The Track and Accompanying Sea Wave Forecasts of the Supertyphoon Mangkhut (2018) by a Real-Time Regional Forecast System
1. Introduction
The northwestern Pacific Ocean (NWP) is one of the areas with the most frequent and intense tropical cyclone (TC) activities, and China is one of the most seriously affected countries along the NWP, with about seven or eight TCs making landfall each year (Wang and Qian 2005). Statistically, there are totally 1898 TCs generated in the NWP including the South China Sea (SCS) from 1945 to 2009 (29.2 per year), and among which 597 TCs landed or approached China (Jia et al. 2010). In the summer and autumn of each year, TCs caused enormous property loss of over $10 trillion with an affected population of 250 million in the southeastern coastal areas of China and the surrounding TC-affected area (Wu and Kuo 1999; Liu et al. 2009). Therefore, an accurate forecasting of TC tracks, especially the landfall locations, is very important for the disaster reduction and prevention in the areas of high TC incidence.
Forecasting of TCs dates back to the 1960s. The original forecast technique was developed using statistical methods including the approaches of “averaging across previous cyclones,” “statistical modeling of previous cyclones,” and so on (Neumann 1979; Roy and Rita 2012). As the simplest technique, it uses only the current and historical movement of a TC, which is unable to handle the anomalous motion of TC (Roy and Rita 2012). With the increase of computational resources, the dynamical approach based on the numerical model became an important and promising technique for TC forecast, which estimates the physical state of the atmosphere through numerical approximation of mathematical equations describing the dynamics and physics of weather systems driving TC’s formation and development (Radford 1994; Fulton 2001; Bender et al. 2007; Hong et al. 2015). This model-based approach, when combined with advanced data assimilation technique that helps the numerical model digest efficiently various observations, can gain higher accuracy than the statistical methods.
With the development of more sophisticated numerical models and data assimilation methods, as well as the increase of computational resources and data sources, significant improvements in TC track forecast have been achieved during the past several decades. However, considerable forecast uncertainty still exists. Various factors, such as poor understanding of the physical processes (Fraedrich and Leslie 1989; Plu 2011; Peng et al. 2014), inaccurate initial conditions (Leslie et al. 1998; Bender et al. 2007; Hsiao et al. 2009), incomplete parameterization schemes (Rao and Prasad 2007; Khain and Lynn 2011), as well as their combination (Zhou et al. 2013), contribute to the errors of TC track forecast. For instance, in 2015, the errors of TC track forecasts of 24, 48, and 72-h from the National Hurricane Center (NHC) are 50, 70 and 100 km, respectively, and those from the China Meteorological Administration (CMA) are 60, 120 and 180 km, respectively. The accuracy of the TC track forecast still has a great potential to be improved.
Because of the high impact of TCs, more and more operational agencies or research institutes participate in the TC forecast of the NWP, including the South China Sea Institute of Oceanology (SCSIO). This paper introduces a real-time Regional Forecast System of the SCS Marine Environment (denoted as RFSSME) established in the SCSIO and its performance in the track and accompanying sea waves forecasts of Supertyphoon Mangkhut (2018). In addition, three sensitivity experiments are conducted to investigate the key techniques employed in RFSSME in improving the track forecasting skill for Typhoon Mangkhut (2018). The paper is organized as follows: in section 2, a description of RFSSME and Typhoon Mangkhut (2018) is given. The validation of track and sea wave forecast by RFSSME against observations is presented in section 3. Section 4 demonstrates the results of the sensitivity experiments. Section 5 gives a summary.
2. A brief description of RFSSME as well as Supertyphoon Mangkhut (2018)
RFSSME was established by the State Key Laboratory of Tropical Oceanology (LTO), SCSIO in 2016. With the newly developed forecasting technologies incorporated, RFSSME was developed from an older version of a real-time air–sea–wave forecasting system, which was also known as the Experimental Platform of Marine Environment Forecasting (EPMEF) (Peng et al. 2015). RFSSME is an atmosphere–ocean–wave forecast system, which consists of three main subsystems: the atmosphere forecast (AF), the ocean forecast (OF) and the wave forecast (WF), with their data assimilation schemes incorporated, as schematically shown in Fig. 1. RFSSME was updated from EPMEF in several aspects: the increase of model resolution, extension of the forecast period, data assimilation schemes, and physical parameterization schemes, and so on. The details of each subsystem are introduced as follows.
Flow chart of RFSSME, in which AF, OF, and WF represent the three subsystems of atmosphere forecast, ocean forecast, and wave forecast, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Flow chart of RFSSME, in which AF, OF, and WF represent the three subsystems of atmosphere forecast, ocean forecast, and wave forecast, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Flow chart of RFSSME, in which AF, OF, and WF represent the three subsystems of atmosphere forecast, ocean forecast, and wave forecast, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The AF was developed on the basis of the Weather Research and Forecasting (WRF) Model (version 3.6) with the Advanced Research WRF (ARW) core (Skamarock and Klemp 2008). A two-domain nesting configuration was designed for the WRF Model. The outer domain of WRF Model (WRF-D01) covers the SCS, the western Pacific Ocean and the eastern Indian Ocean (87°–145°E, −14°–40°N) with a horizontal resolution of 54 km (Fig. 2). The SST used in WRF-D01 is the daily real-time global sea surface temperature (RTG_SST) analysis data provided by the National Centers for Environmental Prediction/Marine Modeling and Analysis Branch (NCEP/MMAB). The initial conditions and lateral boundary conditions are from the Global Forecast System (GFS) provided by NCEP with a horizontal resolution of 1° × 1°. The inner domain (WRF-D02) covers the SCS and the NWP (96°–135°E, 0°–30°N) with a horizontal resolution of 18 km (Fig. 2). Both domains have 30 vertical layers. The three-dimensional variational data assimilation system of WRF (WRF3DVAR; Liu and Barker 2006) incorporated with the scale-selective data assimilation (SSDA) scheme (Peng et al. 2010; Xie et al. 2010) was designed for both domains of the AF. The real-time radiance data from NOAA ATOVS instruments in every forecasting cycle are assimilated through the WRF3DVAR, which is helpful to improve the accuracy of TC track forecast by optimizing the TC structure in the initial fields (Liu et al. 2014). The SSDA scheme involves a low-pass filter which is used to perform the scale separation of the initial/forecasting wind fields from the GFS or the WRF Model outputs. The large-scale components of the wind field from the WRF Model are adjusted by those from the GFS outputs which are assimilated through WRF3DVAR with a preset time interval of 24 h, and then the adjusted large-scale components and the original small-scale components of WRF Model outputs are recombined for the next forecasting cycle. The cutoff wavenumber for the low-pass filter in the RFSSME is set to 4. Readers may refer to Lai et al. (2014) for more details of the SSDA scheme.
Model domains of subsystems in RFSSME, including the outer domain of WRF (WRF-D01), the inner domain of WRF (WRF-D02; thick gray solid line), the domain of POM (thin black dotted line), the outer domain of WW3 (WW3-D01; the same as WRF-D02), and the inner domain of WW3 (WW3-D02; gray dotted line).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Model domains of subsystems in RFSSME, including the outer domain of WRF (WRF-D01), the inner domain of WRF (WRF-D02; thick gray solid line), the domain of POM (thin black dotted line), the outer domain of WW3 (WW3-D01; the same as WRF-D02), and the inner domain of WW3 (WW3-D02; gray dotted line).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Model domains of subsystems in RFSSME, including the outer domain of WRF (WRF-D01), the inner domain of WRF (WRF-D02; thick gray solid line), the domain of POM (thin black dotted line), the outer domain of WW3 (WW3-D01; the same as WRF-D02), and the inner domain of WW3 (WW3-D02; gray dotted line).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The OF was developed on the basis of the Princeton Ocean Model (POM; Blumberg and Mellor 1987). The domain covers the SCS and the NWP (97°–135°E, 0°–30°N) (Fig. 2) with a horizontal resolution of 1/15° and 40 vertical layers from surface to 7000-m depth. The topography data used in this model are the 1 arc min global relief model of Earth’s topography and bathymetry (ETOPO1; Marks and Smith 2006). The daily forecast results from a quasi-global Hybrid Coordinate Ocean Model (HYCOM; Cummings and Smedstad 2013) with a horizontal resolution of 1/6° are used as the boundary conditions. In addition, the Oregon State University Tidal Prediction Software (OTPS; Egbert et al. 1994) provides tidal levels and currents of 13 primary tidal constituents at open boundaries. A multiscale 3DVAR (MS3DVAR) data assimilation scheme (Li et al. 2008a,b; Peng et al. 2016) is employed in POM. The observations assimilated in MS3DVAR system include the real-time daily SST data from Remote Sensing Systems (http://www.remss.com), the along-track SLA data from Copernicus Marine Environment Monitoring Service (http://marine.copernicus.eu), as well as Argo temperature/salinity (T/S) profile data from the French Research Institute for the Sustainable Exploitation of the Sea (IFREMER), which are assimilated once every day at 0000 UTC.
The WRF-D02 is coupled with the POM model through the Ocean Atmosphere Sea Ice Soil III (OASIS3) coupler developed by the European Center for Research and Advanced Training in Scientific Computation (CERFACS; Valcke et al. 2000). The OASIS3 coupler consists of an executable program performing the regridding of the coupling fields and a coupling library performing the coupling exchanges. Using a bilinear interpolation that is flux conservation, the SST outputs from the POM model are transferred to the WRF Model, while the wind stress, moisture, heat and radiation fluxes computed from the WRF Model are passed to the POM model every 10 min.
The WF was developed on the basis of the WAVEWATCH III (WW3) model (Tolman 1997, 1999, 2002). The WW3 model here is also a double-nested system. The outer domain (WW3-D01) shares the same area as WRF-D02 with a horizontal resolution of 1/10°. The inner domain (WW3-D02) covers the surrounding area of Hainan Island (105°–113°E, 16°–23°N) with a horizontal resolution of 1/20° (Fig. 2). The 10-m-height wind field from the WRF-D02 is used to drive the WW3. The WW3-D01 is cooperated with an optimal interpolation data assimilation system, which is able to assimilate the significant wave height (SWH) observations from the wave buoy (Wang 2013).
RFSSME makes a 120-h forecast automatically four times per day (i.e., at 0000, 0600, 1200, 1800 UTC) and provides the forecast results of atmosphere, ocean, and wave. The system has been running stably since it was established and shows good performance in the prediction of the marine environment, especially in the forecasts of TC track and sea waves. In this paper, we show the forecast results of the track and sea waves during Supertyphoon Mangkhut (2018).
During 2018 there are in total 29 TCs generated in the NWP including the SCS, among which the twenty-second, Typhoon Mangkhut (2018), is the most destructive one. Figure 3 shows the best track of Supertyphoon Mangkhut (2018) provided by the Japan Meteorological Agency (JMA). Mangkhut (2018) formed over the NWP about 1500 km west of the international date line as a tropical depression on 7 September. It soon developed into a tropical storm and then intensified further into a typhoon as it moved westward. Favorable environmental conditions, including low vertical wind shear, ample outflow aloft, high sea surface temperature and high ocean heat content, hastened the development of the Mangkhut (2018) and made it transform into a supertyphoon with its central minimum pressure of 905 hPa and maximum wind of 110 kt (1 kt ≈ 0.51 m s−1). When moving westward, Mangkhut (2018) made the first landfall at Baggao, Cagayan Province, Philippines (17.9°N, 122°E) at 1800 UTC 14 September at its peak intensity. Subsequently, Mangkhut (2018) entered the SCS, swept over the northern SCS, and then made the second landfall at Shangchuan Island (or Haiyan town officially), Taishan City, Guangdong Province, China (21.8°N, 112.6°E) at 0900 UTC 16 September with its central minimum pressure of 955 hPa and maximum wind of 75 kt. Mangkhut (2018) dissipated over Guangxi Province, China on 17 September (http://agora.ex.nii.ac.jp/digital-typhoon/summary/wnp/s/201822.html.en). Mangkhut (2018) brought a huge disaster to the regions along its track, including Guam, the Northern Mariana Islands, Philippines, Taiwan, southern China, and Vietnam, with a total economic loss of more than $3.74 billion and at least 134 fatalities (https://en.wikipedia.org/wiki/Typhoon_Mangkhut).
The best track of Mangkhut (2018) provided by JMA, in which the colors of the circles denote the typhoon grade at each moment.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The best track of Mangkhut (2018) provided by JMA, in which the colors of the circles denote the typhoon grade at each moment.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The best track of Mangkhut (2018) provided by JMA, in which the colors of the circles denote the typhoon grade at each moment.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
3. Validation of track and sea wave forecasting during Mangkhut (2018)
RFSSME started the forecast of Mangkhut (2018) from 0600 UTC 8 September when it moved into the model domain. For comparison, the predicted tracks from three other official agencies—that is, CMA, Joint Typhoon Warning Center (JTWC), and JMA—are also included here. Table 1 lists the mean track forecast errors of Mangkhut (2018). In the view of averaging for all forecast periods, the track forecast skill of RFSSME is comparable to those of the three official agencies, with an overall mean track error of 69.9 km that ranks the second. For the longer forecast period of 49–72 h, RFSSME performed best with a mean track error of 95.24 km. In addition, we calculated the mean biases of the landing location and time of the landfall in the coastal island of South China for different forecast periods (Table 2). In an averaging over all forecasting made during 1–72 h, RFSSME outperformed the three official agencies in the prediction of both the landing location and time with mean biases of 76.89 km and 3 h, respectively. In particular, RFSSME ranks the first to predict successfully the landfall location at Shangchuan Island when initialized at 0600 UTC 15 September 2018 (27 h ahead of the landfall, as shown in Fig. 4) with a distance (time) bias of 3.55 km (1 h), the smallest in comparison with an averaging bias of 104.63 km (3 h) for the three official agencies.
Mean errors (km) of the track forecasts from different forecast agencies for different forecast periods. The rank of RFSSME in forecast skill among the four forecast agencies/systems at each forecast period is listed in the rightmost column.
Mean distance (km) and time (h; in parentheses) biases of the landfall location in the coastal island of South China from different forecast agencies for different forecast periods. The rightmost column gives the rank of RFSSME in the skill of predicting the landfall location and time (in parentheses) among the four forecast agencies/systems for each forecast period.
Tracks of Mangkhut (2018) predicted by RFSSME (black), CMA (red), JTWC (blue), and JMA (green), initialized at 0600 UTC 15 Sep 2018. The gray line denotes the best track; the solid circles from east to west represent the locations of Mangkhut (2018) at 1800 UTC 15 Sep, 0600 UTC 16 Sep, and 1800 UTC 16 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Tracks of Mangkhut (2018) predicted by RFSSME (black), CMA (red), JTWC (blue), and JMA (green), initialized at 0600 UTC 15 Sep 2018. The gray line denotes the best track; the solid circles from east to west represent the locations of Mangkhut (2018) at 1800 UTC 15 Sep, 0600 UTC 16 Sep, and 1800 UTC 16 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Tracks of Mangkhut (2018) predicted by RFSSME (black), CMA (red), JTWC (blue), and JMA (green), initialized at 0600 UTC 15 Sep 2018. The gray line denotes the best track; the solid circles from east to west represent the locations of Mangkhut (2018) at 1800 UTC 15 Sep, 0600 UTC 16 Sep, and 1800 UTC 16 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
We next validate the forecast of the significant wave height (SWH) from RFSSME against observations. There was a disposable wave buoy maintained by the National Ocean Technology Center working east off the Luzon strait during the life cycle of Mangkhut (2018) (Fig. 5). The SWH measured by the wave buoy is from 6 to 16 September 2018 with an interval of 1 h. To assess the forecast skills of different forecast periods, we separate the 5-day forecast at 0000 UTC of each day from 6 to 16 September 2018 into five sections of 1–24, 25–48, 49–72, 73–96 and 97–120 h and map them into corresponding observations in terms of time period, resulting in five series of SWH forecasts with different forecast periods, as shown in Fig. 6. It can be seen that the SWH predictions at different forecast periods are similar to each other and are close to the observations, with an overall mean forecast error of 0.54 m. The performance of RFSSME in SWH forecast can be further seen in the scatterplot (Fig. 7), which shows that the predicted SWH has a high correlation with the observed SWH at different forecast periods with an average correlation coefficient up to 0.94. In general, the correlation coefficient of predicted SWH falls dramatically with the forecast period larger than 3 days because of the increase of weather forecast errors. However, the mean errors of predicted SWH by RFSSME just increase slightly from 0.47 m for 1–24 h forecast to 0.53 m (0.53 m) for the 73–96 h (97–120 h) forecast, with the corresponding correlation coefficients decreasing slightly from 0.95 to 0.93 (0.92). These results suggest RFSSME has high skill in sea wave prediction.
The trajectory of the wave buoy (red line) and the best track of Mangkhut (2018) (gray line; the colors of the circle denote the typhoon grades).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The trajectory of the wave buoy (red line) and the best track of Mangkhut (2018) (gray line; the colors of the circle denote the typhoon grades).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The trajectory of the wave buoy (red line) and the best track of Mangkhut (2018) (gray line; the colors of the circle denote the typhoon grades).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Time series of SWH forecasts (black line) and observations (gray line) for different forecast periods of (a) 1–24, (b) 25–48, (c) 49–72, (d) 73–96 and (e) 97–120 h, in which the mean errors are 0.47, 0.58, 0.57, 0.53 and 0.53 m, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Time series of SWH forecasts (black line) and observations (gray line) for different forecast periods of (a) 1–24, (b) 25–48, (c) 49–72, (d) 73–96 and (e) 97–120 h, in which the mean errors are 0.47, 0.58, 0.57, 0.53 and 0.53 m, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Time series of SWH forecasts (black line) and observations (gray line) for different forecast periods of (a) 1–24, (b) 25–48, (c) 49–72, (d) 73–96 and (e) 97–120 h, in which the mean errors are 0.47, 0.58, 0.57, 0.53 and 0.53 m, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Scatterplot (colored dots) and corresponding linear trends (colored lines) between the SWH forecasts and observations for different forecast periods of 1–24, 25–48, 49–72, 73–96 and 97–120 h, in which the correlation coefficients are 0.95, 0.94, 0.93, 0.93 and 0.92, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Scatterplot (colored dots) and corresponding linear trends (colored lines) between the SWH forecasts and observations for different forecast periods of 1–24, 25–48, 49–72, 73–96 and 97–120 h, in which the correlation coefficients are 0.95, 0.94, 0.93, 0.93 and 0.92, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Scatterplot (colored dots) and corresponding linear trends (colored lines) between the SWH forecasts and observations for different forecast periods of 1–24, 25–48, 49–72, 73–96 and 97–120 h, in which the correlation coefficients are 0.95, 0.94, 0.93, 0.93 and 0.92, respectively.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
4. Sensitivity experiments
To further investigate which technique employed in RFSSME, the air–sea coupling or the SSDA scheme, plays an important role in improving the track forecasting skill for Mangkhut (2018), three sensitivity experiments are conducted. The three experiments have the same model setup as the RFSSME except that one of the two techniques, that is, the air–sea coupling and the SSDA scheme, is excluded; they are denoted as COUPLE_X, SSDA_X, and BOTH_X (Table 3). All of the experiments start from 0600 UTC 8 to 0600 UTC 16 September 2018 with four times of forecasting per day as for RFSSME.
A list of sensitivity experiments that include (√) or do not include (×) the air–sea coupling scheme and/or the SSDA scheme in RFSSME.
Table 4 presents the mean track forecast errors at difference forecast periods from RFSSME, COUPLE_X, SSDA_X, and BOTH_X. The results show that, when the air–sea coupling scheme is excluded (COUPLE_X) in RFSSME, the track forecast errors are slightly reduced, with a 5-day mean of 114.0 km as compared with 118.04 km from RFSSME, suggesting that the air–sea coupling scheme does not help improving the track forecasting of Mangkhut (2018); instead it slightly deteriorate the forecast skill; when the SSDA scheme is excluded, however, the track forecast errors become much larger than those from RFSSME, with a 5-day mean of 184.73 km, suggesting that the SSDA scheme plays a vital role in the track forecasting of Mangkhut (2018). Moreover, the difference of track forecast errors between SSDA_X and RFSSME becomes larger and larger with the increase of forecast periods, implying that the SSDA scheme is more effective for longer forecast periods. The results from BOTH_X are similar to, but slightly better than, those from SSDA_X.
Mean track forecast errors (km) from the RFSSME, COUPLE_X, SSDA_X, and BOTH_X forecasting experiments for different forecast periods.
(a) Forecast tracks and (b) corresponding errors of Mangkhut (2018) from the GFS, RFSSME, COUPLE_X, SSDA_X, and BOTH_X, initialized at 0000 UTC 12 Sep 2018. The gray line in (a) denotes the best track; the solid circles in (a) from east to west represent the locations of Mangkhut (2018) at 0000 UTC 13 Sep, 0000 UTC 14 Sep, 0000 UTC 15 Sep, and 0000 UTC 16 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
(a) Forecast tracks and (b) corresponding errors of Mangkhut (2018) from the GFS, RFSSME, COUPLE_X, SSDA_X, and BOTH_X, initialized at 0000 UTC 12 Sep 2018. The gray line in (a) denotes the best track; the solid circles in (a) from east to west represent the locations of Mangkhut (2018) at 0000 UTC 13 Sep, 0000 UTC 14 Sep, 0000 UTC 15 Sep, and 0000 UTC 16 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
(a) Forecast tracks and (b) corresponding errors of Mangkhut (2018) from the GFS, RFSSME, COUPLE_X, SSDA_X, and BOTH_X, initialized at 0000 UTC 12 Sep 2018. The gray line in (a) denotes the best track; the solid circles in (a) from east to west represent the locations of Mangkhut (2018) at 0000 UTC 13 Sep, 0000 UTC 14 Sep, 0000 UTC 15 Sep, and 0000 UTC 16 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Temporal evolution of forecast minimum sea level pressure of Mangkhut (2018) from RFSSME, COUPLE_X, SSDA_X, and BOTH_X, initialized at 0000 UTC 12 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Temporal evolution of forecast minimum sea level pressure of Mangkhut (2018) from RFSSME, COUPLE_X, SSDA_X, and BOTH_X, initialized at 0000 UTC 12 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Temporal evolution of forecast minimum sea level pressure of Mangkhut (2018) from RFSSME, COUPLE_X, SSDA_X, and BOTH_X, initialized at 0000 UTC 12 Sep 2018.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The forecast track from GFS global forecast, with a 5-day mean track forecast error of 174.79 km, is very close to RFSSME or COUPLE_X with the SSDA scheme assimilating the large-scale flow from GFS global forecast, though slightly deviates northward before crossing the Luzon Strait. These results imply that the large-scale environmental flow from GFS global forecast is of great influence on the track forecast of Mangkhut (2018). Figure 10 shows the vertical profiles of domain-averaged 5-day mean root-mean-square errors (RMSEs) of the large-scale u and υ components against the GFS analysis data for RFSSME, COUPLE_X, SSDA_X and BOTH_X initialized from 0000 UTC 12 September 2018. The RMSEs of large-scale components of the flow (hereafter denoted as large-scale flow) are significantly reduced by the SSDA scheme employed in RFSSME and COUPLE_X nearly at all levels from 1000 to 100 hPa, as compared with those without the SSDA scheme (SSDA_X and BOTH_X). The air–sea coupling may benefit the simulation of the large-scale flow in RFSSME or SSDA_X, but this beneficial effect may be offset or overwhelmed by negative impacts of the air–sea coupling on Mangkhut (2018) as discussed above, resulting in slightly larger track forecast errors for Mangkhut (2018) compared to those from COUPLE_X or BOTH_X. Figures 11 and 12 show the 500 hPa large-scale geopotential height at 0600 UTC 13 and 0600 UTC 15 September 2018, respectively, from RFSSME, COUPLE_X, SSDA_X, and BOTH_X initialized at 0000 UTC 12 September 2018, as well as from the GFS analysis data. At 0600 UTC 13 September 2018, the patterns of the 500 hPa large-scale geopotential height from RFSSME, COUPLE_X, SSDA_X, and BOTH_X (Figs. 11b–e) are similar to that from the GFS analysis data (Fig. 11a), except for the occurrence of a low pressure perturbation in the southwest area of the domain (Figs. 11b–e). At this moment, the steering flow of Mangkhut (2018), which is obtained from a height- and area-averaged large-scale flow between 925 and 300 hPa and over a 3°–8° ring area centered at the minimum sea level pressure location (Chan and Gray 1982; Evans et al. 1991), is nearly in the same direction (i.e., northwest) for the GFS analysis data, RFSSME, and sensitivity experiments, as indicated by the black arrow in Figs. 11a–e. However, 48 h later (i.e., at 0600 UTC 15 September 2018), this low pressure perturbation strengthens and develops into a strong low pressure system (LPS) in SSDA_X and BOTH_X (Figs. 12d,e), whereas it is suppressed in RFSSME and COUPLE_X (Figs. 12b,c). The unrealistic LPS may have the so-called Fujiwhara effect (Fujiwhara 1921) on Mangkhut (2018), that is, a cyclonic rotation between Mangkhut (2018) and the LPS. The “Fujiwhara effect” makes the LPS move toward Mangkhut (2018) and forces the typhoon to move northward, as indicated by the steering flows of the LPS and Mangkhut (2018) in SSDA_X and BOTH_X (Figs. 12d,e). With the LPS suppressed, the steering flows of Mangkhut (2018) from RFSSME (Fig. 12b) and COUPLE_X (Fig. 12c) maintained to be northwestward, which is close to that from the GFS analysis data (Fig. 12a). Therefore, the application of the SSDA scheme in RFSSME helps the regional model maintain a more realistic large-scale flow as obtained from the GFS forecasts, leading to a significant improvement in the track forecast of Mangkhut (2018) for longer forecast periods (say, longer than 48 h). Note that the effectiveness of the SSDA scheme is somewhat dependent on the accuracy of the assimilated large-scale flow; statistically, the large-scale flow obtained from the GFS forecasts is supposed to be more accurate than that from a regional model, but occasionally it could be degraded, in which case the SSDA scheme may not work well.
Vertical profiles of 5-day mean RMSEs (m·s−1) for large-scale (a) u and (b) υ components against GFS analysis for RFSSME, COUPLE_X, SSDA_X, and BOTH_X initialized at 0000 UTC 12 Oct 2018. The RMSEs are averaged over all grids in the domain.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Vertical profiles of 5-day mean RMSEs (m·s−1) for large-scale (a) u and (b) υ components against GFS analysis for RFSSME, COUPLE_X, SSDA_X, and BOTH_X initialized at 0000 UTC 12 Oct 2018. The RMSEs are averaged over all grids in the domain.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
Vertical profiles of 5-day mean RMSEs (m·s−1) for large-scale (a) u and (b) υ components against GFS analysis for RFSSME, COUPLE_X, SSDA_X, and BOTH_X initialized at 0000 UTC 12 Oct 2018. The RMSEs are averaged over all grids in the domain.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The 500-hPa large-scale geopotential height at 0600 UTC 13 Sep 2018 from (a) GFS analysis data, (b) RFSSME, (c) COUPLE_X, (d) SSDA_X, and (e) BOTH_X initialized at 0000 UTC 12 Sep 2018. The black arrows represent the steering flows of Mangkhut (2018).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The 500-hPa large-scale geopotential height at 0600 UTC 13 Sep 2018 from (a) GFS analysis data, (b) RFSSME, (c) COUPLE_X, (d) SSDA_X, and (e) BOTH_X initialized at 0000 UTC 12 Sep 2018. The black arrows represent the steering flows of Mangkhut (2018).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
The 500-hPa large-scale geopotential height at 0600 UTC 13 Sep 2018 from (a) GFS analysis data, (b) RFSSME, (c) COUPLE_X, (d) SSDA_X, and (e) BOTH_X initialized at 0000 UTC 12 Sep 2018. The black arrows represent the steering flows of Mangkhut (2018).
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
As in Fig. 11, but for the forecast time of 0600 UTC 15 Sep 2018. The black arrows represent the steering flows of Mangkhut (2018) and the low pressure system.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
As in Fig. 11, but for the forecast time of 0600 UTC 15 Sep 2018. The black arrows represent the steering flows of Mangkhut (2018) and the low pressure system.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
As in Fig. 11, but for the forecast time of 0600 UTC 15 Sep 2018. The black arrows represent the steering flows of Mangkhut (2018) and the low pressure system.
Citation: Journal of Atmospheric and Oceanic Technology 37, 11; 10.1175/JTECH-D-19-0196.1
5. Summary
A real-time regional marine forecasting system, called RFSSME, is introduced, and its performance in the track and accompanying sea wave forecasts of Supertyphoon Mangkhut (2018) is assessed. The results show that, statistically, the performance of RFSSME in track forecasts is comparable to, or slightly better than, those of other agencies in comparison, with a mean track error of 69.9 km for the forecast period of 1–72 h. In particular, RFSSME showed a very high skill in the prediction of the landing location in the coastal island of South China with distance (time) biases of 76.89 km (3 h) averaging over all forecasting made during 1–72 h and only 3.55 km (1 h) for the forecasting initialized at 0600 UTC 15 September 2018 (27 h ahead of the landfall), outperforming the three official agencies. RFSSME also shows a good forecast skill in the sea wave forecast with a mean SWH error of 0.54 m and correlation coefficient of 0.94.
The results from sensitivity experiments demonstrate that the SSDA scheme plays a vital role in improving the track forecast skill for Mangkhut (2018). Further analysis indicates that the SSDA scheme helps the atmospheric model achieve a more accurate prediction of the steering flow of the typhoon through fitting the large-scale flow from the regional model to that from the GFS forecasts, leading to a significant improvement in the track forecast skill for Mangkhut (2018). On the other hand, however, the weakening effect of air–sea coupling on typhoon intensity through wind-induced sea surface cooling could make the typhoon deviate northward due to the barotropic conservation of potential vorticity or the so-called beta drift, leading to a lightly negative impact on the track forecast skill for Mangkhut (2018).
The good performance of RFSSME in the track and accompanying sea wave forecasts of Mangkhut (2018) suggests that RFSSME has a capability to give a valuable early warning of the high impact weather events, TCs, to the government or public in the regions of high TC incidence to reduce the loss of property or life.
Acknowledgments
This work was jointly supported by the National Natural Science Foundation of China (Grants 41931182, 41890851, 41521005, 41776028 and 41676016), Guangdong Special Support Program (2019BT2H594), the Key Special Project for Introduced Talents Team of Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou) (GML2019ZD0303), the Chinese Academy of Sciences (Grants ZDRW-XH-2019-2, ISEE2018PY05, and 133244KYSB20180029), and the Independent Research Project Program of State Key Laboratory of Tropical Oceanography under contract (LTOZZ2004, LTOZZ1902 and LTOZZ1802). The authors gratefully acknowledge the use of the HPCC at the South China Sea Institute of Oceanology, Chinese Academy of Sciences.
REFERENCES
Bender, M. A., I. Ginis, R. E. Tuleya, B. Thomas, and T. Marchok, 2007: The operational GFDL coupled hurricane–ocean prediction system and a summary of its performance. Mon. Wea. Rev., 135, 3965–3989, https://doi.org/10.1175/2007MWR2032.1.
Blumberg, A. F., and G. L. Mellor, 1987: A description of a three-dimensional coastal ocean circulation model. Three-Dimensional Coastal Ocean Models, N. S. Heaps, Ed., Coastal and Estuarine Studies Series, Vol. 4, Amer. Geophys. Union, 1–16.
Chan, J. C. L., and W. M. Gray, 1982: Tropical cyclone movement and surrounding flow relationships. Mon. Wea. Rev., 110, 1354–1374, https://doi.org/10.1175/1520-0493(1982)110<1354:TCMASF>2.0.CO;2.
Cummings, J. A., and O. M. Smedstad, 2013: Variational data assimilation for the global ocean. Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications, S. Park and L. Xu, Eds., Vol. II, Springer, 303–343 pp.
Egbert, G. D., A. F. Bennett, and M. G. G. Foreman, 1994: TOPEX/Poseidon tides estimated using a global inverse model. J. Geophys. Res., 99, 24 821–24 852, https://doi.org/10.1029/94JC01894.
Evans, J. L., G. J. Holland, and R. L. Elsberry, 1991: Interactions between a barotropic vortex and an idealized subtropical ridge. Part I: Vortex motion. J. Atmos. Sci., 48, 301–314, https://doi.org/10.1175/1520-0469(1991)048<0301:IBABVA>2.0.CO;2.
Fraedrich, K., and M. Leslie, 1989: Estimates of cyclone track predictability. I: Tropical cyclones in the Australian region. Quart. J. Roy. Meteor. Soc., 115, 79–92, https://doi.org/10.1002/qj.49711548505.
Fujiwhara, S., 1921: The natural tendency towards symmetry of motion and its application as a principle in meteorology. Quart. J. Roy. Meteor. Soc., 47, 287–292, https://doi.org/10.1002/qj.49704720010.
Fulton, S. R., 2001: An adaptive multigrid barotropic tropical cyclone track model. Mon. Wea. Rev., 129, 138–151, https://doi.org/10.1175/1520-0493(2001)129<0138:AAMBTC>2.0.CO;2.
Hong, J. S., C. T. Fong, L. F. Hsiao, Y. C. Yu, and C. Y. Tzeng, 2015: Ensemble typhoon quantitative precipitation forecasts model in Taiwan. Wea. Forecasting, 30, 217–237, https://doi.org/10.1175/WAF-D-14-00037.1.
Hsiao, L. F., M. S. Peng, D. S. Chen, K. N. Huang, and T. C. Yeh, 2009: Sensitivity of typhoon track predictions in a regional prediction system to initial and lateral boundary conditions. J. Appl. Meteor. Climatol., 48, 1913–1928, https://doi.org/10.1175/2009JAMC2038.1.
Jia, X., C. T. Lu, and J. Lu, 2010: Statistical characteristics of typhoons affecting China coast and numerical simulation of typhoon waves (in Chinese). J. Waterw. Harbor, 31, 433–436.
Khain, A., and B. Lynn, 2011: Simulation of tropical cyclones using spectral bin microphysics. Recent Hurricane Research—Climate, Dynamics, and Societal Impacts, A. Lupo, Ed., Intech Open, 197–226.
Lai, Z. J., S. Hao, S. Q. Peng, B. Liu, X. Q. Gu, and Y. K. Qian, 2014: On improving tropical cyclone track forecasts using a scale-selective data assimilation approach: A case study. Nat. Hazards, 73, 1353–1368, https://doi.org/10.1007/s11069-014-1155-y.
Leslie, L. M., J. F. LeMarshall, R. P. Morison, C. Spinoso, R. J. Purser, N. Pescod, and R. Seecamp, 1998: Improved hurricane track forecasting from the continuous assimilation of high quality satellite wind data. Mon. Wea. Rev., 126, 1248–1258, https://doi.org/10.1175/1520-0493(1998)126<1248:IHTFFT>2.0.CO;2.
Li, Z. J., Y. Chao, J. C. McWilliams, and K. Ide, 2008a: A three-dimensional variational data assimilation scheme for the Regional Ocean Modeling System: Implementation and basic experiments. J. Geophys. Res., 113, C05002, https://doi.org/10.1029/2008JC004928.
Li, Z. J., Y. Chao, J. C. McWilliams, and K. Ide, 2008b: A three-dimensional variational data assimilation scheme for the Regional Ocean Modeling System. J. Atmos. Ocean. Tech., 25, 2074–2090, https://doi.org/10.1175/2008JTECHO594.1.
Liu, B., S. Z. Lu, Y. K. Qian, and S. Q. Peng, 2014: An application of ATOVS radiance in data assimilation of typhoon (in Chinese). J. Trop. Oceanogr., 31, 44–53.
Liu, D. F., L. Pang, and B. T. Xie, 2009: Typhoon disaster in China: Prediction, prevention, and mitigation. Nat. Hazards, 49, 421–436, https://doi.org/10.1007/s11069-008-9262-2.
Liu, Z. Q., and D. Barker, 2006: Radiance assimilation in WRF-Var: Implementation and initial results. Seventh WRF User’s Workshop, Boulder, CO, DTC, 4.2.
Marks, K. M., and W. H. F. Smith, 2006: An evaluation of publicity available global bathymetry grids. Mar. Geophys. Res., 27, 19–34, https://doi.org/10.1007/s11001-005-2095-4.
Neumann, C. J., 1979: A guide to Atlantic and eastern Pacific models for the prediction of tropical cyclone motion. NOAA Tech. Memo. NWS 93226 pp.
Peng, S. Q., L. Xie, B. Liu, and F. Semazzi, 2010: Application of scale-selective data assimilation to regional climate modeling and prediction. Mon. Wea. Rev., 138, 1307–1318, https://doi.org/10.1175/2009MWR2974.1.
Peng, S. Q., and Coauthors, 2014: On the mechanisms of the recurvature of Super Typhoon Megi. Sci. Rep., 4, 4451, https://doi.org/10.1038/srep04451.
Peng, S. Q., and Coauthors, 2015: A real-time regional forecasting system in the South China Sea and its performance in the track forecasts of tropical cyclones during 2011–2013. Wea. Forecasting, 30, 471–485, https://doi.org/10.1175/WAF-D-14-00070.1.
Peng, S. Q., X. Z. Zeng, and Z. J. Li, 2016: A three-dimensional variational data assimilation system for the South China Sea: Preliminary results from observing system simulation experiments. Ocean Dyn., 66, 737–750, https://doi.org/10.1007/s10236-016-0946-y.
Plu, M., 2011: A new assessment of the predictability of tropical cyclone tracks. Mon. Wea. Rev., 139, 3600–3608, https://doi.org/10.1175/2011MWR3627.1.
Radford, A. M., 1994: Forecasting the movement of tropical cyclones at the Met. Office. Meteor. Appl., 1, 355–363, https://doi.org/10.1002/met.5060010406.
Rao, D. V. B., and D. H. Prasad, 2007: Sensitivity of tropical cyclone intensification to boundary layer and convective processes. Nat. Hazards, 41, 429–445, https://doi.org/10.1007/s11069-006-9052-7.
Rossby, C. G., 1948: On displacement and intensity changes of atmospheric vortices. J. Mar. Res., 7, 175–196.
Roy, C., and K. Rita, 2012: Tropical cyclone track forecasting techniques—A review. Atmos. Res., 104–105, 40–69, https://doi.org/10.1016/J.ATMOSRES.2011.09.012w.
Skamarock, W. C., and J. B. Klemp, 2008: A time-split nonhydrostatic atmospheric model for weather research and forecasting applications. J. Comput. Phys., 227, 3465–3485, https://doi.org/10.1016/j.jcp.2007.01.037.
Tolman, H. L., 1997:A new global wave forecast system at NECP. Ocean Wave Measurements and Analysis, B. L. Edge and J. M. Helmsley, Eds., Vol. 2, ASCE, 777–786.
Tolman, H. L., 1999: User manual and system documentation of WAVEWATCH-III version 1.18. NOAA Tech. Note 166, 110 pp.
Tolman, H. L., 2002:User manual and system documentation of WAVEWATCH-III version 2.22. NOAA Tech. Note 222, 133 pp.
Valcke, S., L. Terray, and A. Piacentini, 2000: The OASIS Coupler user’s guide: Version 2.4. CERFACS Tech. Rep. TR/CGMC/00-10, 85 pp.
Wang, A. M., 2013: Typhoon wave assimilation model establishment and South China Sea typhoon wave characteristics under the background of winter monsoon. M.S. thesis, Institute of Oceanology, Chinese Academy of Sciences, 63 pp.
Wang, B., R. L. Elsberry, Y. Q. Wang, and L. G. Wu, 1998: Dynamics in tropical cyclone motion: A review. Chin. J. Atmos. Sci., 22, 416–434.
Wang, J. B., and W. H. Qian, 2005: Statistic analysis of tropical cyclone impact on the China mainland during the last half century. Chin. J. Geophys., 48, 1069–1077, https://doi.org/10.1002/cjg2.750.
Wu, C. C., and Y. H. Kuo, 1999: Typhoons affecting Taiwan: Current understanding and future challenges. Bull. Amer. Meteor. Soc., 80, 67–80, https://doi.org/10.1175/1520-0477(1999)080<0067:TATCUA>2.0.CO;2.
Wu, L. G., B. Wang, and S. A. Braun, 2005: Impacts of air–sea interaction on tropical cyclone track and intensity. Mon. Wea. Rev., 133, 3299–3314, https://doi.org/10.1175/MWR3030.1.
Xie, L., B. Liu, and S. Q. Peng, 2010: Application of scale-selective data assimilation to tropical cyclone track simulation. J. Geophys. Res., 115, D17105, https://doi.org/10.1029/2009JD013471.
Zhou, H., W. Zhu, and S. Q. Peng, 2013: The impacts of different micro-physics schemes and boundary layer schemes on the simulated track and intensity of Super Typhoon Megi (2013) (in Chinese). J. Trop. Meteor., 28, 599–608.
