The China Meteorological Administration (CMA) best-track data (Ying et al., 2014; Lu et al., 2021) is mainly used. Benefiting from more observations, it is one of the best best-track data that gives reliable estimates near China’s water. It provides elementary TC information including 6-hourly TC position, minimum sea-level pressure (MSLP) near the TC center, 2-min mean maximum sustained wind, etc. The best-track data from the Japan Meteorological Agency (JMA) and Joint Typhoon Warning Center (JTWC) are used for comparison. In any case, the selection of best-track data does not matter in the analysis of the PTC-STC in this study.

The National Centers for Environmental Prediction—Climate Forecast System Reanalysis (NCEP-CFSR) data is used for the numerical model (Section 2.3) initial condition. It is a global, high resolution (6-hourly, 0.5° × 0.5°, 37 levels), coupled atmosphere-ocean-land surface-sea ice system designed to provide the best estimates of the state of these coupled domains. It is one of the mainstreams with high recognition and confidence nowadays.

The Weather Research and Forecasting Model Version 4.0 (WRF v4.0; Powers et al., 2017) is employed. A two-way and triple-nested domain with 43 vertical levels is configured (Figure 1). The horizontal resolutions of the outermost, intermediate, and innermost domains are 36, 12, and 4 km, respectively. For the model physics, the eta microphysics (NOAA, 2001), Rapid Radiative Transfer Model (RRTMG) longwave and shortwave radiation schemes (Mlawer et al., 1997; Iacono et al., 2008), eta similarity surface layer physics (Janjić, 1994), Noah-MP (multi-physics) land surface model (Niu et al., 2011; Yang et al., 2011), Mellor-Yamada-Janjić (MYJ) planetary boundary layer scheme (Mesinger, 1993), and Kain-Fritsch (KF) cumulus scheme (Kain, 2004) are adopted in all domains. The model initial condition is derived from the CFSR and no boundary condition is updated during the simulations. All simulations are run for 72 h which is long enough as mainly the TCG stage of STC is focused on.

A statistically representative PTC-STC event (PTC Neil and STC Nameless) in 1999 is chosen for the case study (Figure 1). It is explicitly selected from the climatology over 1980–2020 based on the following criteria: 1) the event is in the season that has the most occurrences of PTC-STC events (i.e., July, August or September); 2) the PTC is in the region of the highest spatial density of PTCs; 3) the STC appears after the PTC approaches and/or hits the Philippines; 4) the time difference between PTC and STC is no more than 4 days (Schenkel, 2016); 5) the relative spatial distance between PTC and STC is no more than 2,500 km (Schenkel, 2016).

In the control experiment (CTRL), we run the model with the CFSR data and test the model confidence with the CMA best-track data. The CTRL experiment intends to refine the reanalysis data (from ∼55 km downscaled to 4 km) and produce dependable model results. This not only provides a finer and reliable understanding of the physical processes, but also serves as a benchmark for the comparison with the sensitivity experiments (SEA and IN; see below). We hypothesize that the terrain of the Philippines and the intensity of PTC could matter the TCG in the SCS.

In order to test the importance of the terrain of the Philippines, we replace it by sea from the CTRL experiment, namely, SEA experiment. In addition, to test the dependence on the PTC intensity, we intensify the PTC by inserting a stronger TC-like vortex (maximum wind = 40 m s⁻¹, radius of maximum wind = 50 km) to the CTRL experiment using TC bogus scheme from the WRF model, namely, IN experiment. The initial MSLP of PTC in the CTRL and IN experiments are about 996 hPa and 984 hPa, respectively (Figure 4D).

Figures 1, 2 show the CMA best tracks and reanalyzed lower-tropospheric synoptic evolutions of the PTC-STC event in this study. In general, a PTC Neil (1999) formed over the WNP in late July 1999. It moved westward towards Luzon at first, and then intensified, lingered, and recurved northward on the Philippine Sea. A STC Nameless (1999) subsequently formed and developed over the SCS (Figure 2). Later, Neil moved northeastward and dissipated, while Nameless moved northwestward and made landfall on South China.

The CTRL experiment starts at 18 UTC July 21, that is, the time when the PTC is in the Philippine Sea and before the STC forms. The model results show that when the PTC is approaching to the east of Luzon, a low-pressure disturbance emerges, develops, and becomes the STC gradually at the adjacent water west of Luzon (Figure 3). Although the STC in the late stage of simulation makes landfall on Luzon which is not in line with the best tracks (Figure 4), the successful simulation of TCG and the development of STC is promising which largely agree with the reanalysis and best-track data (cf. Figures 1–4). Hence, hereafter, the results in the early-to-mid simulation will be mainly examined which is long enough to cover the main focus, PTC-STC, of this study.

Dynamically, Figures 5, 6 suggest that the TCG of STC is largely driven by the horizontal advection and vertical propagation of absolute vorticity in the northwest water of Luzon. At early simulation, a weak embryo V around the northern tip of Luzon is found in the lower troposphere (925 hPa; Figure 5A). An embryo V is identified and traced by the local maximum of absolute vorticity. From 19 UTC July 21 to 00 UTC July 22, as the PTC moves westward, the embryo strengthens, and the lower-tropospheric wind nearby changes gradually from northeasterly to northerly or northwesterly such that the embryo is advected cyclonically correspondingly (Figure 5B). The area of embryo shrinks while it intensifies and extends upward to 500 hPa. The meso-α scale averaged vorticity about the center of embryo increases below 850 hPa (Figure 6A, box 1), which is related to the wave pouch which represents a wider cyclonic wind field that could favor the aggregation of vorticity (Dunkerton et al., 2009; Montgomery et al., 2010; Raymond and López Carrillo, 2011; Wang et al., 2012). Soon, the averaged vorticity increases in both meso-α (Figure 6A, box 2) and meso-β (Figure 6B, box 1; related to the self-development of the vortex or local fine perturbations) scales between 1 and 3 km. These imply the advected vorticity from the PTC is mainly confined in the low-to-mid troposphere (below 700 hPa) at the early stage. It is noted that examining the meso-α and meso-β scale averaged vorticity here helps to understand whether the vorticity is prevailed locally or from the larger-scale environment explicitly.

Later, when the embryo is advected southward or southeastward, hits the mountain and lingers at the periphery (Figure 5C), it intensifies below 1 km after 09 UTC July 22 (Figure 6B, box 2), which is possibly related to the increasing horizontal wind shear caused by the friction of the steep mountain range near the west slope. From 06 to 18 UTC July 22, the vorticity of the embryo continues to transport downward and downscales from meso-α scale to meso-β scale. Meanwhile, the strengthened southwesterly convergence flow generates positive vorticity fragments below 700 hPa in the southern flank of the PTC causing the east-west vorticity fragment belt at 15°N spirals into the PTC center, and thus PTC intensifies gradually.

From 20 UTC July 22 to 03 UTC July 23 (Figure 5D), the horizontal wind shear over the SCS increases with the PTC intensity, provides and propagates positive vorticity upward, and feeds the embryo further. A strong vorticity in meso-β scale soon develops upward to 700 hPa gradually, just along the west side of the mountain range (Figure 6B, box 3). A structural, closed cyclonic circulation begins to appear in the lower troposphere (Figure 5D). At 06 UTC July 23, the meso-α scale averaged vorticity about the center of embryo increases sharply and soon develops throughout low-to-mid troposphere, showing the horizontally expansion and uplift of vorticity (Figure 6A, box 3).

Thermodynamically, no apparent warm core structure appears in the northwest water of Luzon in the early simulation until 05 UTC July 22 (Figure 7A). At 11 UTC July 22, a closed low-pressure disturbance forms at 925 hPa near the southwest side of the mountain range (Figure 7B). This disturbance expands northward at 13 UTC and aligns with the warm air mass near the northwest side of the mountains at 14 UTC (not shown). There is no significant deepening of disturbance from 14 to 22 UTC, but warm air at 500 hPa extends westward (Figure 7C). Subsequently, the disturbance deepens in the lower troposphere, and the warm core low-pressure center emerges in 700–925 hPa due to the adiabatic heating (Bao et al., 2018) and develops upward gradually. The disturbance becomes a tropical depression and coheres vertically with an obvious warm core structure in the low-to-mid troposphere (Figure 9A).

The low-to-mid troposphere is humid during the emergence of STC in the simulation. From the 18–23 UTC July 21, there is strong easterly above 3.5 km (300–650 hPa) with a moisture tongue transporting water vapor westward (Figure 8A), and relatively weak westerly underneath (650–975 hPa) over the SCS (Figure 8A). As the westerly strengthens, the water vapor below 3 km gradually accumulates near the mountains from 00 to 05 UTC July 22 (Figure 9A). With the easterly shear and uplifting over the SCS, the moisture is transported upward along the mountains. At 10 UTC July 22, the water vapor in the lower troposphere extends upward to more than 7 km above the mountain, and a deep column of wet air through the middle troposphere forms (Figure 9A) and tilts westward with height gradually. Later, the inclined wet air column moves westward to 118°E, and another column of wet air extends up from the west side of the mountain to the middle troposphere (Figure 9A).

Inspired by both the observations and CTRL experiment (i.e., the TCG location, and the dynamic and thermodynamic evolutions of STC), we hypothesize that the terrain of the Philippines could be decisive to the TCG of STC in the SCS. The SEA experiment shows that if the terrain of the Philippines is replaced by sea, no STC forms in the SCS (Figures 4B, 10).

In the SEA experiment, 1) the intensity of PTC, so does its cyclonic vorticity, is stronger than that in the CTRL experiment (Figures 4D, 11); 2) a weaker cyclonic vorticity is found near the west coast of the Philippines which corresponds to the absence of southward vorticity advection of the aforementioned vorticity embryo in the CTRL experiment (Figure 11); 3) no significant accumulation of moisture is found near the position of mountains (Figure 9B), compared to the CTRL experiment; 4) the uplift due to the topographic upslope is absent (Figure 9B); and 5) no apparent warm-core low is found in the SCS (Figure 9B). All these remarkable differences suggest that the terrain of the Philippines (steep mountain range of Luzon in particular) is geographically essential to the PTC-STC type of TCG in the SCS because it acts like a physical barrier that can accumulate moisture (Figure 9B), constrain the vorticity embryo, allow upslope lifting, and facilitate cyclonic development in the lower troposphere of the SCS, providing favorable TCG conditions for the STC.

Climatologically, we discover several similar PTC-STC events in recent decades (1980–2020; not shown). Among them, most of the PTCs are not very strong (below typhoon level) and do not move into the SCS. Hence, we further hypothesize that the TCG of STC in the SCS could be PTC intensity and track dependent. The IN experiment shows that if the PTC is strong, the PTC could move into the SCS, the occurrence of TCG of STC is unlikely (Figures 4B, 12).

In the IN experiment, the stronger PTC (Figure 13A) moves northward along the east coast of Luzon from 18 UTC July 21 to 10 UTC July 22, and gradually approaches the island westward (Figures 4, 12A). With the strengthening of PTC, the northeasterly near the west coast of Luzon gets stronger as well and no obvious warm core appears on the leeside. The PTC makes landfall on the northeast coast of Luzon at 17 UTC July 22 (Figure 12B). As it encounters mountains, its warm-core low structure is distorted and weakened (Figure 14B). During the landfall of PTC, a new weak lower-tropospheric low-pressure center emerges on the west side of the mountain range from 14 UTC July 22 and moves southward gradually (Figures 12, 14A, 14B). As less positive vorticity arises near the west coast of Luzon, the subsequent new center soon dissipates (Figure 13B). Instead, the cyclonic low-pressure structure of PTC crosses Luzon and enters the SCS. A new center of PTC appears in the lower troposphere at 00 UTC July 23 and redevelops (Figures 13C, 14C). No standalone STC forms.

In the process of the intensified PTC approaching and passing through Luzon, the water vapor mainly transports from east to west in 300–650 hPa (Figure 8B), and the water vapor flux in 650–975 hPa is larger than that in CTRL (cf. Figures 8A, B). The wet air on the east side of the mountains is lifted to the middle troposphere, and moves westward gradually (Figure 9C). The water vapor in the lower troposphere over the SCS is also lifted on the west side of the mountains. Compared to the CTRL experiment, the stronger easterly wind above 3 km and less moisture accumulating below 3 km to the west of Luzon in the early stage (cf. Figures 9A, C) could suppress the development of local convective activities over the SCS and thus does not favor the formation of STC.