-
We used the ERA5 reanalysis data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) (Hersbach et al., 2020). ERA5 is the fifth generation ECMWF reanalysis for global climate and weather and has a horizontal resolution of 0.25° × 0.25° and a time interval of 1 h. Surface rain gauge observations were provided by the National Meteorological Information Center of the China Meteorological Administration. We also used the operational S-band dual polarization radar observation at Zhengzhou station. The radar operates in the volume coverage pattern 21 (VCP-21) scanning mode, consisting of nine elevation angles: 0.5°, 1.5°, 2.4°, 3.4°, 4.3°, 6.0°, 9.9°, 14.6°, and 19.5°. The temporal resolution of radar reflectivity data is 6 min.
-
A three-dimensional and non-hydrostatic WRF model, version 4.2, was used to simulate the “21.7” Henan rainstorm. The domains and configurations of the model are shown in Fig. 3a and Table 1. The ERA5 reanalysis data was used as the initial and boundary conditions, and the model integrates for 48 h starting from 0000 UTC on 19 July 2021. The integration step of the outer domain was 9 s, and the inner domain was 3 s. In order to introduce large-scale fields consistent with the driving fields, the four-dimensional data assimilation (FDDA) functions were activated by performing grid analysis nudging throughout the model integration. In addition, sea surface temperature (SST) data with a resolution of 1° × 1° from the National Centers for Environmental Prediction (NCEP) was used for SST update during the model integration. A modified Morrison two-moment microphysical scheme was used in our study, in which the cloud droplet number concentration was set to 500 cm−3 for polluted continental cases, and the fall speed parameters of graupel were modified based on the sensitivity experiments.
Parameter Description Model WRF V4.2 Horizontal grid spacing Domain 1: 3 km
Domain 2: 1 kmNesting Two−way nesting Grid points Domain 1: 700×700×51
Domain 2: 601×601×51Model top pressure 50 hPa Cloud microphysical scheme Morrison 2−mom Cumulus convective scheme No Planetary boundary layer scheme YSU Land surface scheme Noah land−surface model Longwave radiation scheme RRTMG Shortwave radiation scheme RRTMG Table 1. Design of the numerical simulation experiment.
-
The methods for calculating PE can be divided into two categories: one is LSPE, which reflects the large-scale environment, and the other is CMPE, which reflects the cloud microphysical processes. The calculation of LSPE is based on the surface precipitation equation by Gao et al. (2005) and Cui and Li (2006), while CMPE is calculated based on the budget of cloud hydrometeors during the precipitation process (Li et al., 2002; Sui et al., 2005). The 3D WRF-based LSPE and CMPE are derived by Mao et al. (2018) as follows:
Surface rainfall budget can be expressed as:
$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{T}} $ ,$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ ,$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{E}} $ ,$ {Q}_{\mathrm{C}\mathrm{M}} $ , and$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ can be calculated by:Then, LSPE and CMPE can be written as:
Here,
$ {P}_{\mathrm{s}} $ , the surface rain rate, is equal to the sum of water vapor processes, including local atmospheric drying/moistening ($ {Q}_{\mathrm{W}\mathrm{V}\mathrm{T}} $ ), water vapor flux convergence/divergence ($ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ ), surface evaporation ($ {Q}_{\mathrm{W}\mathrm{V}\mathrm{E}} $ ), and hydrometeor loss/convergence or hydrometeor gain/divergence ($ {Q}_{\mathrm{C}\mathrm{M}} $ ).$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ is the net amount of water vapor consumed by microphysical processes.$ H $ is the Heaviside function, where$ H\left(F\right)=1 $ when$ F > 0 $ , and$ H\left(F\right)=0 $ when$F \leqslant 0$ . For example, H(QWVT ) = 1 when QWVT > 0, and H(QWVT ) = 0 when QWVT ≤ 0. H(QWVF), H(QWVS) and H(QCM) are interpreted in the same way, in order to ensure that the calculated LSPE and CMPE range from 0 to 100%.$\left[\left( \right)\right]={\int }_{{z}_{{\rm{b}}}}^{{z}_{{\rm{t}}}}\rho \left( \right){\rm{d}}z$ denotes the mass vertical integration, where$ {z}_{\mathrm{t}} $ and$ {z}_{\mathrm{b}} $ are the top and bottom height of the model atmosphere, respectively.${q}_{\mathrm{v}},\;{q}_{\mathrm{c}},\;{q}_{\mathrm{r}},\;{q}_{\mathrm{i}},\;{q}_{\mathrm{s}},\;\mathrm{ }\;\mathrm{a}\mathrm{n}\mathrm{d}\;{q}_{\mathrm{g}}$ are the mixing ratios of water vapor, cloud water, rainwater, cloud ice, snow, and graupel.$ u\;\mathrm{a}\mathrm{n}\mathrm{d}\;v $ are the zonal and meridional winds, and$ {E}_{\mathrm{s}} $ is the surface evaporation. The microphysical processes in the Morrison scheme are summarized in Table 2.Notation Description $ \mathrm{E}\mathrm{P}\mathrm{R}\mathrm{D} $ Sublimation of cloud ice $ \mathrm{E}\mathrm{P}\mathrm{R}\mathrm{D}\mathrm{G} $ Sublimation of graupel $ \mathrm{E}\mathrm{P}\mathrm{R}\mathrm{D}\mathrm{S} $ Sublimation of snow $ \mathrm{P}\mathrm{R}\mathrm{D} $ Deposition of cloud ice $ \mathrm{P}\mathrm{R}\mathrm{D}\mathrm{G} $ Deposition of graupel $ \mathrm{P}\mathrm{R}\mathrm{D}\mathrm{S} $ Deposition of snow $ \mathrm{P}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{G} $ Rain-graupel collection $ \mathrm{P}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{S} $ Rain-snow collection $ \mathrm{P}\mathrm{S}\mathrm{A}\mathrm{C}\mathrm{R} $ Conversion due to collection of snow by rain $ \mathrm{P}\mathrm{I}\mathrm{A}\mathrm{C}\mathrm{R} $ Change QR, ice-rain collection $ \mathrm{P}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{I} $ Change QI, ice-rain collection $ \mathrm{P}\mathrm{I}\mathrm{A}\mathrm{C}\mathrm{R}\mathrm{S} $ Change QR, ice-rain collision, added to snow $ \mathrm{P}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{I}\mathrm{S} $ Change QI, ice rain collision, added to snow $ \mathrm{P}\mathrm{G}\mathrm{S}\mathrm{A}\mathrm{C}\mathrm{W} $ Conversion to graupel due to collection droplet
by snow$ \mathrm{P}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{S} $ Conversion to graupel due to collection rain by snow $ \mathrm{P}\mathrm{S}\mathrm{A}\mathrm{C}\mathrm{W}\mathrm{G} $ Change in Q collection droplets by graupel $ \mathrm{P}\mathrm{S}\mathrm{A}\mathrm{C}\mathrm{W}\mathrm{I} $ Change Q droplet accretion by cloud ice $ \mathrm{P}\mathrm{S}\mathrm{A}\mathrm{C}\mathrm{W}\mathrm{S} $ Change Q droplet accretion by snow $ \mathrm{M}\mathrm{N}\mathrm{U}\mathrm{C}\mathrm{C}\mathrm{R} $ Contact freezing of rain $ \mathrm{E}\mathrm{V}\mathrm{P}\mathrm{M}\mathrm{S} $ Melting and evaporation of snow $ \mathrm{E}\mathrm{V}\mathrm{P}\mathrm{M}\mathrm{G} $ Melting and evaporation of graupel $ \mathrm{P}\mathrm{G}\mathrm{M}\mathrm{L}\mathrm{T} $ Melting of graupel $ \mathrm{P}\mathrm{S}\mathrm{M}\mathrm{L}\mathrm{T} $ Melting of snow $ \mathrm{P}\mathrm{R}\mathrm{A} $ Accretion droplets by rain $ \mathrm{P}\mathrm{R}\mathrm{C} $ Auto-conversion of droplets $ \mathrm{P}\mathrm{R}\mathrm{E} $ Evaporation of rain $ \mathrm{P}\mathrm{C}\mathrm{C}( > 0) $ Condensation of cloud droplets $ \mathrm{P}\mathrm{C}\mathrm{C}( < 0) $ Evaporation of cloud droplets $ \mathrm{N}\mathrm{P}\mathrm{R}\mathrm{C}1 $ Change NR autoconversion droplets $ \mathrm{N}\mathrm{P}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{S} $ Change N rain-snow collection $ \mathrm{N}\mathrm{N}\mathrm{U}\mathrm{C}\mathrm{C}\mathrm{R} $ Change N due to contact freezing of rain $ \mathrm{N}\mathrm{R}\mathrm{A}\mathrm{G}\mathrm{G} $ Self-collection of rain $ \mathrm{N}\mathrm{I}\mathrm{A}\mathrm{C}\mathrm{R} $ Change N, ice-rain collection $ \mathrm{N}\mathrm{I}\mathrm{A}\mathrm{C}\mathrm{R}\mathrm{S} $ Change N, ice-rain collision, added to snow $ \mathrm{N}\mathrm{P}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{G} $ Change N collection rain by graupel $ \mathrm{N}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{C}\mathrm{S} $ Change N conversion to graupel due to collection
rain by snow$ \mathrm{N}\mathrm{S}\mathrm{U}\mathrm{B}\mathrm{R} $ Loss of NR during evaporation $ \mathrm{N}\mathrm{S}\mathrm{M}\mathrm{L}\mathrm{T}\mathrm{R} $ Change N melting snow to rain $ \mathrm{N}\mathrm{G}\mathrm{M}\mathrm{L}\mathrm{T}\mathrm{R} $ Change N melting graupel to rain Table 2. List of microphysical processes in Morrison microphysics scheme.
-
From the analysis above, the model generally reproduced the spatial and temporal distribution of the precipitation, the convective systems producing the rainstorm, and the mesoscale dynamical and thermodynamic environment. Therefore, the simulation data could be used to analyze the PE of the heavy rainfall in the following subsections.
-
Figure 8 shows the spatial distribution of surface rainfall rate, LSPE, and CMPE during the different stages of precipitation. In general, the regions with high LSPE and CMPE corresponded to the regions of heavy precipitation. For example, at 1200 UTC on 19 July (Figs. 8a−c), rainfall mainly occurred in the south of Zhengzhou and Shangqiu, showing a northwest–southeast zonal distribution. The distributions of LSPE and CMPE showed similar patterns to surface rainfall, and CMPE was greater than LSPE. At 2100 UTC on 19 July (Figs. 8a1−c1), as the rainband moved to the central Henan province, the corresponding LSPE and CMPE also showed similar distribution patterns. It should be noted that the relationship between PE and rainfall rate was not linear with one-to-one correspondence. Based on the formulas of LSPE and CMPE, the regions of weak precipitation may have weak moisture convergence and surface evaporation, which may lead to relatively high LSPE to some extent.
-
Figure 9 shows the time series of the area-averaged (34.3º–34.8ºN, 113º–114.1ºE) hourly rainfall rate, LSPE, and CMPE. As seen from Fig. 9a, the observed hourly rainfall rate (black line) showed a fluctuating upward trend staring from 0000 UTC 19 July, with a peak of 29 mm h−1 at 0900 UTC on 20 July, which was similar to the trend of hourly precipitation seen at Zhengzhou station in Fig. 3b. The corresponding simulations (red line) showed similar variation trends, with a maximum value of 27 mm h−1 at 1000 UTC on 20 July. Thus, the simulation generally replicated the evolution and magnitude of the observed precipitation from the perspective of regional average and could be used to calculate LSPE and CMPE. From Fig. 9b, it was found that CMPE was always greater than LSPE, and the average LSPE and CMPE were about 33.34% and 58.13%, respectively. According to the studies by Mao et al. (2018), the average LSPE and CMPE for the “7.21” rainstorm in Beijing were 36.09% and 60.66% in the warm sector precipitation stage and were 32.12% and 54.02% in the cold frontal precipitation stage, respectively. The average LSPE and CMPE of the “21.7” Henan rainstorm were between the values of warm-sector and cold frontal precipitation during the “7.21” rainstorm. It was noteworthy that the highest LSPE and CMPE were 58.26% and 88.54%, respectively, at 0300 UTC on 20 July, although the strongest hourly rainfall occurred at 0900 UTC 20 July.
-
To determine the key physical factors influencing LSPE, the numerator and denominator terms of LSPE were analyzed (Figs. 10a, 10c). In general, the
$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ was the main positive contributor to$ {P}_{\mathrm{s}} $ , accounting for 86.75%. The contribution rates of the other terms were relatively small (Table 3). Therefore,$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ had a significant impact on the surface rainfall. From the denominator term of LSPE (Fig. 10c), the water vapor flux convergence$H\left({Q}_{\mathrm{W}\mathrm{V}\mathrm{F}}\right)$ ${Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ played a key role in the supply of water vapor, accounting for 48.19%, followed by the local atmospheric drying$H\left({Q}_{\mathrm{W}\mathrm{V}\mathrm{T}}\right)$ ${Q}_{\mathrm{W}\mathrm{V}\mathrm{T}} $ , contributing 36.49%.Figure 10. Time series of area-averaged numerator terms (units: mm h−1) of (a) LSPE and (b) CMPE; Time series of area-averaged denominator terms (units: mm h−1) of (c) LSPE and (d) CMPE. The black lines in (a) and (b) denote the surface rainfall rate. Area average is over the range of the black box in Fig. 8.
Terms Contribution rates (%) LSPE numerator CMPE numerator LSPE denominator CMPE denominator $ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ 86.75 − 48.19 − $ {Q}_{\mathrm{W}\mathrm{V}\mathrm{T}} $ 7.30 − 36.49 − $ {Q}_{\mathrm{C}\mathrm{M}} $ –3.31 –3.31 13.51 34.28 $ {Q}_{\mathrm{W}\mathrm{V}\mathrm{E}} $ 9.26 − 1.81 − $ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ − 103.31 − 65.72 Table 3. Contribution rates of different terms in LSPE and CMPE.
To understand the influence of
$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ on LSPE,$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ can be further decomposed into (a) the water vapor convergence ($ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ ) and (b) the water vapor advection ($ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ ):It is clear from Fig. 11a that
$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ was almost completely contributed by$ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ , and the contribution of$ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ was very small.$ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ reached a maximum value at 0800 UTC, when heavy precipitation happened, indicating that strong convergence of water vapor ($ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ > 0) played a significant role in the formation of heavy rainfall. Furthermore, the vertical profiles of the non-integral$ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ and$ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ are given in Fig. 11b. Water vapor convergence ($ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ > 0) occurred below the 0°C layer, with the strongest convergence near the surface, while moisture divergence ($ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ < 0) occurred above the 0°C layer, with its peak between the heights of 6−7 km. The advection of water vapor ($ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ ) mainly occurred below 6 km, with moist advection ($ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ > 0) extending from the ground to 1.5 km and dry advection ($ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ < 0) between 1.5−6 km, indicating that there was intrusion of dry and cold air in the lower and middle troposphere.Figure 11. (a) Time series and (b) vertical profiles of area-averaged
$ \mathrm{Q}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{V} $ (red bar and line),$ \mathrm{V}\mathrm{G}\mathrm{R}\mathrm{A}\mathrm{D}\mathrm{Q} $ (blue bar and line), and$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{F}} $ (black line); (c) Time series and (d) vertical profiles of area-averaged tendency of hydrometeors, including cloud water ($ \mathrm{Q}\mathrm{C}\mathrm{T}\mathrm{E}\mathrm{N} $ , red bar and line), rainwater ($ \mathrm{Q}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ , blue bar and line), ice crystal ($ \mathrm{Q}\mathrm{I}\mathrm{T}\mathrm{E}\mathrm{N} $ , yellow bar and line), snow ($ \mathrm{Q}\mathrm{S}\mathrm{T}\mathrm{E}\mathrm{N} $ , orange bar and line), graupel ($ \mathrm{Q}\mathrm{G}\mathrm{T}\mathrm{E}\mathrm{N} $ , green bar and line), and$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ (black line) . The black dotted line represents the height of the 0°C layer. Area average is over the range of the black box in Fig. 8. -
The impacts of physical factors on CMPE are represented in Figs. 10b and d. It can be found that the mean contribution rates of
$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ and$ {Q}_{\mathrm{C}\mathrm{M}} $ to$ {P}_{\mathrm{s}} $ were 103.31% and –3.31%, respectively, while$ H\left({Q}_{\mathrm{W}\mathrm{V}\mathrm{S}}\right){Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ and$ {H(Q}_{\mathrm{C}\mathrm{M}}){Q}_{\mathrm{C}\mathrm{M}} $ accounted for 65.72% and 34.28%, respectively, in the denominator terms of CMPE (Fig. 10d). Clearly,$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ was the key physical factor that influenced CMPE.The net consumption of water vapor by microphysical processes (
$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ ) is equal to the sum of the net increase in hydrometeors ($ \mathrm{Q}\mathrm{C}\mathrm{T}\mathrm{E}\mathrm{N} $ ,$ \mathrm{Q}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ ,$ \mathrm{Q}\mathrm{I}\mathrm{T}\mathrm{E}\mathrm{N} $ ,$ \mathrm{Q}\mathrm{S}\mathrm{T}\mathrm{E}\mathrm{N} $ ,$ \mathrm{Q}\mathrm{G}\mathrm{T}\mathrm{E}\mathrm{N} $ ), so we analyzed the tendency of hydrometeors to indirectly study the impacts of$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ on CMPE. In terms of magnitude, the$ \mathrm{Q}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ term had the greatest influence on$ {Q}_{\mathrm{W}\mathrm{V}\mathrm{S}} $ , followed by$ \mathrm{Q}\mathrm{C}\mathrm{T}\mathrm{E}\mathrm{N} $ and$ \mathrm{Q}\mathrm{S}\mathrm{T}\mathrm{E}\mathrm{N} $ (Fig. 11c), revealing that most of the consumed water vapor was transformed to cloud water between 5–7 km and snow and graupel between 5–10 km, and then converted to raindrops near the 0°C layer through melting and collision-coalescence (Fig. 11d).Furthermore, we decompose
$ \mathrm{Q}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ to study the impact of rain-related microphysical processes on CMPE. The calculation of$ \mathrm{Q}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ is as follows:As seen from Fig. 12a, rainwater was mainly formed through the accretion of cloud water (
$ \mathrm{P}\mathrm{R}\mathrm{A} $ ), followed by melting of snow ($ \mathrm{P}\mathrm{S}\mathrm{M}\mathrm{L}\mathrm{T} $ ) and graupel ($ \mathrm{P}\mathrm{G}\mathrm{M}\mathrm{L}\mathrm{T} $ ). The primary sink of rainwater was evaporation ($ \mathrm{P}\mathrm{R}\mathrm{E} $ ). From the vertical profiles, during 0000–0100 UTC on 20 July (Fig. 12b), rain mixing ratio was mainly contributed by$ \mathrm{P}\mathrm{S}\mathrm{M}\mathrm{L}\mathrm{T} $ at about 4–5 km, followed by$ \mathrm{P}\mathrm{R}\mathrm{A} $ below the melting level. During 0800–0900 UTC, when the heavy rainfall occurred (Fig. 12b1), the conversion rates of$ \mathrm{P}\mathrm{G}\mathrm{M}\mathrm{L}\mathrm{T} $ and$ \mathrm{P}\mathrm{R}\mathrm{A} $ increased significantly and became the main contributors to rain mass. The budget of rain number concentration showed similar results (Figs. 12c–c1). During the heavy precipitation period, a great number of raindrops were produced by auto-conversion from cloud droplets ($ \mathrm{N}\mathrm{P}\mathrm{R}\mathrm{C}1 $ ) between 5–8 km and between 2–3 km. Near the melting layer, rain number concentration was mainly contributed by$ \mathrm{P}\mathrm{G}\mathrm{M}\mathrm{L}\mathrm{T} $ . The number concentration of rain decreased rapidly between 4–5 km through self-collection ($ \mathrm{N}\mathrm{R}\mathrm{A}\mathrm{G}\mathrm{G} $ ). The budget analysis suggested that$ \mathrm{P}\mathrm{R}\mathrm{A} $ and$ \mathrm{P}\mathrm{G}\mathrm{M}\mathrm{L}\mathrm{T} $ were the main contributors to rain mixing ratio and number concentration during the heavy precipitation stage. In the Morrison microphysical scheme,$ \mathrm{P}\mathrm{R}\mathrm{A} $ was proportional to the mixing ratios of cloud water and rainwater. During the heavy precipitation period, both cloud water and rainwater mixing ratios increased significantly, thus the production rate of$ \mathrm{P}\mathrm{R}\mathrm{A} $ also increased.Figure 12. (a) Time series and vertical profiles of source and sink terms of (b, b1) rain mixing ratio (
$ \mathrm{Q}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ ) and (c, c1) rain number concentration ($ \mathrm{N}\mathrm{R}\mathrm{T}\mathrm{E}\mathrm{N} $ ). (b, c) are averaged during 0000–0100 UTC on 20 July, (b1, c1) are averaged during 0800–0900 UTC on 20 July. -
From the analysis above, it was found that water vapor convergence was the key physical factor in LSPE, and the rain-related source terms were crucial in CMPE. In this section, the possible mechanisms of LSPE and CMPE are further explored.
-
Figure 13 shows the horizontal and vertical distributions of water vapor flux divergence derived from ERA5 data. Figure 2 illustrates that water vapor was transported by a strong southeasterly air flow from the East China Sea to Henan under the influence of the subtropical high and Typhoon In-Fa (2021). Meanwhile, the southerly air flow of Typhoon Cempaka (2021) also transported water vapor to Henan. The two water vapor transport channels converged in Zhengzhou, providing sufficient moisture for the heavy rainstorm. From Fig. 13, it can be seen that water vapor was blocked by the Taihang and Funiu Mountains in western Henan province, and strong moisture convergence formed on the windward slope in front of the mountain. From the vertical cross section, it can be seen that at 0000 UTC on 20 July the moisture convergence extended from near the ground to the level of 500 hPa, with the strongest convergence located at 925 hPa.
Figure 13. Vertical integral of divergence of moisture flux (blue shading, units: kg m−2 s−1), wind field at 850 hPa (barbs, full barb denotes 4 m s−1), and terrain height (gray shading, units: m) at (a) 0000 UTC and (b) 1200 UTC on 20 July 2021; Zonal–vertical distribution of divergence of moisture flux (shading, units: 10−7 kg m−2 s−1 hPa−1) and wind field (arrows, units: m s−1) along 34.6°N at (c) 0000 UTC and (d) 1200 UTC. The red line indicates the location of cross sections.
A sensitivity experiment was performed to further examine the blocking of moisture by the Funiu and Taihang Mountains. In the sensitivity experiment, the terrain height of the mountainous area (32.5°–36.0°N, 110.5°–114.0°E) in western Henan province was modified to half of the original height. The simulation results showed that the rainfall center in the sensitivity experiment moved northward to Jiaozuo city and the maximum accumulated precipitation was about 756 mm, which was much lower than that simulated in the control experiment (about 1041 mm, figure omitted). It was found that the moisture convergence over Zhengzhou in the sensitivity experiment was slightly weaker than that in the control experiment, suggesting the topographic blocking of the Taihang and Funiu Mountains enhanced moisture convergence in Zhengzhou and prompted heavy rainfall. The blocking effect of mountains and its role in the convective initiation in the southwest of Zhengzhou are fully discussed by Yin et al. (2022).
-
To understand the probable mechanism of CMPE, a diagnostic study of environmental and microphysical variables was performed. It is shown in Fig. 14a that there was a high relative humidity (RH) layer between 4–8 km over Zhengzhou, which may be the results of large-scale vapor transport and updrafts induced by local topography. RH deceased significantly at about 0800–0900 UTC on 20 July when heavy rainfall happened. In terms of dynamic fields, both the updrafts and downdrafts reached their peaks at 0900−1000 UTC 20 July, and so did the mass contents of liquid water (sum of cloud water and rainwater) and ice phase particles (sum of cloud ice, snow, and graupel). From the thermodynamic fields, we found there was a negative potential temperature perturbation between 3–5 km starting from 1200 UTC on 19 July and reaching its maximum at 0000 UTC on 20 July, which may be associated with the diabatic cooling of evaporation and melting processes.
Figure 14. Time–height distribution of (a) relative humidity (units: %); (b) updraft (units: m s−1); (c) downdraft (units: m s−1); (d) mass content of liquid water (units: g kg−1); (e) mass content of ice phase particles (units: g kg−1); (f) potential temperature perturbation (units: K). The black solid line is the hourly precipitation, and the red solid lines are the 0°C and −20°C isotherms.
According to the studies by Browning and Golding (1995) and Browning (1997), dry intrusion is a coherent region of air descending from near-tropopause level. The dry intrusion is characterized by high potential vorticity and relatively low wet-bulb potential temperature and can lead to convective instability where it overruns the warm air. In our study, dry intrusion air is defined as the air with RH less than 60% (Yao et al., 2007). Figure 15 shows the wind field and RH at 500 hPa derived from ERA5 reanalysis data. At 0000 UTC on 20 July (Fig. 15a), the low vortex at 500 hPa was located in western Henan province and Zhengzhou is controlled by the southwesterly flow with RH greater than 100%. At 0800 UTC, when the strong precipitation occurred (Fig. 15b), dry air (RH less than 60%) intruded from the northwestern part of the vortex into Zhengzhou (indicated by the brown arrow), and the RH over Zhengzhou decreased to 90%. The dry intrusion was also significant at 300 hPa (figure omitted). Figures 15c–d depict the zonal–vertical cross sections of RH and potential vorticity (PV) along Zhengzhou. At 0000 UTC, the low levels over Henan were dominated by easterly winds, and strong updrafts were induced by orographic lifting. The air over Zhengzhou was very moist, with RH over 100% extending from the surface to upper troposphere. Regions of high PV were mainly located above 12 km and in the middle-to-upper levels between 107°–111°E, which was a result of downward movement of high PV from the upper levels. At 0800 UTC, there was dry air intrusion in the middle and upper levels of Zhengzhou. The dry and cold air was superimposed on the warm and moist air in the lower levels, which enhanced the convective instability in Zhengzhou. Moreover, the descending of high PV from the upper troposphere interacted with the high PV in the lower levels, which promoted the development of a low vortex and also enhanced precipitation (Yao et al., 2007). The intrusion of cold and dry air also resulted in the supersaturation and condensation of water vapor, which contributed to rainfall through
$ \mathrm{P}\mathrm{R}\mathrm{A} $ .Figure 15. (a–b) Relative humidity (shading, units: %) and wind field (barbs, full barb denotes 4 m s−1) at 500 hPa derived from ERA5 data; (c–d) Zonal–vertical distribution of relative humidity (shading, units: %), potential vorticity (blue contours, units: PUV) and wind field (vectors) along 34.9°N at (a, c) 0000 UTC and (b, d) 0800 UTC on 20 July. The red triangles denote the location of Zhengzhou.
Parameter | Description |
Model | WRF V4.2 |
Horizontal grid spacing | Domain 1: 3 km Domain 2: 1 km |
Nesting | Two−way nesting |
Grid points | Domain 1: 700×700×51 Domain 2: 601×601×51 |
Model top pressure | 50 hPa |
Cloud microphysical scheme | Morrison 2−mom |
Cumulus convective scheme | No |
Planetary boundary layer scheme | YSU |
Land surface scheme | Noah land−surface model |
Longwave radiation scheme | RRTMG |
Shortwave radiation scheme | RRTMG |