Aerosol-Induced Invigoration of Cumulus Clouds—A Review

Abstract

This paper is based on the keynote talk that I presented at the International Congress on Clouds and Precipitation (ICCP, 2021), wherein I was awarded a lifetime membership of ICCP. I focus on the invigoration of cumulus clouds by high concentrations of ice nuclei and hygroscopic aerosol. As far as ice nuclei are concerned, I discuss the hypothesized invigoration of cumulus clouds by seeding with high concentrations of ice nuclei or what has been called rainfall enhancement by means of “dynamic seeding”. As to the effects of enhanced concentrations of hygroscopic aerosol on cumulus dynamics and rainfall, I discuss two mechanisms, (1) “mixed-phase invigoration” and (2) “condensational invigoration”. I conclude that the concept of invigoration of convective clouds using high concentrations of hygroscopic aerosol by means of “condensational invigoration” is the dominant response of cumuli to enhanced concentrations of hygroscopic aerosol. Moreover, the invigorated cumulus clouds produce more rainfall.

1. Introduction

This is a review of my personal experience researching the potential impacts of aerosol on the dynamics and rainfall of convective clouds. In particular, I focus on the invigoration of cumulus clouds by high concentrations of ice nuclei and hygroscopic aerosol. It includes research conducted during my PhD studies at Pennsylvania State University (Penn State), at the Experimental Research Laboratory (EML) of NOAA under the direction of Dr. Joanne Simpson, and as a faculty member of the Department of Atmospheric at Colorado State University. For an independent review on the subject of aerosol invigoration of cumuli, see [1].

2. Invigoration of Cumuli and Rainfall Enhancement

As a PhD student of Meteorology at Penn State, I was mentored by Drs. Charlie Hosler, Larry Davis, and Ron Lavoie. As a consequence of their mentoring, I became enthusiastic about the potential of rainfall enhancement by seeding cumulus congestus clouds with high concentrations of AgI aerosol to invigorate their dynamics. AgI is an aerosol that is often used for cloud seeding, which acts as an ice-forming nuclei (IFN). As a result, I wrote my PhD dissertation on the subject and published two reviewed papers on the subject [2,3]. In that study, I merged a detailed (for the time) ice microphysics model with a Lagragian parcel cloud model. Dr. Joanne Simpson referred to my model as being like “putting a jet engine on a trolley car”! Pioneering research on the subject was published by [4]. A review of the basic concepts of dynamic cloud seeding can be found in [5]. Essentially, seeding cumulus congestus clouds with AgI results in the freezing of supercooled droplets and the latent heat of fusion. It is hypothesized that the additional heat enhances the buoyancy of the cloud and causes the cloud to grow taller. As ref. [6] demonstrated, in a population of natural clouds, taller clouds produce more rain, on average. Thus, it is hypothesized that the taller seeded clouds will rain more than unseeded clouds.

Observations in congestus clouds seeded with enhanced ice-forming nuclei (IFN) over Florida showed that seeding did glaciate clouds [7,8,9,10]. While limited exploratory studies of cumuli seeded to invigorate cloud growth showed an increase in rainfall [11], extensive areal-wide randomized field experiments did not confirm those exploratory experiments [12,13,14,15,16]. Following subsequent field experiments seeding cumulus congestus clouds over Thailand and west Texas, refs. [17,18] suggested that the seeding of 183 convective clouds increased the height of clouds by 7%, the area of cells by 43%, their duration by 36%, and rain volumes by 130%. The low amounts of enhanced vertical development are worth noting, which they attributed to increased precipitation mass loading, which results in only modest increases in updraft strength and cloud top height. They argued that the rapid conversion of supercooled liquid water into graupel particles is essential for rainfall enhancement. This process is facilitated in clouds that are warm-based and maritime with a broad cloud droplet distribution and supercooled raindrops.

In summary, the hypothesis for the enhancement of rainfall by means of AgI seeding via cloud invigoration is a complex process that requires the use of modern-day cloud observing tools and multidimensional cloud models (many of which are in operational use today) to better quantify the response of clouds to ice nuclei cloud seeding. Unfortunately, research funding in this area essentially terminated at the time when these tools became available.

3. Invigoration of Cumulus Clouds by Hygroscopic Aerosol

3.1. My First Experience with Dynamic Invigoration of Cumuli Using Hygroscopic Aerosol

My first experience in which enhanced concentrations of hygroscopic aerosol led to the invigoration of the dynamics of cumulus congestus and moderate-intensity cumulonimbi were in simulations of the impacts of dust incursion over the Florida peninsula during the NASA Crystal-FACE field campaign [19]. In that study, we simulated the evolution of sea-breeze-driven convection influenced by dust acting as CCN (aerosol particles that nucleate cloud droplets, generally at supersaturations of less than 1%) [20,21,22]. We also simulated the effects of giant CCN (GCCN are wettable particles greater than 5 μm in diameter) and IFN. The simulations were performed with dust acting together as CCN, GCCN, and IFN, and then independently by those mechanisms. We found that enhanced CCN affected the early stages of convection, leading to invigorated convection and enhanced rainfall. My interpretation of those simulations was that “enhanced concentrations of CCN result in the formation of high concentrations of cloud droplets, which suppresses warm rain by droplet collision and coalescence. As a result, greater quantities of cloud droplets are transported to supercooled levels. These cloud droplets freeze, releasing greater amounts of the latent heat of fusion. The enhanced buoyancy invigorates the updrafts of cumuli. The invigorated updrafts vertically transport more water thereby enhancing rainfall”. This concept can be referred to as the “mixed-phase” response to hygroscopic aerosols. Note that this concept follows directly from the original dynamic seeding concept, except that it is enhanced CCN rather that IFN! Refs. [23,24,25] also interpreted the results of simulations of enhanced CCN concentrations on cumulus clouds based on the “mixed-phase” concept.

Subsequent simulations of urban aerosol pollution carried out in our group revealed similar cumulus dynamic responses. Simulations were performed of the effects of enhanced CCN concentrations on rainfall over and downwind of St. Louis, MO. These simulations were based on observations reported from METROMEX. As in the earlier studies, enhanced CCN led to convective invigoration and enhanced rainfall early in the convective period [26].

Likewise, simulations over and downwind of Houston revealed that enhanced CCN concentrations led to convective invigoration and enhanced rainfall [27,28]. Note that in both urban studies, urban land-use changes or the urban heat island had a dominant impact on rainfall. But the urban aerosol effect was discernable.

We also carried out simulations of the impacts of high concentrations of CCN on the strength of winds associated with a derecho-producing mesoscale convective system (MCS) [29,30]. Likewise, we carried out a number of studies examining the influence of high concentrations of CCN on the intensity of tropical cyclones (TCs). The simulated TCs weakened if enhanced CCN resulted in convective invigoration in the outer rainbands and strengthened in the inner rainbands [31,32,33,34]. Except for the [35] study, I interpreted the results of those simulations in terms of the “mixed-phase” concept!

I would like to comment on an interesting anecdote regarding the Henian Zhang [31] study. Henian was a PhD candidate at the University of Illinois under the supervision of Dr. Greg McFarquhar. She came to me and asked if she could use our model, RAMS [36], to study the impact of aerosol on the dynamics of tropical cyclones. I concurred and subsequently was invited to serve on her PhD committee. I thus participated in her PhD defense along with Greg McFarquhar and Bob Rauber. So she had three well-known cloud physicists on her committee. In her oral presentation, Henian concluded that enhanced CCN concentrations altered the dynamics of the simulated tropical cyclone as a result of greater amounts of condensation and the enhanced latent heat of condensation associated with high CCN concentrations. All three meteorologists on her committee objected to her conclusion on the grounds that supersaturations in clouds are typically too small to provide much additional latent heat associated with high cloud droplet concentrations! As we shall see, I now think she was right!

3.2. Condensational Invigoration

Another concept of hygroscopic aerosol invigoration of clouds is called “condensational invigoration”. The proponents of this concept are [37,38,39,40,41]. The theory is that “numerous pollution-sized aerosol particles result in the formation of high concentrations of cloud droplets which exhibit greater net surface areas upon which condensation occurs. Thus, vapor deposition rates are enhanced, which leads to enhanced latent heat of condensation in cumuli”. Early work by [37] suggested that this effect saturates at droplet concentrations of a few hundred CCN per cubic centimeter.

I was at first reluctant to accept this hypothesis, because it is well known that peak supersaturations a few hundred meters above the cloud base are less than 1% [20,21,42,43,44]. But then I remembered a study by [45], who showed that when vigorous droplet collision and coalescence are prevalent higher in the cloud, supersaturations can rise well above near-cloud-base nominal values. A number of recent modeling studies have shown that at heights of roughly 3 km above the cloud base, where droplet collection can be prevalent, cloud droplet concentrations are reduced, and the net surface area upon which condensation occurs is reduced. Supersaturations (always referenced with respect to liquid in this paper) can then exceed nominal near-cloud-base values. As a result, an appreciable enhancement of condensation occurs in a polluted cloud relative to a clean cloud [46,47,48,49,50,51,52,53,54,55,56].

It is important to note the study by [57], in which they demonstrated the importance of water loading to the response of cumuli to high concentrations of hygroscopic aerosol. They found that high concentrations of hygroscopic aerosol led to the expected enhanced freezing of supercooled water aloft and latent heat release, but it also increased the amount of water loading at those elevated levels, which contributed to a decrease in average updraft speeds above 6 km. This response is consistent with the interpretation of AgI seeding experiments in Texas [17,18], where it was suggested that the observed low amounts of enhanced vertical development were due to increased precipitation mass loading, resulting in only modest increases in updraft strength and cloud top height.

These studies motivated me and my colleague Bob Walko to investigate the condensational invigoration hypothesis further [58]. The global-to-storm-scale model called the Ocean–Land–Atmosphere Model (OLAM) was used. OLAM is a global nonhydrostatic weather and climate prediction model [59,60,61,62]. OLAM is an outgrowth of the Regional Atmospheric Modeling System (RAMS; [36,63], which was developed for investigating meso- and cloud-scale phenomena. The cloud physics scheme used in RAMS [64,65]; ref. [36] is incorporated into OLAM. Thus, OLAM is well equipped to simulate moist convective systems at very high resolutions (e.g., down to scales of tens of meters). OLAM includes recent upgrades in cloud physics [66,67,68,69,70] which enable the explicit representation of aerosols and their impact on liquid and ice nucleation (including aerosol size, concentration, and chemistry via the hygroscopicity parameter kappa) [71]. Bob Walko implemented a Lagrangian aerosol bin model into OLAM, in which aerosol activation and vapor diffusional growth are represented explicitly as the model is integrated forward in time. Once supersaturation ceases to increase, integration is halted. At that time, the number of newly activated aerosols has been determined. This scheme allows for the characterization of multiple aerosol species with different values of hygroscopicity, as characterized by their kappa value.

To examine the relative role of aerosol-induced condensational versus mixed-phase invigoration in convective intensity and rainfall, idealized large-eddy simulations (LESs) of deep convective clouds over south Florida were performed. The GEOS-Chem global atmospheric chemistry model was run with and without anthropogenic aerosol sources. Aerosol concentrations and chemistry were represented in OLAM by the GEOS-Chem output. The results of those simulations clearly showed that higher aerosol concentrations enhanced precipitation, produced larger amounts of cloud liquid water content, enhanced updraft velocities—particularly in the latter part of the simulation and resulted in a modest enhancement of the latent heating of condensation. Those results support the concept that convective cloud invigoration is mainly due to “condensational invigoration” and not “mixed-phase invigoration”. Furthermore, those results suggest that “condensational invigoration” can produce an appreciable precipitation enhancement of ordinary warm-based convective clouds such as those that are common in locations like south Florida.

3.3. Implications

• Including the potential dynamical response of cumulus clouds to natural variability and human-induced changes in hygroscopic aerosol in large-scale climate and forecast models is extremely challenging. It is much more challenging than if the response of aerosol variability were a simple static response to, say, variability in CCN concentrations.
• The conclusion by [72,73] that rainfall downwind of a paper pulp in South Africa was due to the effluent of giant CCN was likely misinterpreted. This paper motivated considerable research and operational cloud seeding around the world [74]. By analyzing radar-defined cells in exploratory hygroscopic seeding experiments, ref. [74] suggested that after 20–30 min, seeded clouds developed higher rain masses and maintained those higher rain masses for another 25–30 min. [75] performed an independent evaluation of the South African hygroscopic seeding experiments and found that the seeding storms clearly lasted longer than the unseeded storms. He suggested that there was a clear dynamic signature of seeding. Ref. [75], however, did not infer that a dynamic enhancement of clouds was the result of condensational invigoration. None the less, his conclusion is consistent with the idea that hygroscopic seeding using flares, invigorated the dynamics of those clouds and thereby enhanced rainfall.
• The potential of rainfall enhancement in cumulus clouds by seeding with high concentrations of ultra-fine particles or aerosol particles of lower hygroscopicity than what we call CCN, is much greater than seeding with either AgI aerosol or GCCN.

3.4. Conclusions and Suggestions for Future Research

This review supports the concept of invigoration of convective clouds using hygroscopic aerosol or “condensational invigoration”. That is, high concentrations of cloud droplets that are formed on numerous pollution-sized aerosols exhibit greater net surface areas upon which condensation occurs, thereby enhancing the net vapor deposition rates, which leads to enhanced latent heat release by means of condensation in cumuli. The process is facilitated by the coexistence of very hygroscopic aerosols that are activated within a few hundred meters of the cloud base (CCN) and aerosols that have lesser hygroscopicity that can escape activation near the cloud base and be available for activation higher in the cloud, where cloud droplet collision and coalescence processes have depleted the concentrations of cloud droplets. Thus, in those elevated regions, supersaturations will rise sufficiently to activate the aerosols with lesser hygroscopicity and contribute to large concentrations of cloud droplets with higher net surface areas, thus resulting in enhanced latent heating of condensation in the cloud.

It is clear that we need to measure those simulated high supersaturations in elevated regions of cumuli. Moreover, we need to explore if there is a cut-off point where the CCN concentrations are so high that cloud droplet coalescence is so suppressed that the elevated activation of lesser hygroscopic aerosols cannot occur.