Price 2009 model were reconstructed by the combination of MODIS observations and SCSPOD14 climatological ocean thermal structure, as described in
Section 2.1.
Figure 4 shows the monthly variability of the MLD and temperature gradient at four selected locations based on the SCSPOD14. It can be seen that the MLD and temperature gradient vary considerably between summer (May to September) and winter (October to March), in particular at the shallow water area (P2 and P3); this is probably due to enhanced surface heating in summer and strong wind mixing in winter. In this study, to account for the possible variability of the MLD and temperature gradient, the annual mean values were used to reconstruct the temperature profiles for the Price 2009 model experiments.
Figure 5 shows the synthetic profiles for the four selected points prior to Hato’s passage. It is noteworthy that at shallow water locations P2 and P3, the synthetic temperature profiles extend until they reach the bottom depth (blue and red curves in
Figure 5). For all the Price 2009 model experiments, salinity remains at a constant of 33 psu throughout the water column.
In addition to the Price 2009 model, a sophisticated full ocean model, the Luzon Strait Nowcast/Forecast System (LZSNFS) [
39,
40,
41,
42,
43,
44,
45], developed by the US Naval Research Laboratory (NRL), was used to examine the ocean response to typhoon Hato. The LZSNFS model deals with all three-dimensional ocean processes, including coastal and bottom boundary layer effects that the Price 2009 model does not account for. Therefore, it would provide more thorough insights into the SST cooling in the shallow water area. Furthermore, to explore the impact of shallow water on Hato’s rapid intensification, the advanced hurricane version of the Weather Research and Forecasting (WRF) model, hereafter AHW, was employed to perform a series of numerical experiments on Hato’s intensity. The descriptions of the LZSNFS and AHW models will be introduced later in the corresponding sections.
2.3. Air-Sea Enthalpy Flux
Based on the under-typhoon SST estimated from the Price 2009 model, the corresponding sensible (
) and latent (
) heat fluxes, or total heat fluxes, can be calculated through the bulk aerodynamic formulas [
21,
46,
47]:
where
and
are the exchange coefficients for sensible and latent heat, both taken as 1.3 × 10
−3 based on measurements under high wind condition [
48],
and
the SST and near-surface air temperature,
and
the surface and air specific humidities calculated based on
,
and dew-point temperature, and
and
the air heat capacity and latent heat of vaporization, respectively. Throughout this paper, heat fluxes that are transferring from the ocean to the atmosphere are defined as positive.