Cloud Seeding Won't Save the Great Salt Lake

There have been some media reports lately talking about the potential for expanding cloud seeding to help with the drought and the Great Salt Lake (e.g., Utah looks to expand cloud seeding to help with drought, Great Salt Lake).  These articles claim that "studies have shown up to 10% more precipitation can be generated out of a winter storm thanks to cloud seeding" and that "it's one of the best, most effective, cheapest methods we can do to increase our water supply today."

What they don't do is talk about how much runoff is actually generated by cloud seeding, and this is the most critical part of the water supply problem.  Let's take a closer look.

Does cloud seeding work?

Several weeks ago we discussed how water can exist in clouds at temperatures below 0˚C due to a lack of particles, known as ice nuclei, that can serve as a catalyst for ice formation (see The Formation of Drizzle and Freezing Drizzle).  We refer to water that exists at such temperatures as supercooled or supercooled liquid water.   

The basic idea behind cloud seeding as done in the western United States is to introduce more ice-nucleating particles into clouds with a lot of supercooled liquid water.  This is known as glaciogenic cloud seeding and causes some of the supercooled cloud droplets to freeze.  Once that happens, those newly created ice particles can grow into snowflakes since the relative humidity for ice is lower than water.  Basically, these ice crystals grow "at the expense" of the remaining supercooled liquid water droplets.  This is known a the Wegener–Bergeron–Findeisen process (named after three scientists who first described it).  

The ability of cloud seeding to initiate the freezing of cloud droplets and snow growth in clouds is not disputed. It has been shown repeatedly in lab, field, and computer modeling studies.  Some of the pioneering lab work was done by Bernard Vonnegut, brother of the famous author Kurt Vonnegut, Vincent Shaefer, and Irving Langmuir at General Electric's "House of Magic" in Schenectady New York (https://www.sciencefriday.com/segments/kurt-vonnegut-in-the-house-of-magic/).  Eventually they worked with the military to seed clouds with dry ice, showing a conversion of supercooled cloud droplets into ice particles that grew and fell downward, creating a gap in the clouds.

Source: US Army Signal Corps via https://www.sciencefriday.com/segments/kurt-vonnegut-in-the-house-of-magic/

This led to widespread research on cloud seeding in the 1960s, 1970s, and 1980s, often in arid regions like the western United States, Israel, and Australia.  Studies continue today and in some instances are beginning to use powerful computer modeling, which is probably a critical pathway to improving seeding efficiency.  

Nevertheless, it has proven very difficult to clearly quantify the contribution of cloud seeding to precipitation in actual storms and more importantly, the contribution to runoff.  Much like work done in medicine, there is no "control" case.  Careful randomization and statistical analysis needs to be done over many many cases to yield confident results and that costs money.  

As summarized in a recent review article by Rauber et al. (2019), "there have been seven randomized scientifically based projects studying the physical effects of cloud seeding for the purpose of increasing seasonal mountain snowpack."  These studies all used ground-based silver iodide generators as is commonly done in Utah and the western United States (silver iodide is a commonly used ice-nucleating particle).  However, four of them had design flaws, making their results questionable.  Of the other three, two were confirmatory. One (in Australia) reported statistically significant increases in precipitation. The other, the Wyoming Weather Modification Pilot Project (WWMPP) in the Medicine Bow and Sierra Madre Mountains was conducted from 2008–2013 and is probably the most relevant major study for Utah.  It involved the collection of data in 118 randomized precipitation cases that met quality control and seeding criteria over size years.  I put seeding criteria in italics because it is important to recognize that seeding does not work in all storms and there is a difference between the precipitation increase in seedable storms and the precipitation increase in all storms (or total seasonal snowfall).  Seeded storms meeting quality control and seeding criteria exhibited increases in precipitation, but they were not statistically significant.  Essentially, one could not distinguish if the increases were a seeding effect or background weather "noise." 

Computer modeling insights

Computer modeling is now being used to improve and evaluate the veracity of cloud seeding.  Having "digital twins" in which one can have seeded and unseeded storms offers new insights into seeding effects.  The National Center for Atmospheric Research led an effort to use computer modeling to evaluate the WWMPP.  I highlight here two key findings:

• Simulations produced a mean enhancement of precipitation during storms meeting the quality cotrol and seeding criteria of 5%.  Seeding produced enhancements between 3 and 7% in half of the storms. The maximum enhancement was 25%.  

• Approximately 30% of the annual precipitation was produced by clouds identified as seedable during the experiment.  Thus, they estimated a seeding impact on annual precipitation of 1.5%.

Hard realities

These results illustrate a hard reality for glaciogenic cloud seeding today.  Even if seeding can augment precipitation in some storms, not all storms are potential candidates for effective cloud seeding.  Most claims for seeding-based precipitation estimates are often for seedable storms and not for annual (or seasonal) precipitation.  

In the Fox13 article above, it is claimed that "up to 10% more precipitation can be generated out of a winter storm thanks to cloud seeding."  I have italicized two key phrases that seem pretty harmless, but are really the crux of the problem.  When you see a sentence like this, you should be asking the critical questions: How good is that estimate? What fraction of storms are potentially seedable?  How much is the increase in seasonal precipitation?  How large is the increase in runoff?

I conclude with a quote from a recent paper by Geerts and Rauber (2022).  

"Public confidence that cloud seeding 'works' is generally high in regions with operational seeding, notwithstanding decades of scientific reports indicating that the changes in precipitation are uncertain."

Cloud seeding will not end this drought, nor will it save the Great Salt Lake. We should not be fooled into thinking otherwise or conflating precipitation increases from "some" storms as seasonal or runoff increases.