Monsoon Modeling

Off the coast of Acapulco, blowing over the waters touching Mexico and drawing inland from the sea, the North American Monsoon rolls into the Southwest each year to dump its last drops of rain.

The summer rainy season in the Southwest is part of a larger climate pattern affecting Northern Mexico.
Credit: University Corporation for Atmospheric Research

“That’s a tremendous amount of energy going northward,” said David Mitchell, associate research professor in atmospheric sciences at the Desert Research Institute.

Complex monsoon processes make storm intensity hard to predict from season to season, and climate change raises even more uncertainty for the future of the monsoon. Capturing the monsoon in a global climate model proves difficult because the small-scale processes driving monsoons are difficult to catch in the large-scale grids used in global climate models.

Monsoons flood Arizona and New Mexico with 35 and 45 percent of the states’ precipitation, respectively, and shower Mexico with 60 percent of their annual rainfall, according to the Desert Research Institute, an environmental research institute based in Nevada.

The North American Monsoon provides rain for one of the most rapidly growing regions—northern Mexico and the Southwest U.S.—yet it is “perhaps the least understood of all large-scale circulation patterns” affecting the continent, according to a 2004 report prepared by the Climate Prediction Center.

“If we had a better understanding of the monsoon, we could probably do a better job predicting it,” Mitchell said. Although its annual visit is predictable, individual years vary by storm arrivals and seasonal rainfall.

Predicting monsoon intensity in a changing climate remains difficult. Driving forces of climate include complex, interconnecting systems in the air, land, and sea. Atmosphere and ocean interactions influence climate, and vegetation, soil moisture, and land elevation also play a part in monsoon events.

Because the monsoon provides almost half of the yearly rainfall for parts of the Southwest, understanding the monsoon’s relationship to climate provides valuable information to managers of agriculture, water systems, wildlife, fire mitigation, and cities. Forecasting monsoon rainfall can help determine when floods or fires might accompany the summer heat and help stakeholders prepare for extreme events.

Can global climate models accurately predict monsoons?

Global climate models, or global circulation models, assemble current scientific understanding of Earth’s climate into computers and simulate a future climate. The models forecast climate patterns ranging from the next season to half a century away.

Mitchell researches monsoon causes and acknowledges the complications in trying to predict the monsoon with a global climate model.

“I guess I remain skeptical that the models can do a decent job,” Mitchell said.

Research by Mitchell and other scientists suggests sea surface temperature drives monsoon activity. Temperatures in the Pacific Ocean hover around 66 degrees Fahrenheit, while those in the Gulf of California, or Sea of Cortez, float around 82 to 86 degrees.

Mitchell’s research shows the Gulf of California sea surface temperatures playing a lead role in the local monsoon’s intensity and timing. According to his study, heavy rain poured in the Southwest when sea surface temperatures reached about 85 degrees or more in the northern Gulf. When temperatures later dropped lower than this temperature, the monsoons waned. Substantial warm season rainfall over the Sierra Madre in northwestern Mexico did not occur until sea surface temperatures in the Gulf exceeded 79 degrees.

Warm water from the Gulf of California may “provide the fuel” to power monsoon storms in the Southwest, especially Arizona, Mitchell explained.

But a global climate model’s spatial resolution does not pick up Gulf of California sea surface temperature, Mitchell explained.

“So it’s apples and oranges,” he said referring to the Pacific and Gulf comparison. Global climate model spatial resolution is often a few degrees in latitude and longitude. The models’ broad grid-scale blends the Gulf into the Pacific, as if the 20 degree temperature difference doesn’t exist.

Unresolved resolution

Climate models require relatively fine resolution to represent regional weather and climate processes. Most global atmospheric models use a resolution too coarse to represent seasonal climate forecasting and climate change projection.

The warm season in Arizona is the most difficult to simulate, as a typical global atmospheric model simply cannot represent thunderstorms and thus the rainfall they cause during the summer monsoon, said Christopher Castro, assistant professor in atmospheric sciences at The University of Arizona.

“You can’t go out and just simulate a cloud,” Castro said. “Even if the grid-scale is 30 kilometers [18.6 miles], there’s no way you can represent that… because you can’t increase the resolution of the model down to a one-millimeter grid spacing to represent the formation of precipitation, which is what it would take.”

Global atmospheric models typically use 100-kilometer (62-mile) grid spacing or larger. “Within that scale is really where you need to know what’s happening,” Castro said.

Landscape features like the Mogollon Rim in Arizona and New Mexico play an important role in the formation and spatial patterns of summer monsoon season storms, and can be difficult to represent in global and regional climate models.
Credit: ©Jim Wark, Airphoto

But simply increasing resolution cannot eliminate the dependence on model parameterizations. Model parameterization involves choosing key variables, or drivers—precipitation, land-atmosphere interactions, and atmospheric radiation—affecting climate, and applying equations to these variables so the computer can simulate Earth’s natural systems.

“Four different parameterization schemes to represent precipitation will give four different results, even in the same atmospheric model. So then it is up to the model user to choose which parameterization gives the most reasonable result,” Castro said.

Parameterization uncertainty is and will probably always be a barrier in the use of atmospheric models to represent weather and climate at any scale, Castro said.

“Even our basic physical understanding is still in process,” Castro said referring to monsoon field research. Field research, like the North American Monsoon Experiment (NAME), allows scientists to learn about monsoon processes like the daily evolution of thunderstorms. Such detailed observations can then be compared with results from atmospheric models and can potentially improve existing model parameterizations or lead to the development of new ones.

If the models can’t represent the monsoon physical process well in a given grid spacing, just increasing resolution may not solve the problem. “It’s not a panacea,” Castro said.

In spite of these inherent difficulties, Castro has used a regional atmospheric model to retrospectively simulate North American summer climate and its interannual variability over the past fifty 50 years, suggesting regional models may add substantial value over global models. Castro recently obtained funding from the National Science Foundation to do this.

Regional perspectives on monsoon activity, coupled with continual research on the physical processes, should help stakeholders who need to know how the monsoon will behave in the future.