-
The EMPM model developed by Kerstein (1992), Krueger et al. (1997), Su et al. (1998), Krueger et al. (2008), and Tölle and Krueger (2014) has been extensively utilized to study the entrainment-mixing processes between cloud and the ambient air (e.g., Krueger et al., 1997, 2008; Su et al., 1998; Lu et al., 2013b, 2018b, 2020; Tölle and Krueger, 2014; Luo et al., 2020, 2021). The EMPM can track the history of every droplet due to the variation of the local environment, ranging from the model integral scale to the model Kolmogorov scale (~1 mm). During the entrainment-mixing process, four processes are considered: parcel ascent, entrainment, turbulent mixing (i.e., turbulent deformation and molecular diffusion), and droplet condensation/evaporation. Turbulent deformation is a key process and is implemented by random rearrangement events as a finite-rate turbulent mixing. Using the “triplet map” introduced by Kerstein (1991), each event can be realized. The “triplet map” replaces the scalar field within the randomly selected segment with three compressed copies of the scalar field and then inverts the central copy (Krueger et al., 1997; Su et al., 1998). This treatment changes the scalar gradient within the segment and realizes the effect of compressive strain in a turbulent flow (Krueger et al., 1997; Su et al., 1998). The finite-rate mixing has been supported by observations (Gerber et al., 2008) and is beneficial to the broadening of CDSD. This setting allows individual droplet to experience different local supersaturation environment and contribute to the different condensation/evaporation rate (Su et al., 1998).
When entrainment occurs, a randomly selected region of the parcel is replaced with environmental air. The subsequent turbulent mixing process includes two stages. In the first stage, the entrained air breaks down under the action of turbulent eddies and randomly distributes within the cloudy parcel. This stage increases the interfacial area between the cloudy air and entrained air. In the second stage, when entrained air size approaches the Kolmogorov microscale, the molecular diffusion process takes effect and rapidly smooths out the scalar gradients (Krueger et al., 1997; Su et al., 1998; Tölle and Krueger, 2014). A more detailed description of the EMPM has been provided by Krueger et al. (1997), Su et al. (1998), Krueger et al. (2008), and Tölle and Krueger (2014). The length, width, and height of the EMPM domain are 20 m, 1 mm, and 1 mm, respectively. Su et al. (1998) tested the sensitivity to sizes equaling 100 m and 20 m. The simulations indicated that “the 100-m results are similar to those from the 20-m case”. Further, the EMPM works as follows: the entire EMPM domain containing droplets is treated as a cloudy parcel and ascends at a specific vertical velocity across the entire domain. When the entrainment-mixing process occurs, environmental air replaces the same-sized cloudy parcel at the entrainment height. During the subsequent ascending process, the entrained environmental air mixes with the cloudy air at a finite turbulent rate (“Mix 1”). Subsequently, the cloud grows adiabatically (“Adiabatic”) until another entrainment-mixing process occurs (“Mix 2”). Notably, the cloud continues to ascend during the three stages.
The initial droplets in the cloudy parcel are assumed to follow a gamma size distribution (Liu et al., 2002; McFarquhar et al., 2015; Lu et al., 2020; Bera, 2021), such that the droplet number concentration n(r) for a droplet radius (r) is given as:
where N0, β, and μ represent the intercept, slope, and shape parameters, respectively. The narrow initial CDSD has μ and d of 40.0 and 0.16, respectively, and is binned by 49 bins, ranging from 1 to 25 μm in radius (Fig. S1 in the Electronic Supplementary Material, ESM). During ascent, the first entrainment-mixing process is set to occur near the beginning of the simulations to examine the effects of entrainment-mixing on the narrow CDSD. Since the model output frequency is 0.75 s, the first entrainment process is chosen to occur at 0.75 s. During the subsequent adiabatic process, d first increases and then decreases. The formation of small droplets increases d and the growth of small droplets into big droplets by condensation decreases d, as per the theoretical expectation of droplet growth (Wallace and Hobbs, 2006). The second entrainment-mixing process occurs when d reaches its maximum. Such a choice is employed to weaken the significant effect of the condensation process, which results in a decrease in d. Herein, we mainly focus on the relationship between d and λ after the second entrainment-mixing process.
The cloudy parcel has initial pressure, water vapor mixing ratio, and temperature of 963.95 hPa, 15.73 g kg−1, and 293.56 K (Raga et al., 1990; Tölle and Krueger, 2014), respectively. The baseline case of the EMPM has a vertical velocity (w) of 1.0 m s−1, relative humidity of entrained environmental air (RHe) of 88%, turbulence dissipation rate (ɛ) of 5×10−3 m2 s−3, initial droplet number concentration (ni) of 119.4 cm−3, initial liquid water content (LWCi) of 0.5 g m−3, and initial mean volume radius (rvi) of 10 μm. Further, entrained environmental air is assumed to be without and with CCN (Cases without and with CCN are considered.). According to Su et al. (1998) and Krueger et al. (2008), the CCN distribution is composed of two log-normal size distributions. The first one is 18 categories of salt with the radius range of 0.14–3.73 μm; the mean and standard deviation are 0.9 μm and 0.63 μm, respectively. The second one is 31 categories of ammonia bi-sulfate with the radius range of 0.02–0.75 μm; the mean and standard deviation are 0.09 μm and 0.05 μm, respectively; the CCN concentration is 49.65 cm−3. These CCN are assumed to be at equilibrium sizes, and the competition for water vapor between CCN and droplets is considered. The growth of each droplet is dependent on its local meteorological fields, and the curvature and solution effects of droplet are considered (see Appendix for the droplet growth equations).
Previous measurements and numerical simulations in shallow cumulus clouds indicate that ɛ can range from 10−5 to 10−2 m2 s−3 (Siebert et al., 2006a, b; Hoffmann et al., 2014); w can increase from near 0 at cloud edge to 6 m s−1 in the core of the convective cloud (Jonas, 1990; Burnet and Brenguier, 2007; Hudson et al., 2012); LWC is between 0.1 g m−3 and 1.2 g m−3 (Gerber et al., 2008; Hudson et al., 2012); ni can vary from 20 to 700 cm−3 (Burnet and Brenguier, 2007; Gerber et al., 2008; Small and Chuang, 2008; Hudson et al., 2012); and RHe can be in the range of 70%–95% (Axelsen, 2005; Burnet and Brenguier, 2007; Lu et al., 2018c). To explore the effects of impact factors on the relationship between d and λ, the w values in the domain are set to 0.5, 1.0, and 1.5 m s−1, respectively; the RHe values are set to 77%, 88%, and 93.5%, respectively; the ɛ values are set to 5×10−4, 5×10−3, and 1×10−2 m2 s−3, respectively; the ni values are set to 69.1, 119.4, and 552.6 cm−3, computed from LWCi of 0.5 g m−3 and rvi of 12, 10, and 6 μm, respectively. LWCi of 0.25, 0.5, and 0.75 g m−3 are also included with a fixed rvi of 10 μm. The above parameter settings are listed in Table 1. These values in the sensitivity tests are within reasonable ranges and represent typical shallow convective clouds with distinct environmental thermodynamics and air pollution amounts.
Cases Entrained Cloud
Condensation
Nuclei, CCNVertical
velocity,
w (m s−1)Relative humidity
of entrained
air, RHe (%)Turbulence
dissipation rate,
ɛ (m2 s−3)Initial droplet
number
concentration,
ni (cm−3)Initial liquid
water content,
LWCi (g m−3)Entrained air blob
number of the
second entrainment-
mixing process (N2)Baseline case Yes 1.0 88 5×10−3 119.4 0.5 2, 4, 6, 8, 10 Case 1:
Entrained CCN effectNo 1.0 88 5×10−3 119.4 0.5 2, 4, 6, 8, 10 Case 2:
w effectYes 0.5, 1.0, 1.5 88 5×10−3 119.4 0.5 2, 4, 6, 8, 10 Case 3:
RHe effectYes 1.0 77, 88, 93.5 5×10−3 119.4 0.5 2, 4, 6, 8, 10 Case 4:
ɛ effectYes 1.0 88 5×10−4, 5×10−3,
1×10−2119.4 0.5 2, 4, 6, 8, 10 Case 5:
ni effectYes 1.0 88 5×10−3 69.1, 119.4,
552.60.5 2, 4, 6, 8, 10 Case 6:
LWCi effectYes 1.0 88 5×10−3 119.4 0.25, 0.5, 0.75 2, 4, 6, 8, 10 Table 1. Parameters of sensitivity simulations.
-
As mentioned above, we concentrate on the second entrainment-mixing process and further investigate the relationship between d and λ. Here, λ is calculated using the method developed by Lu et al. (2012):
where L is the length of EMPM; χ is the integrated fraction of the adiabatic cloud;
$\chi^{*}_1$ and$\chi^{*}_2$ are the fractions of the adiabatic cloud at the first and second entrainment heights, respectively; and N1 and N2 are the entrained environmental air blob numbers for the first and second entrainment-mixing processes, respectively. Evidently, λ is determined by N2, given each entrained environmental air blob size (l) of 0.5 m, N1 of 10, and the entrainment height above cloud base (h). The baseline case is taken as an example to calculate λ. The corresponding height of the initial CDSD is 204.5 m above the cloud base and is calculated by assuming that the cloud parcel with LWCi = 0.5 g m−3 moves downward until LWCi = 0 g m−3. As the second entrainment occurs at 42.75 s and w is 1 m s−1, h is equal to 42.75 m + 204.5 m (i.e., 247.25 m). As expected, a larger N2 corresponds to a larger λ (Fig. S2 in the ESM), and λ exists in a reasonable range, from 1.35 to 2.29 km−1 (Lu et al., 2012; de Rooy et al., 2013). N2 is used hereafter to refer to λ. For cases with and without entrained CCN, the relationship between d and λ is investigated by setting different N2 values. In the cases of baseline and all sensitivity simulations, N2 are set to 2, 4, 6, 8, and 10, respectively, with a fixed N1 of 10. Different N1 values (2, 4, 6, 8, 10) are also tested, and as expected, a larger N1 leads to a larger d after the first entrainment-mixing process (Fig. S3 in the ESM); this is because the initial adiabatic CDSD is narrow (Tölle and Krueger, 2014; Gao et al., 2018; Luo et al., 2020). Therefore, d and λ are positively correlated. Since this relationship between d and λ is well known for adiabatic narrow CDSD, we focus on the second entrainment-mixing process and select the widest CDSD at the beginning of the second entrainment-mixing process (i.e., at the end of the first entrainment-mixing process). The CDSD after the first entrainment-mixing process for N1 = 10 is the widest and has the biggest difference from the adiabatic CDSD. Therefore, N1 is set to 10.
Cases | Entrained Cloud Condensation Nuclei, CCN | Vertical velocity, w (m s−1) | Relative humidity of entrained air, RHe (%) | Turbulence dissipation rate, ɛ (m2 s−3) | Initial droplet number concentration, ni (cm−3) | Initial liquid water content, LWCi (g m−3) | Entrained air blob number of the second entrainment- mixing process (N2) |
Baseline case | Yes | 1.0 | 88 | 5×10−3 | 119.4 | 0.5 | 2, 4, 6, 8, 10 |
Case 1: Entrained CCN effect | No | 1.0 | 88 | 5×10−3 | 119.4 | 0.5 | 2, 4, 6, 8, 10 |
Case 2: w effect | Yes | 0.5, 1.0, 1.5 | 88 | 5×10−3 | 119.4 | 0.5 | 2, 4, 6, 8, 10 |
Case 3: RHe effect | Yes | 1.0 | 77, 88, 93.5 | 5×10−3 | 119.4 | 0.5 | 2, 4, 6, 8, 10 |
Case 4: ɛ effect | Yes | 1.0 | 88 | 5×10−4, 5×10−3, 1×10−2 | 119.4 | 0.5 | 2, 4, 6, 8, 10 |
Case 5: ni effect | Yes | 1.0 | 88 | 5×10−3 | 69.1, 119.4, 552.6 | 0.5 | 2, 4, 6, 8, 10 |
Case 6: LWCi effect | Yes | 1.0 | 88 | 5×10−3 | 119.4 | 0.25, 0.5, 0.75 | 2, 4, 6, 8, 10 |