Making Rain in the American Southwest: How Irrigation Strengthens the Monsoon

The evaporative tower, or  'aerological accelerator' of Starr and Anati, as depicted in the 1973 Scientific American article 'The Control of the Water Cycle' by Jose P. Peixoto and M. Ali Kettani
The water vapor tower, or ‘aerological accelerator’ of Starr and Anati, as depicted in the 1973 Scientific American article ‘The Control of the Water Cycle’ by Jose P. Peixoto and M. Ali Kettani

Mad science. That’s what I thought when I first read the 1973 Scientific American classic ‘The Control of the Water Cycle’ by Professors Jose Peixoto and Ali Kettani. The two discussed a radical idea, originally proposed by their colleagues Victor Starr and David Anati of MIT.

Why not build giant, solar-heated water vapor towers on the shores of our arid coastlines?  Warmed air would rise inside the monster chimneys, mimicking natural convection and drawing in moist air from the nearby sea.   Like sky-high smokestacks, the towers would inject water vapor high into the atmosphere, so that somewhere in the downwind direction on the land (hopefully locally), it would condense out and form precipitation.

Nutty professors, I thought.  Trying to engineer the natural evaporation-condensation cycle for benefit of humans.  A sort of crazy, anthropogenic-driven loop in the water cycle.

Fast forward, 40 years.  Were they really that crazy?  Or rather, were they just incredibly prescient?  After all, they knew that the recipe for making rain includes as a key ingredient, an ample supply of water vapor. Sprinkle in a good handful of cloud condensation nucleii (the tiny atmospheric dust particles that provide a ‘seed’ for condensation and the formation of rain drops) and add a generous amount of upward air motion  (to cool the air, bring it to its dew point and ‘squeeze’ out the moisture) and you’ve got yourself a vigorous storm system.

Little did they know that we were already, very likely running such human-driven loops, albeit unintentionally (and without the sci-fi towers).

Recently, my former Ph.D. student and postdoctoral researcher Prof. Min-Hui Lo and I published a paper in Geophysical Research Letters, in which we explored the far-field climate impacts of large-scale irrigation in California’s Central Valley. (The paper is freely available to the general public. Please download it. I paid extra for that.)  We’d already done work with satellites to quantify the rapid rates of groundwater depletion in the Valley. Now we wanted to explore where all that water was going, and what was happening along the way.

Honestly, what we found surprised us, but probably would not have surprised Prof. Peixoto and colleagues at all.  Using a global climate model (the computer models like those used in the IPCC), we added realistic rates of summertime irrigation to the Central Valley (about 40% of which was groundwater) and then ran the model for 90 years.

The nutty professors. Jay Famiglietti and Min-Hui Lo in Irvine, CA
The nutty professors. Jay Famiglietti and Min-Hui Lo in Irvine, CA. January, 2012.

The result?  Evaporation in the Valley doubled, leading to a major export of water vapor downwind, to the arid American Southwest.  Here, the extra water vapor collided with the powerful North American monsoon system that spans the 4-state region of Colorado, Utah, Arizona and New Mexico, and is home to much of the Colorado River basin.

Once in the region, the extra moisture ‘strengthened’ the monsoon by increasing summer rainfall by 15%, and by triggering further upward motion in the monsoon circulation, thereby drawing in even more water vapor from the Gulfs of California and Mexico. The wetter soil in the 4-state region generated 8% more evaporation, which also contributed to the increased monsoon precipitation.

All of that extra rainfall resulted in a 28% increase in runoff to the Colorado River, or more than 11 billion gallons of water.  That’s enough to supply about 3 million people with water for a year. Certainly an intriguing result for those of us who live in the west and rely on the river for a major fraction of our regional water supply.  Where I live in southern California for example, 25% of our water supply comes from the Colorado.

When we consider that in the Lower Colorado River Basin, all of that extra streamflow winds up in Lake Mead, and that most of the water allocated from Lake Mead is shipped to California through water supply conveyances like the All-American Canal and the Colorado River Aqueduct, one of the things that captured our attention was that, really, we’ve created a regional, anthropogenic loop in the water cycle.  Sounding familiar?

How the Water Loop Works

The anthropogenic loop works like this.  Summer irrigation in the Central Valley doubles evaporation there. The additional water vapor travels to the 4-corners region, where it encounters the active monsoon system, and, like throwing fuel on a fire, it strengthens the monsoon by increasing precipitation, evaporation and runoff.

OC Register graphic by Sonya Quick that accompanied science writer Pat Brennan's great article about our paper.
OC Register graphic by Sonya Quick and Maxwell Henderson that accompanied science writer Pat Brennan’s great article about our paper.

The extra runoff increases Colorado River streamflow, which ultimately flows into Lake Mead.  Some of that water, which started out as surface or groundwater in the Central Valley, completes its round trip journey by returning to California via one of our water supply canals.

In short, according to our computer model simulations, the Valley is acting like Peixoto et al.’s water towers, injecting water vapor into the atmosphere (click here to see an animation from our study) and kicking off a chain of downwind phenomena that end up returning the water, by both natural and artificial means, pretty close to its original point of departure.

When building and running computer simulation models however, there’s an important question that we should always ask ourselves.  Are these results realistic, or are they just computer model fiction?

The mechanisms are surely plausible.  As I outlined in the recipe for rainfall above, we know what it takes for precipitation to form, and our findings fall well within the limits of feasibility.

Irrigated rice fields, Sacramento River Valley, Sutter County, CA. Photo by Tom Myers

Next Steps

However, it is absolutely imperative that the scientific community work towards better representing water management in its hydrology and climate models.  Many more of these types of simulations must be conducted so that the significance and transferability of our findings can be verified.

Likewise, we need to analyze available datasets on regional irrigation rates, evaporation, vapor transport, precipitation, runoff and monsoon intensity.  That unfortunately, is easier said than done.  Many of these variables are not measured frequently enough, or with enough spatial coverage, to really determine irrigation’s remote impacts.  Or, even worse, some of them are just not measured at all.

So, we start with computer models and hope that our work will stimulate serious discussion on improving our simulation tools and our network of water cycle observations on the ground.

Let’s assume for the moment that the irrigation impacts that we have highlighted are real. What then are the implications for regional water availability and management?  Let me list a just few here.

Suppose that, in the coming decades, as groundwater depletion continues in the Central Valley, agriculture at its current mega-scale is no longer sustainable. As planted acreage and irrigation decrease, a significant source of moisture for the monsoon and Colorado River streamflow will begin to disappear.  Will we see yet another stressor on the river flow that will lead its further decrease?  More generally, can the strength of the monsoon, and by extension, runoff into the Colorado River, be modulated by managing human contributions to vapor transport into the region?

Many water-stressed regions, like the Middle East for example, already recognize the importance of water lost through evaporation and transport of the vapor out of the region.  Israel has invested heavily in minimizing evaporative and water vapor losses by employing precision drip irrigation (below the surface and almost directly on roots), by covering surface soils with plastic sheeting, and by extensive use of greenhouses.

Supplementary Figure 3 from Lo and Famiglietti, 2013. The change in averaged evapotranspiration for the Central Valley is △ETCV; △ETSWUS, △PSWUS, and △RSWUS are the changes in averaged evapotranspiration, precipitation, and runoff across the 4-state SW U.S. region respectively; the four blue arrows around the boundaries of the 3-D box are the changes in the low-level water vapor transport into and out of the SW U. S. The change in averaged runoff to the Colorado River is ∆R_CRB. The percentage symbol (%) indicates the increased ratio compared to the control run (no irrigation) in our simulation.
Supplementary Figure 3 from Lo and Famiglietti, 2013. It shows in percent (%) the simulated increases in summer precipitation, evapotranspiration and runoff that result from Central Valley irrigation, compared to a simulation with no irrigation. All units are cubic kilometers and are averages over the summer (June, July and August). The change in averaged evapotranspiration for the Central Valley is shown by the blue arrow labeled  △ET_CV; the pink arrows △ET_SWUS, △P_SWUS, and △R_SWUS are the changes in averaged evapotranspiration, precipitation, and runoff across the 4-state SW U. S. region respectively; the four blue arrows around the boundaries of the 3-D box are the changes in water vapor transport into and out of the SW U. S. The change in averaged runoff to the Colorado River is ∆R_CRB. 

Will we in the United States begin to view water vapor as worthy of such investment to minimize its export?  If so, what are the implications for the downwind regions?  Will downwind neighbors clamor over water vapor rights as downstream neighbors now do with surface waters?  Will water policy need to address vapor transport as an emerging issue?

Human Impacts and the Water Cycle

A more subtle implication is this.  Clearly, humanity has been controlling the terrestrial component of the water cycle for millennia, through our global network of reservoirs, conveyances, groundwater pumping, irrigation, etc. But the consequences to the rest of the water cycle have received much less attention.

In some cases, the impacts are local: it’s no secret that Phoenix is more humid because of its reservoirs, and that the additional humidity provides a powerful amount of regional greenhouse warming.  Anecdotally, Central Valley farmers have been talking about greater levels of humidity for decades.

More irrigated rice fields, Sacramento River Valley, Sutter County, CA. Photo by Tom Myers

In other cases, we have not even considered the remote impacts, because, well, they are remote.  More importantly though, it’s also because the science has not yet led us to consider them.  Hopefully, ours and similar studies will change that.  When water management is practiced at the massive scales that it is today, there are without question, consequences for regional and global climate.

It stands to reason then that there are many other such anthropogenic water cycle loops in operation around the world, either closed (with the water returning to the source region) or open (with the water headed to a different destination).  We just haven’t identified them or tried to quantify how much water goes where.  It’s a complicated bit of accounting, but, it needs to be done.

Which brings me to my final point.  Society is struggling to cope with sustainable food and energy production in the face of global warming and water cycle change.  The only way to truly minimize our footprint is to build the knowledge base that will allow us to anticipate the full consequences of large-scale water management practices.

We haven’t tried to fully control the water cycle, and I am not suggesting that we can or that we should try. However, our work suggests that we are unintentionally exerting some control over the strength of the monsoon, Colorado River flows, and the return of the water back to California. Before we decide on and implement future large-scale water management options, we must begin to work towards a full understanding of the scope of their local and remote consequences, to all of the land, atmospheric and oceanic branches of the water cycle.

Peixoto and Kettani were hardly mad scientists. They were without a doubt way ahead of their time.  Likewise, it is not mad science to consider the full implications of individual environmental management practices.  It is responsible, modern science, and effective management and policy cannot be developed without it.  There are many knobs, dials and levers that can be adjusted in our Earth system.  Right now, most of them are being operated independently.



Jay Famiglietti is a hydrologist and Senior Water Scientist at NASA's Jet Propulsion Laboratory. He is also a professor of Earth System Science at the University of California, Irvine, where he was Founding Director of the UC Center for Hydrologic Modeling. Jay's research group uses satellites and develops computer models to track changing freshwater availability around the globe. Jay is a frequent speaker and an active science communicator. His team's research is often featured in the international news media, including the New York Times, the Los Angeles Times, The Economist, CNN/Fareed Zakaria GPS, Al Jazeera, National Public Radio, BBC Radio and others. Jay also appears in the water documentary called 'Last Call at the Oasis.'