1. Introduction
The Pearl River Delta (PRD), located in the coastal area of Southern China, known as the “World Factory” (Q. Zhang et al. 2009), is associated with a considerable amount of the economy and population, as well as their rapid increase, and also is highly threatened by floods due to typhoons (Zhang et al. 2011). It is one of the rainiest regions globally, with annual precipitation exceeding 2000 mm (Ding 1992; Zhang et al. 2016). There are two precipitation seasons: the early one is from May to June, mainly due to the onset of the South China Sea (SCS) summer monsoon; the latter is after August, mainly induced by typhoons (Chen et al. 2010; Meng and Wang 2016; Yuan et al. 2019; Li et al. 2020) and other marine weather systems (Ding 1992).
A typhoon has a persistent and remote impact on the PRD’s precipitation. For example, Typhoon Nida in 2016 and Typhoon Pakhar in 2017 caused heavy rainfall in the PRD when they had already landed in Guangxi province and decreased to tropical depressions (Yang and Sha 2017; Pan et al. 2018). The same situation appeared after Typhoon Sarika entered Fujian Province in 2011 (Luo et al. 2012). Weather forecasters sometimes underestimate or even miss a rainfall episode distant from a typhoon. He et al. (2022) reported that the bias of precipitation simulation increased, and the numerical model forecasting ability decreased when a typhoon landed. In the midlatitudes, the westerly trough modulates a precipitation episode stimulated by a remote typhoon (Cong et al. 2011; Schumacher et al. 2011; Chen and Xu 2017). While at low latitudes, a heavy precipitation episode is usually connected with moisture flux, static stability, and large-scale circulation patterns (He et al. 2016; Wu et al. 2020; Meng and Wang 2016; Luo et al. 2020), all of which can be influenced by a distant typhoon.
One of the important factors determining the accuracy of numerical models in predicting precipitation in southern China is the quality of the initial conditions (Zhong et al. 2017, Chen et al. 2019; Lu et al. 2020). Assimilating techniques improve the initial field, further improving precipitation prediction. In situ observations, including conventional meteorological variables, sounding profiles, and radar-derived winds, improved precipitation forecast by 1 mm to approximately 10 mm, mainly due to the correction of specific humidity below 600 hPa (C. Zhang et al. 2009). The precipitation patterns in a numerical model have been significantly improved by assimilating cloud characteristics and corresponding specific humidity (Wu et al. 2018). The Hong Kong Observatory completed the first dropsonde observation of a tropical cyclone (TC) over the South China Sea (Chan et al. 2018). The dropsonde observations revealed an increase in relative humidity in the middle of the troposphere from the southwestern side to the northeastern side of the TC and successfully improved the path prediction of the TC. Besides specific humidity, Zhang et al. (2016) found that the precipitation prediction using the Global/Regional Assimilation and Prediction System (GRAPES) is sensitive to initial wind profiles. Additional radiosonde assimilation conducted by Hattori et al. (2016) around the South China Sea significantly increased the wind speed, temperature, and specific humidity in the lower troposphere, improving the detecting ability of disturbance in a typhoon’s early development. All these studies confirmed that observing atmospheric vertical profiles helps build a realistic initial field for weather forecasts (Inoue et al. 2015; Cardinali and Healy 2014).
The impact of assimilation on forecasting skill lies in the information offered by observation (Weissmann et al. 2011; Zhang and Wang 2018; Feng and Wang 2019; Schindler et al. 2020). High-resolution inner-core reconnaissance observations positively affected the intensity forecast in weak storms by using the operational Hurricane Weather Research and Forecasting (HWRF) Model (Tong et al. 2018). Assimilating dropsonde observations offer a more realistic atmospheric structure of atmospheric rivers, reducing the root-mean-square error in integrated vapor transport and inland precipitation by more than 70% (Lavers et al. 2018; Zheng et al. 2021a,b). Sounding stations in southern China’s coastal and offshore regions are sparse because of the heavy air traffic (Zhang et al. 2016; C. Zhang et al. 2009). Since the water vapor feeding a typhoon rainfall comes from the South China Sea, a sounding observation there might help build a more realistic initial condition for numerical simulations.
Typhoon Mun in 2019 caused heavy rainfall in the PRD region. Additional sounding observations were carried out at the coast of the PRD by researchers at Sun Yat-sen University. This study aims to assess the extent to which the additional soundings can improve the precipitation prediction of the typhoon with the help of assimilation techniques. Descriptions of the observation experiment, study case, numerical experiment design, and diagnosis method are given in section 2. Section 3 presents the experimental results and some discussion. The conclusions are summarized in section 4.
2. Data and method
a. Observations and atmosphere background field
The best track of Typhoon Mun includes 6-hourly track and intensity analyses obtained from the China Typhoon Network (https://tcdata.typhoon.org.cn; Lu et al. 2021), as shown in Fig. 1, and analyzed by the China Meteorological Administration (CMA). CMA also offered the in-site precipitation station data.
The typhoon track, the additional sounding location, and the simulation domain. The red line is the track of Typhoon Mun, with red-filled and red open circles representing its intensity greater and weaker than a tropical storm, respectively. The red asterisk is the location of the radiosonde observation. The parent and nest domains are also given. The terrain of the simulated area (color shaded; m) is painted in the background. A photo of Dr. Sun Genghou releasing the radiosonde is shown in the inset.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
Grid precipitation observation data over Southern China and its neighboring oceans were collected for verification. The Data Service Centre of CMA offers the hourly 0.1° × 0.1° grid precipitation field (Pan et al. 2012) over southern China. It is a merged product of the automatic weather stations and CMORPH (land coverage only). The Climate Prediction Center (CPC) of the National Oceanic and Atmospheric Administration of America offers CPC morphing technique (CMORPH) precipitation fields with 8-km spatial resolution and 30-min temporal resolution (Joyce et al. 2004). It is bilinearly interpolated into 0.1° × 0.1° as the sea portion, then combined with data on land.
During the activity of Typhoon Mun, additional intensive observations by radiosonde (Vaisala RS41-SG) were performed every 12 h near Huizhou Seaport on the eastern coast of Guangdong (Fig. 1) from 2 to 4 July 2019. The air pressure (P; accuracy: 0.04–1.00 hPa), temperature (T; accuracy: 0.15–0.30 K), relative humidity (RH; accuracy: 2%–4%), wind direction (accuracy: 2°), and wind speed (accuracy: 0.15 m s−1) were measured every second.
The atmospheric background field is the operational global assimilation analysis data from the National Environmental Forecast Center [NCEP GDAS/FNL 0.25° Global Tropospheric Analyses and Forecast Grids (FNL data herein); NCEP 2015]. The FNL owns 6-h intervals and continuously assimilates observational data from global telecommunication systems (GTS) and other sources. The FNL data are widely used in numerical simulation and data assimilation of storms and typhoons (Wang et al. 2009; Abhilash et al. 2012; Yang et al. 2012; Di et al. 2020). The analysis variables used in this study include geopotential height, temperature, relative humidity, and horizontal winds.
b. Numerical experiments design
We chose the Weather Research and Forecasting (WRF) Model, version 4.2.0 Skamarock et al. 2019), and its corresponding three-dimensional variational data assimilation (3D-Var) system (Barker et al. 2004) to use in this study. The “Tropical” physics suite option approved by the official WRF User’s Guide (https://www2.mmm.ucar.edu) is referenced, which has been highly tested and has shown pleasing results for low latitudes (Zhang et al. 2018). The detailed model configurations, including physics options in the Tropical suite, are listed in Table 1. Two domains were set in this study and communicated with each other (Fig. 1). The horizontal grid resolution has been tested, and the resolution of a parent domain (D01) of 18 km and nest (D02) of 6 km seemed the best. A higher resolution, like 4 km (convection resolving), gave a worse precipitation simulation.
WRF configuration.
We designed four numerical experiments (Table 2) to analyze the impact of assimilation on rainfall simulation. The control experiment (CTL) uses the first guess generated from FNL data as the initial field and boundary conditions to predict ongoing precipitation. The other three are sensitivity experiments (SEs) using new initial field and boundary conditions by assimilating the additional sounding data into the first guess. The initial time of CTL and the sensitive experiment with assimilation of all variables (SEA) are 0000 and 1200 UTC, followed by a 12-h simulation. For assimilation, the dense vertical data are subsampled into about 106 layers with a spacing of 5 hPa for pressure levels below 900 hPa and 10 hPa for pressure levels between 900 and 50 hPa. At the initial step, SEA assimilated all variables from additional radiosonde observations, including wind, temperature, and humidity profiles from the surface to 50 hPa, based on vertical position information (height, pressure). In contrast, the second SE (SED) used the initial conditions by assimilating the observed wind profiles; and the third (SET) assimilated thermodynamical variables, including air temperature and humidity, and was based on vertical position information.
Experimental design; U, V, T, Td, and RH represent the zonal and meridional wind components, temperature, dewpoint temperature, and relative humidity, respectively.
For the sake of calculation economy and fast repetition, the NCEP background error covariance (option CV3 for 3D-Var) estimated by the NMC method (Parrish and Derber 1992) is chosen for assimilation. The background-error scale length and variance values were tuned. According to the test results (not shown), the variance scaling factor is reduced by 50%, the horizontal length scale factor is increased by 50%, and the vertical length scale factor is increased by 50%. The 3D-Var default observation errors are chosen. In addition, observations with differences relative to FNL greater than three times the observation error will not be included in assimilation.
c. Diagnosing method
3. Results and discussion
a. Case overview
Typhoon Mun formed at 1200 UTC 2 July 2019 in the western part of the South China Sea and moved westward after its birth (Fig. 1). The intensity of Mun was of a tropical storm, with a maximum wind speed of 18 m s−1 near its center. It landed on Hainan Island at about 0000 UTC 3 July and then entered the Beibu Gulf at about 1200 UTC 3 July. After its second landing in Vietnam, Typhoon Mun gradually weakened into a tropical depression.
Using FNL data, the atmospheric environmental fields of 500 and 850 hPa distinctly illustrate the typhoon’s low geopotential height, robust cyclonic circulation, and high potential temperature, as well as strong specific humidity (Fig. 2). The focus area of this study is around the PRD (the black box in Fig. 2), located to the northeast of the typhoon during the observation experiment. The southeast wind on the east side of the typhoon brought plenty of water vapor from the South China Sea to the PRD, resulting in heavy rainfall there (Fig. 2c).
Atmospheric circulation analysis of FNL at (top) 500 and (bottom) 850 hPa. The solid blue line is the geopotential height with an interval of 20 gpm. The barbs indicate the wind, with a full barb of 4 m s−1. Color fills show (a),(b) the potential temperature (K) and (c),(d) specific humidity (g kg−1). The red rectangle indicates the focus area in this study.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
When Typhoon Mun moved far away from the South China Sea to the Beibu Gulf at 1200 UTC 3 July, the potential temperature and specific humidity at the PRD decreased, and the geopotential height increased. Moreover, the wind direction at the PRD turned southwest at 500 and 850 hPa (Figs. 2b,d), exhibiting the influence of the subtropical high (5880 gpm). Based on a rough analysis of these changes, precipitation in the PRD is expected to experience a dramatic decrease. However, observations of precipitation indicate that this weakening process is occurring at a slow pace.
Typhoon Mun caused heavy rainfall in the PRD area. Most meteorology stations within the PDR recorded 3-day accumulated precipitation > 70 mm; precipitation at seven of them even exceeded 105 mm. The eastern coast of Guangdong is the primary precipitation area induced by Typhoon Mun (Fig. 3a). The core period is from 1200 UTC 2 July to 0000 UTC 4 July. Take the following three stations as examples: Lufeng (22.97°N, 115.65°E), Shanwei (22.80°N, 115.37°E), and Haifeng (23.02°N, 115.31°E) (Fig. 3b). Their precipitation records show accumulated precipitation > 30 mm per 12 h, reaching the standard of storm rainfall grade (Qiao et al. 2012). Cong et al. (2011) divided tropical cyclone–induced precipitation into TC circulation rainfall and remote precipitation. In this study, the sounding site is at a remote rainfall center relative to the distance of the typhoon center.
Precipitation records from (a) meteorology stations from 2 to 5 Jul, and (b) 12-h accumulated precipitation at three stations.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
b. Precipitation prediction assessment
Near the PRD, there were two parallel rain belts: the intensive coastal rain belt and the weak inland one (Figs. 3 and 4a). The predictions from the CTL and SEA captured the two rain belts but underestimated the precipitation intensity (Figs. 4b,c), especially missing the western portion of the coastal rain belt. Considering the location of the sounding observation, we focused on the assimilation effect in the eastern coastal rain belt.
(a) Observed and (b) simulated, (c) precipitation patterns from 1200 UTC 2 Jul to 0000 UTC 4 Jul (core period). The black rectangle and asterisk represent the focus area and the sounding location on the east coast of Guangdong, respectively. Average 12-h accumulated precipitation over the focus area, including observation (black line) and simulations of the CTL (blue line) and the SEA (red line), as well as the differences between the CTL and the SEA (green bars) (d). Initial time 0312 represents the time of 1200 UTC 3 Jul, and the same for others.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
The observed average precipitation in the eastern coastal rainband peaked at 24.8 mm in the 12 h after 1200 UTC 2 July, and then gradually decreased to 16.5 mm in the 12 h after 1200 UTC 3 July (Fig. 4d). Compared to observations, the CTL captures the rain peak time, but the accumulated precipitation is 17.5 mm. Meanwhile, the CTL overestimates the rate of decrease in precipitation after the peak, with its 12-h forecast accumulated precipitation of 8.6 and 2.1 mm after 0000 and 1200 UTC 3 July, respectively. When the typhoon is far from the PRD region, underestimating rainfall intensity or even missing rainfall by a numerical model has been reported before (Yang and Sha 2017; Pan et al. 2018; Luo et al. 2012). This problem is partly corrected in the SEA, with forecast accumulated precipitation of about 12.0 and 6.1 mm after these two moments. Although rainfall in both the SEA and CTL is underestimated compared to observations (OBS), the SEA shows some improvement compared with CTL, especially after 1200 UTC 3 July (see bar graph in Fig. 4d).
c. Atmospheric vertical structure
We divided the assimilation effects into local and nonlocal ones. The local effect is the change in the atmospheric vertical structure around the sounding observation site (Figs. 5 and 6). Since the SEA is closer to the observation than the CTL, the difference between the SEA and CTL reflects the bias in the initial field of the CTL.
The profiles of potential temperature (θ; solid line) and specific humidity (q; dotted line) from the sounding observation, CTL, and SEA simulations at (a) 1200 UTC 2 Jul, (b) 0000 UTC 3 Jul, and (c) 1200 UTC 3 Jul. (d)–(f) The potential temperature differences (θ_dif) and specific humidity differences (q_dif) of SEA minus CTL.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
The wind profiles and their differences in the x direction (u; solid line) and y direction (υ; dotted line) from the observation (black), CTL (blue), and SEA (red) at (a) 1200 UTC 2 Jul, (b) 0000 UTC 3 Jul, and (c) 1200 UTC 3 Jul. (d)–(f) The u differences (u_dif) and υ differences (υ_dif) of SEA minus CTL.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
At 1200 UTC 2 July (Fig. 5d), when peak rainfall started, the SEA gives a warmer but drier atmosphere layer below 800 hPa and from 700 to 500 hPa, and a colder but wetter one near 600 hPa, compared with the CTL. At this moment, assimilation increases the static instability under 700 hPa but decreases the precipitable water (i.e., vertically integrated humidity) under 500 hPa. These two opposite effects cancel each other and make the rainfall in the following 12 h identical between the SEA and the CTL.
In contrast, at 0000 UTC 3 July, when the rainfall peak passed, the assimilation caused a wet and warm effect under 850 hPa relative to the CTL, causing increased precipitation in the following 12 h. At 1200 UTC 3 July, although assimilation raised the air temperature by about 1 K above 750 hPa, increasing the static stability, the air in the SEA was significantly wetter than the CTL (by more than 1.5 g kg−1 under 800 hPa), still favorable to trigger condensation in the lower troposphere. As a result, the precipitation in the SEA during the following 12 h is still more intensive than that in the CTL. Multiple reanalysis datasets (e.g., ERA5, JRA-55, NCEP–DOE) underestimated the temperature and humidity of the lower atmosphere in the SCS and its surrounding areas (Hattori et al. 2016; Chan et al. 2018; Deng et al. 2022).
The assimilation also corrected the wind profile. FNL underestimated the wind speed under 925 hPa during the observation period, and such underestimation has been significantly corrected in the SEA. Over time, the winds at 700–925 hPa shift from southeast to south and then to the southwest (Figs. 6a–c). However, the change in wind direction below 925 hPa is slower than above. At 1200 UTC 3 July, wind analysis by FNL is southerly, while observations indicate southeasterly wind. Assimilation corrected this bias of FNL (Figs. 6c,f). At 1200 UTC 2 July, assimilation reduces the vertical wind shear under 700 hPa (Figs. 6a,d), while at the other two moments, assimilation seems to enhance the vertical shear of u but to reduce the vertical shear in υ.
The assimilation helps provide a more realistic atmospheric boundary layer (ABL) structure and the turbulent transport within it. We define the atmospheric boundary layer height (ABLH) as the level where the bulk Richardson number [Eq. (4)] reaches 0.25 from the surface upward. At 1200 UTC 2 July, ABLHs from observation, the CTL, and the SEA were 278, 240, and 282 m, respectively. The turbulent transport within the ABL is similar between the CTL and the SEA. In contrast, at 1200 UTC 3 July, the ABLHs from observation, the CTL, and the SEA were 726, 416, and 603 m, respectively. Since the humidity in the SEA is much greater than in the CTL, an intensified vertical transport due to a deeper ABL in the SEA is more likely to trigger strong condensation, partly explaining the more intensive precipitation in the SEA than in the CTL. Therefore, the local effect of assimilation is helping give a more realistic ABL structure.
The SEAs showed that the thermal structure under 850 hPa, containing the planetary boundary layer (PBL), is essential to precipitation prediction. Therefore, the PBL parameterization used in FNL needs to be noted. A hybrid eddy diffusivity–mass flux (EDMF) PBL parameterization was developed and implemented in FNL from January 2015 to December 2019 (Siebesma et al. 2007; Han et al. 2016). However, some studies pointed out that this scheme might be more proper for a strongly unstable PBL than a stable PBL (Chen 2022; Griffin et al. 2021), like that within the peak precipitation period (1200 UTC 2 July). However, when the PBL became stable after the precipitation peak, the EDMF scheme significantly underestimated the PBL height and turbulence intensity. Meanwhile, assimilation partly corrected these underestimations, thereby improving the ongoing precipitation simulation. Therefore, the assimilation of sounding observations can help improve precipitation prediction by correcting the PBL structures provided by a model.
d. Atmospheric horizontal field
Assimilation causes a regional synchronous adjustment of the atmospheric field. Taking 925 hPa as an example, Fig. 7 shows the increment of the atmospheric field of the SEA minus the CTL from the initial moment to the forecast of 9 h. At the initial moment, assimilation causes an increment in specific humidity of about 1.8 g kg−1 at the sounding position, and the increment, although it decreases with distance from the sounding position, spreads to the whole domain. At the same time, a southeast wind increment of about 1.0 m s−1 (Fig. 7a) occupies most domain areas and seems stronger over land than sea. The wind increment shows a cyclonic vortex west of the sounding. At the 3-h forecast, the specific humidity increment weakens around the sounding position. The wind increment appears as a southerly enhancement (Fig. 7b). Assimilation leads to increased moisture in the air within the PRD region where the sounding location is situated. Additionally, the southern side of the PRD (offshore) experiences an increase in humidity. Subsequently, onshore winds transport this water vapor toward the PRD. A stronger supply of water vapor contributes to heavier precipitation in the PRD region.
The increment of specific humidity (shaded; g kg−1) and wind field (barb; full barb indicates 1 m s−1) of SEA relative to CTL. The red asterisk is the location of the radiosonde observation.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
At the 6-h forecast, the humidity increment is further weakened, and the wind increment is presented as southwesterly (Fig. 7c) like the 850-hPa wind in CTL (Fig. 2d). At the 9-h forecast, the specific humidity increment shows no identical pattern over the ocean but still positive over the continent. At the same time, the wind increment is becoming chaotic (Fig. 7d). After 6 h, the impact of water vapor changes caused by assimilation on the PRD region tends to dissipate, indicating that the change in precipitation in the area also tends to disappear. It can also be noted that downstream of the sounding position, that is, at the inland rain zone location (near 24.5°N, 113°E), assimilation also caused an increase in humidity in the area (Figs. 7a–c) and led to an increase in precipitation simulations in this area (Figs. 4b,c).
e. Moisture flux patterns
Since surface evaporation is usually slight during rainfall, precipitation generally equals the horizontal MFC (Zhu et al. 2007). The MFC difference between the SEA and the CTL is mainly concentrated at 925 and 950 hPa in the eastern PRD at 1200 UTC 3 July, close to the observation site (Fig. 8), favoring precipitation enhancement. The MFC differences at 950 hPa are the largest, about 3 × 10−3 g kg−1 s−1, or 60% of the background. It becomes small at 925 hPa, with some negative values in the PRD region. The MFC difference is negative at 850 and 700 hPa in the PRD region (Figs. 8a,b), corresponding to an enhanced divergent circulation capping on the condensation center. Du and Chen (2019) reported that the mesoscale lifting due to low-level convergence (950 hPa) and midlevel divergence (700 hPa) is essential for heavy rainfall events near the PRD. Therefore, the opposite response in MFC at 950 and 700 hPa due to assimilation will likely enhance the precipitation intensity.
The sum of the 12-h forecast moisture fluxes convergence (10−4 g kg−1 s−1) by the CTL on different pressure layers initially at 1200 UTC 3 Jul (contours), and the difference between SEA and CTL (shaded; SEA minus CTL). The black rectangle represents the east coastal rain belt area, and the red asterisk is the location of the radiosonde observation.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
To show the duration of the assimilation effect, we analyzed the time series of regional averaged MFC differences (SEA minus CTL) after 1200 UTC 3 July (Fig. 9). After 2–3 h of simulation, the precipitation difference peaks at 0.8 mm h−1, and the MFC differences at 950 and 925 hPa both increase to their peaks (∼3 × 10−4 g kg−1 s−1). The MFC difference at 925 hPa gradually decreased to 0 at 1900 UTC, when the MFC differences at 850 and 950 hPa were still opposite and significant. At the same time, the rainfall difference was ignorable (Fig. 9b). Therefore, it seems that the MFC at 925 hPa is critical for PRD precipitation under the influence of a remote typhoon.
Hourly moisture flux (a) convergence difference and (b) precipitation difference of SEA minus CTL after 1200 UTC 3 Jul.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
We further divided the MFC difference into dynamic and thermodynamical terms following Eq. (3) (Fig. 9). At 925 and 950 hPa, the dynamic term contributes to the main MFC differences, but the thermodynamical term is more significant at 850 hPa. Generally, the thermodynamical term’s contribution is less significant than that of the dynamic term. Therefore, the dynamic response seems more sensitive and important than the thermodynamical response when the initial fields have been changed. Here, we note that the dynamic variables might not contribute to the primarily dynamic response.
f. Sensitivity experiments of assimilated observation factors
Hou et al. (2013) demonstrated that the choices of variables in assimilation are essential for the precipitation forecasting skill in southern China. Meanwhile, since radiosonde observation is improper for regions with heavy air traffic, a wind radar and microwave radiometer usually independently monitor the wind and thermal profiles around PRD regions. Therefore, to design a more functional observation network around the PRD, it is important to figure out which kind of observation is more helpful for typhoon prediction. For this purpose, we designed the SED and SET.
The MFC differences referring to the CTL are presented in Fig. 10. The differences in MFC and precipitation from the SET are comparable to those of the SEA (Figs. 9 and 10). In contrast, the precipitation difference is close to zero (Fig. 10b), and the MFC difference at 925 hPa remained negative in the SED until 1600 UTC (Fig. 10a). Therefore, it seems that the atmospheric thermodynamical variables are more critical than dynamic ones when using assimilation techniques to improve precipitation prediction. Moreover, if people want to improve typhoon rainfall prediction by upgrading the atmosphere monitoring networks, a radiometer (which can give atmospheric thermal profiles with high frequency) seems a better choice than a wind radar. However, it should be noted that the precipitation performance of the SET and the SED is worse than that of the SEA. In other words, assimilation with atmospheric thermodynamical and dynamic variables is still superior to just considering one of them. Similar results have been reported by Zhao and Xue (2009), who found that the best prediction of Hurricane Ike (2008) is given when both reflectivity and radial velocity data are assimilated.
Hourly moisture flux (a) convergence difference and (b) precipitation difference of the SET and the SED relative to the CTL after 1200 UTC 3 Jul.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0024.1
The enhancement of MFC in the SEA (Fig. 9a) and the SET (Fig. 10a) relative to the CTL was mainly contributed by the atmospheric wind fields’ response. The atmospheric thermal structure (due to assimilation) corresponds to atmospheric available potential energy, which is usually two orders of magnitude greater than the kinematic energy, that is, wind fields. Therefore, it is easy to understand that the SET caused a much more significant change in MFC and precipitation than did the SED. Meanwhile, the MFC change in the SED is opposite to that in the SET at 925 hPa (Fig. 10a). Zheng et al. (2021a,b) pointed out that the assimilation impact of dropsonde observations also depends on how the observations are filling the existing observations gaps. Zhao and Xue (2009) found that the assimilation of observed wind fields helps forecast the hurricane track, and the radar reflectivity, connecting to condensation heating, helps improve the rainfall forecast. Therefore, assimilating the observed thermal structure in the lower troposphere is helpful for typhoon precipitation prediction (Chen et al. 2010).
The assimilation effect, including the MFC and precipitation differences of the SET and the SEA relative to the CTL disappears within the PRD after 6 h. However, the impact of assimilation is significant even after 9 h of integration (Fig. 7). One explanation is that the sounding observation is within the rain center, so the atmospheric advection under a wind speed of about 10 m s−1 (Fig. 2) will erase the local assimilation effect in a few hours. Therefore, additional sounding observations at the upstream area of the rain center seemed a good choice to improve the assimilation effect in rainfall prediction. Noting that the precipitation from SEA is still underestimated compared to observation on 3 July (Fig. 4), a further improvement of precipitation prediction might rely on a better-designed sounding observation network within the coastal ocean.
4. Summary and conclusions
The forecasting of typhoon-related precipitation is challenging. We studied the rainfall in the PRD regions caused by Typhoon Mun in 2019. A series of numerical experiments were designed to evaluate the impact of assimilating one site of radiosonde observation over the coastal area of the PRD. The CTL directly used the FNL data as the initial conditions, while the SEs used the initial fields by assimilating the radiosonde observation. The SEA reduced the bias in the precipitation forecast of the CTL by 24%.
The assimilation can significantly correct the local atmospheric structure under 850 hPa and give a more realistic PBL structure than the CTL, which was named the local effect. The unreasonable PBL structure in FNL is mainly from its hybrid EDMF schemes. The PBL is higher and moister in the SEs than the CTL after the rain peak, which is beneficial for sustaining rainfall intensity.
Besides the local effect, assimilation also caused a nonlocal effect by modulating the spatial patterns of the atmospheric circulation, as well as by modulating the MFC at the lower troposphere. Moreover, the dynamic (wind) field response contributed to the primary change in MFC. However, assimilating the meteorological thermal profiles, including temperature and humidity, caused a much more significant and positive modulation in MFC than assimilating wind profiles only.
The experimental results demonstrate that the assimilation has an impact of approximately 6 h. If observations are conducted at 6-h intervals and cycling 3D-Var is employed, it could potentially yield improved precipitation simulation (Hsiao et al. 2012; Lagasio et al. 2019). It should be noted that the importance of assimilating atmospheric variables might rely on the choice of assimilation techniques. For example, we found that a hybrid, four-dimensional, ensemble-variational data assimilation (hybrid 4D-EnVar) shows a different importance of the wind profile compared to this study (Weissmann et al. 2011; Zheng et al. 2021b). Assessing different assimilation techniques will be finished in our subsequent work.
Acknowledgments.
The authors are grateful to the team members of Sun Yat-sen University’s 2019 South China Sea voyage for their work. NCAR is also acknowledged for the WRF, 3D-VAR systems, and FNL data. This study was supported by the National Key R&D Program of China (2020YFA0608804, 2019YFA0607004), the Program of Marine Economy Development Special Fund under the Department of Natural Resources of Guangdong Province [GDNRC(2022)18], and the Guangdong Basic and Applied Basic Research Foundation (2020B1515020025).
Data availability statement.
The track data of Typhoon Mun used during this study are openly available at https://tcdata.typhoon.org.cn, with technical notes from Lu et al. (2021). The CMORPH precipitation data are openly available at ftp://ftp.cpc.ncep.noaa.gov/precip. The FNL data are openly available at https://rda.ucar.edu/datasets/ds083.3. Merged precipitation grid data, additional radiosonde data, and precipitation station data archiving for supporting this paper can be found online at https://doi.org/10.6084/m9.figshare.22046900; https://doi.org/10.6084/m9.figshare.22046963; and https://doi.org/10.6084/m9.figshare.22047005.
REFERENCES
Abhilash, S., A. K. Sahai, K. Mohankumar, J. P. George, and S. Das, 2012: Assimilation of Doppler weather radar radial velocity and reflectivity observations in WRF-3DVAR system for short-range forecasting of convective storms. Pure Appl. Geophys., 169, 2047–2070, https://doi.org/10.1007/s00024-012-0462-z.
Banacos, P. C., and D. M. Schultz, 2005: The use of moisture flux convergence in forecasting convective initiation: Historical and operational perspectives. Wea. Forecasting, 20, 351–366, https://doi.org/10.1175/WAF858.1.
Barker, D. M., W. Huang, Y.-R. Guo, A. J. Bourgeois, and Q. N. Xiao, 2004: A three-dimensional variational data assimilation system for MM5: Implementation and initial results. Mon. Wea. Rev., 132, 897–914, https://doi.org/10.1175/1520-0493(2004)132<0897:ATVDAS>2.0.CO;2.
Cardinali, C., and S. Healy, 2014: Impact of GPS radio occultation measurements in the ECMWF system using adjoint-based diagnostics. Quart. J. Roy. Meteor. Soc., 140, 2315–2320, https://doi.org/10.1002/qj.2300.
Chan, P. W., N. G. Wu, C. Z. Zhang, W. J. Deng, and K. K. Hon, 2018: The first complete dropsonde observation of a tropical cyclone over the South China Sea by the Hong Kong Observatory. Weather, 73, 227–234, https://doi.org/10.1002/wea.3095.
Chen, L., and Y. Xu, 2017: Review of typhoon very heavy rainfall in China. Meteor. Environ. Sci., 40, 3–10, https://doi.org/10.16765/j.cnki.1673-7148.2017.01.001.
Chen, L., Y. Li, and Z. Cheng, 2010: An overview of research and forecasting on rainfall associated with landfalling tropical cyclones. Adv. Atmos. Sci., 27, 967–976, https://doi.org/10.1007/s00376-010-8171-y.
Chen, T., J. Sun, Y. Chen, Y. Guo, and J. Xu, 2019: Study on the numerical predictivity of localized severe meso-scale rainstorm in Guangzhou on 7 May 2017. Meteor. Mon., 45, 1199–1212, https://doi.org/10.7519/j.issn.1000-0526.2019.09.002.
Chen, X., 2022: How do planetary boundary layer schemes perform in hurricane conditions: A comparison with large‐eddy simulations. J. Adv. Model. Earth Syst., 14, e2022MS003088, https://doi.org/10.1029/2022MS003088.
Chou, C., and C.-W. Lan, 2012: Changes in the annual range of precipitation under global warming. J. Climate, 25, 222–235, https://doi.org/10.1175/JCLI-D-11-00097.1.
Cong, C., L. Chen, X. Lei, and Y. Li, 2011: An overview on the study of tropical cyclone remote rainfall (in Chinese). J. Trop. Meteor., 27, 264–270, https://doi.org/10.3969/j.issn.1004-4965.2011.02.016.
Deng, H., and Coauthors, 2022: Assessment on the water vapor flux from atmospheric reanalysis data in the South China Sea on 2019 summer. J. Hydrometeor., 23, 847–858, https://doi.org/10.1175/JHM-D-21-0210.1.
Di, Z., Q. Duan, C. Shen, and Z. Xie, 2020: Improving WRF typhoon precipitation and intensity simulation using a surrogate-based automatic parameter optimization method. Atmosphere, 11, 89, https://doi.org/10.3390/atmos11010089.
Ding, Y., 1992: Summer monsoon rainfalls in China. J. Meteor. Soc. Japan, 70, 373–396, https://doi.org/10.2151/jmsj1965.70.1B_373.
Du, Y., and G. Chen, 2019: Heavy rainfall associated with double low-level jets over southern China. Part II: Convection initiation. Mon. Wea. Rev., 147, 543–565, https://doi.org/10.1175/MWR-D-18-0102.1.
Ek, M. B., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, V. Koren, G. Gayno, and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108, 8851, https://doi.org/10.1029/2002JD003296.
Feng, J., and X. Wang, 2019: Impact of assimilating upper-level dropsonde observations collected during the TCI field campaign on the prediction of intensity and structure of Hurricane Patricia (2015). Mon. Wea. Rev., 147, 3069–3089, https://doi.org/10.1175/MWR-D-18-0305.1.
Griffin, S. M., J. A. Otkin, S. E. Nebuda, T. L. Jensen, P. S. Skinner, E. Gilleland, T. A. Supinie, and M. Xue, 2021: Evaluating the impact of planetary boundary layer, land surface model, and microphysics parameterization schemes on cold cloud objects in simulated GOES‐16 brightness temperatures. J. Geophys. Res. Atmos., 126, e2021JD034709, https://doi.org/10.1029/2021JD034709.
Han, J., M. L. Witek, J. Teixeira, R. Sun, H.-L. Pan, J. K. Fletcher, and C. S. Bretherton, 2016: Implementation in the NCEP GFS of a hybrid eddy-diffusivity mass-flux (EDMF) boundary layer parameterization with dissipative heating and modified stable boundary layer mixing. Wea. Forecasting, 31, 341–352, https://doi.org/10.1175/WAF-D-15-0053.1.
Hattori, M., J. Matsumoto, S.-Y. Ogino, T. Enomoto, and T. Miyoshi, 2016: The impact of additional radiosonde observations on the analysis of disturbances in the South China Sea during VPREX2010. SOLA, 12, 75–79, https://doi.org/10.2151/sola.2016-018.
He, B., Z. Yu, Y. Tan, Y. Shen, and Y. Chen, 2022: Rainfall forecast errors in different landfall stages of super Typhoon Lekima (2019). Front. Earth Sci., 16, 34–51, https://doi.org/10.1007/s11707-021-0894-9.
He, L., T. Chen, and Q. Kong, 2016: A review of studies on prefrontal torrential rain in South China. J Appl. Meteor. Sci., 27, 559–569, https://doi.org/10.11898/1001-7313.20160505.
Hong, S., and J. J. Lim, 2006: The WRF single-moment 6-class microphysics scheme (WSM6). Asia-Pac. J. Atmos. Sci., 42, 129–151.
Hong, S.-Y., N. Yign, and D. Jimy, 2006: A new vertical diffusion package with an explicit treatment of entrainment processes. Mon. Wea. Rev., 134, 2318–2341, https://doi.org/10.1175/MWR3199.1.
Hou, T., F. Kong, X. Chen, and H. Lei, 2013: Impact of 3DVAR data assimilation on the prediction of heavy rainfall over southern China. Adv. Meteor., 2013, 129642, https://doi.org/10.1155/2013/129642.
Hsiao, L.-F., D.-S. Chen, Y.-H. Kuo, Y.-R. Guo, T.-C. Yeh, J.-S. Hong, C.-T. Fong, and C.-S. Lee, 2012: Application of WRF 3DVAR to operational typhoon prediction in Taiwan: Impact of outer loop and partial cycling approaches. Wea. Forecasting, 27, 1249–1263, https://doi.org/10.1175/WAF-D-11-00131.1.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Inoue, J., A. Yamazaki, J. Ono, K. Dethloff, M. Maturilli, R. Neuber, P. Edwards, and H. Yamaguchi, 2015: Additional Arctic observations improve weather and sea-ice forecasts for the Northern Sea route. Sci. Rep., 5, 16868, https://doi.org/10.1038/srep16868.
Joyce, R. J., J. E. Janowiak, P. A. Arkin, and P. Xie, 2004: CMORPH: A method that produces global precipitation estimates from passive microwave and infrared data at high spatial and temporal resolution. J. Hydrometeor., 5, 487–503, https://doi.org/10.1175/1525-7541(2004)005<0487:CAMTPG>2.0.CO;2.
Lagasio, M., F. Silvestro, L. Campo, and A. Parodi, 2019: Predictive capability of a high-resolution hydrometeorological forecasting framework coupling WRF cycling 3DVAR and continuum. J. Hydrometeor., 20, 1307–1337, https://doi.org/10.1175/JHM-D-18-0219.1.
Lavers, D. A., M. J. Rodwell, D. S. Richardson, F. M. Ralph, J. D. Doyle, C. A. Reynolds, V. Tallapragada, and F. Pappenberger, 2018: The gauging and modeling of rivers in the sky. Geophys. Res. Lett., 45, 7828–7834, https://doi.org/10.1029/2018GL079019.
Li, H., Q. Wan, D. Peng, X. Liu, and H. Xiao, 2020: Multiscale analysis of a record-breaking heavy rainfall event in Guangdong, China. Atmos. Res., 232, 104703, https://doi.org/10.1016/j.atmosres.2019.104703.
Lu, C., L. Lin, and F. Zhou, 2020: Analysis of the source of forecast errors for a heavy precipitation in the southwest of Guangdong province (in Chinese). Chin. J. Atmos. Sci., 44, 1337–1348, https://doi.org/10.3878/j.issn.1006-9895.2008.20130.
Lu, X., H. Yu, M. Ying, B. Zhao, S. Zhang, L. Lin, L. Bai, and R. Wan, 2021: Western North Pacific tropical cyclone database created by the China Meteorological Administration. Adv. Atmos. Sci., 38, 690–699, https://doi.org/10.1007/s00376-020-0211-7.
Luo, C., J. He, and Y. Zhang, 2012: Causal analysis of the torrential rain associated with a dissipating away tropical cyclone. Guangdong Meteor., 34, 1–5, https://doi.org/10.3969/j.issn.1007-6190.2012.03.001.
Luo, Y., and Coauthors, 2020: Science and prediction of heavy rainfall over China: Research progress since the reform and opening-up of new China. J. Meteor. Res., 34, 427–459, https://doi.org/10.1007/s13351-020-0006-x.
Meng, W., and Y. Wang, 2016: A diagnostic study on heavy rainfall induced by landfalling Typhoon Utor (2013) in South China: 2. Postlandfall rainfall. J. Geophys. Res. Atmos., 121, 12 803–12 819, https://doi.org/10.1002/2015JD024647.
NCEP, 2015: NCEP GDAS/FNL 0.25 Degree Global Tropospheric Analyses and Forecast Grids. Research Data Archive at the NCAR, Computational and Information Systems Laboratory, accessed 23 April 2020, https://doi.org/10.5065/D65Q4T4Z.
Pan, Q., W. Li, and R. Guo, 2018: Analysis of the causation of a heavy rain at the rear portion of Typhoon Pakhar (1714). Guangdong Meteor., 40, 6–10, https://doi.org/10.3969/j.issn.1007-6190.2018.04.002.
Pan, Y., Y. Shen, J. Yu, and P. Zhao, 2012: Analysis of the combined gauge-satellite hourly precipitation over China based on the OI technique. Acta Meteor. Sin., 70, 1381–1389, https://doi.org/10.11676/qxxb2012.116.
Parrish, D. F., and J. C. Derber, 1992: The National Meteorological Center’s spectral statistical-interpolation analysis system. Mon. Wea. Rev., 120, 1747–1763, https://doi.org/10.1175/1520-0493(1992)120<1747:TNMCSS>2.0.CO;2.
Qiao, L., Y. Li, J. Fu, C. Tian, B. Bi, and Q. Zhou, 2012: The State Standard of the People’s Republic of China–Grade of Precipitation GB/T 28592–201 (in Chinese). Standards Press of China, 3 pp.
Schindler, M., M. Weissmann, A. Schäfler, and G. Radnoti, 2020: The impact of dropsonde and extra radiosonde observations during NAWDEX in autumn 2016. Mon. Wea. Rev., 148, 809–824, https://doi.org/10.1175/MWR-D-19-0126.1.
Schumacher, R. S., T. J. Galarneau Jr., and L. F. Bosart, 2011: Distant effects of a recurving tropical cyclone on rainfall in a midlatitude convective system: A high-impact predecessor rain event. Mon. Wea. Rev., 139, 650–667, https://doi.org/10.1175/2010MWR3453.1.
Seidel, D. J., Y. Zhang, A. Beljaars, J.-C. Golaz, A. R. Jacobson, and B. Medeiros, 2012: Climatology of the planetary boundary layer over the continental United States and Europe. J. Geophys. Res., 117, D17106, https://doi.org/10.1029/2012JD018143.
Sicard, M., C. Pérez, F. Rocadenbosch, J. M. Baldasano, and D. García-Vizcaino, 2006: Mixed-layer depth determination in the Barcelona coastal area from regular lidar measurements: Methods, results and limitations. Bound.-Layer Meteor., 119, 135–157, https://doi.org/10.1007/s10546-005-9005-9.
Siebesma, A. P., P. M. M. Soares, and J. Teixeira, 2007: A combined eddy-diffusivity mass-flux approach for the convective boundary layer. J. Atmos. Sci., 64, 1230–1248, https://doi.org/10.1175/JAS3888.1.
Skamarock, W. C., and Coauthors, 2019: A description of the Advanced Research WRF Model version 4. NCAR Tech. Note NCAR/TN-556+STR, 145 pp., https://doi.org/10.5065/1dfh-6p97.
Tong, M., and Coauthors, 2018: Impact of assimilating aircraft reconnaissance observations on tropical cyclone initialization and prediction using operational HWRF and GSI ensemble–variational hybrid data assimilation. Mon. Wea. Rev., 146, 4155–4177, https://doi.org/10.1175/MWR-D-17-0380.1.
Wang, Y., Y. Wang, and H. Fudeyasu, 2009: The role of Typhoon Songda (2004) in producing distantly located heavy rainfall in Japan. Mon. Wea. Rev., 137, 3699–3716, https://doi.org/10.1175/2009MWR2933.1.
Weissmann, M., and Coauthors, 2011: The influence of assimilating dropsonde data on typhoon track and midlatitude forecasts. Mon. Wea. Rev., 139, 908–920, https://doi.org/10.1175/2010MWR3377.1.
Wu, N., Z. Wen, W. Deng, L. Lin, and G. Chen, 2020: Advances in warm-sector heavy rainfall during the first rainy season in South China. J. Meteor. Sci., 40, 605–616, https://doi.org/10.3969/2020jms.0077.
Wu, Y., W. Meng, D. Chen, W. Lin, and L. Zhu, 2018: A study of the impact of initial conditions on the predictability of a warm-sector torrential rain over South China. Acta Meteor. Sin., 76, 323–342, https://doi.org/10.11676/qxxb2018.001.
Yang, G., and T. Sha, 2017: Analysis of the cause of intense rain in Pearl River delta long after Typhoon Nida (1604) left. Guangdong Meteor., 39 (3), 1–5, https://doi.org/10.3969/j.issn.1007-6190.2017.03.001.
Yang, S.-C., E. Kalnay, and T. Miyoshi, 2012: Accelerating the EnKF spinup for typhoon assimilation and prediction. Wea. Forecasting, 27, 878–897, https://doi.org/10.1175/WAF-D-11-00153.1.
Yuan, J., J. Lü, D. Feng, M. Mao, T. Feng, J. Yin, and L. Zuo, 2019: Heavy rainfall events in southern China associated with tropical cyclones in the Bay of Bengal: A case study. Atmosphere, 10, 574, https://doi.org/10.3390/atmos10100574.
Zhang, C., and Y. Wang, 2017: Projected future changes of tropical cyclone activity over the western North and South Pacific in a 20-km-mesh regional climate model. J. Climate, 30, 5923–5941, https://doi.org/10.1175/JCLI-D-16-0597.1.
Zhang, C., and Y. Wang, 2018: Why is the simulated climatology of tropical cyclones so sensitive to the choice of cumulus parameterization scheme in the WRF model? Climate Dyn., 51, 3613–3633, https://doi.org/10.1007/s00382-018-4099-1.
Zhang, C., Q. Wan, Y. Huang, Z. Chen, and W. Ding, 2009: Experiments on assimilation of initial values in numerical prediction of a warm-sector precipitation in South China. J. Trop. Meteor., 15, 73–77, https://doi.org/10.3969/j.issn.1006-8775.2009.01.012.
Zhang, L., J. Gong, and R. Wang, 2018: Diagnostic analysis of various observation impacts in the 3DVAR assimilation system of global GRAPES. Mon. Wea. Rev., 146, 3125–3142, https://doi.org/10.1175/MWR-D-17-0182.1.
Zhang, Q., C.-Y. Xu, Y. D. Chen, and T. Yang, 2009: Spatial assessment of hydrologic alteration across the Pearl River Delta, China, and possible underlying causes. Hydrol. Processes, 23, 1565–1574, https://doi.org/10.1002/hyp.7268.
Zhang, Q., W. Zhang, Y. D. Chen, and T. Jiang, 2011: Flood, drought, and typhoon disasters during the last half-century in the Guangdong province, China. Nat. Hazards, 57, 267–278, https://doi.org/10.1007/s11069-010-9611-9.
Zhang, X., Y. Luo, Q. Wan, W. Ding, and J. Sun, 2016: Impact of assimilating wind profiling radar observations on convection-permitting quantitative precipitation forecasts during SCMREX. Wea. Forecasting, 31, 1271–1292, https://doi.org/10.1175/WAF-D-15-0156.1.
Zhao, K., and M. Xue, 2009: Assimilation of coastal Doppler radar data with the ARPS 3DVAR and cloud analysis for the prediction of Hurricane Ike (2008). Geophys. Res. Lett., 36, L12803, https://doi.org/10.1029/2009GL038658.
Zheng, M., and Coauthors, 2021a: Data gaps within atmospheric rivers over the northeastern Pacific. Bull. Amer. Meteor. Soc., 102, E492–E524, https://doi.org/10.1175/BAMS-D-19-0287.1.
Zheng, M., and Coauthors, 2021b: Improved forecast skill through the assimilation of dropsonde observations from the Atmospheric River Reconnaissance Program. J. Geophys. Res. Atmos., 126, e2021JD034967, https://doi.org/10.1029/2021JD034967.
Zhong, X., D. Zhang, and Y. Guan, 2017: Analysis of a heavy rain process in the coastal area of southwestern Guangdong May 16–17, 2015. Guangdong Meteor., 39, 14–18, https://doi.org/10.3969/j.issn.1007-6190.2017.01.004.
Zhu, Q., J. Lin, S. Shou, and D. Tang, 2007: Weather Principles and Methods. 4th ed. China Meteorological Press, 649 pp.