Multi-Scale Land-Atmosphere Interactions: Modeling Convective Processes from Plants to Planet

Principal Investigator(s):
Scott Denning, Colorado State University

Thunderstorms are very important to the Earth’s climate, providing most of the water in summer rain, reflecting solar radiation back to space, releasing huge amounts of heat from the condensation of water vapor, and vertically mixing pollutants and other components of the air. Atmospheric scientists call these vigorous processes “convective processes.” Big thunderstorm updrafts are typically less than a few miles across, yet climate models have grid cells that are as much as 100 miles wide. Partly as a result of this mismatch between huge grid cells and much smaller clouds, much of the uncertainty in projections of future climate change comes from differences in the effects of clouds in different climate models.

Another feature of the Earth system that varies a lot on small scales is the landscapes with which we’re all familiar. Fields of corn, soybeans, or wheat are arranged in a complicated mosaic with suburban landscapes, golf courses, parking lots, fallow fields, and reservoirs. Each of these different landscapes interacts differently with the air above, producing variations in temperature, humidity, and wind over scales of a few hundreds of meters. Like clouds, this kind of small-scale variation is hard to represent in climate models whose grid cells are the size of Connecticut, yet these variations influence local climate quite a lot.

We are studying the way these two kinds of fine-scale features (convective clouds and landscapes) interact with one another to influence the larger scale climate, with the aim of improving the way these small-scale processes are represented in climate models. We hope that this research also leads to better fundamental understanding of the way convective storms form and get organized, which may also lead to improvements in weather forecasts.

There are three main components to the work we’re doing.

First, we will look in great detail at a specific day in which a line of thunderstorms moved across a site in central Oklahoma at which DOE maintains a range of instruments to observe clouds, rain, winds, temperatures, and other weather properties. We will compare the results of very fine-scale models of these convective clouds with the observations to test how well the models work. We will then use the models to analyze the role of small-scale variations in the landscape such as plant cover, soil moisture, and temperature in the way the storms developed and moved on that day. In particular, we’re interested in the role of landscape features in breaking up pools of cold air that flow out of the bottom of the thunderstorm clouds. The “cold pools” are an important part of how convective storms spread and move across the region, and can play a major role in the development of severe storms.

Second, we will use three different models of clouds and rainfall to simulate an entire summer at DOE’s central Oklahoma facility. One of the models is the one that we will have previously tested for that one very well-observed storm system. A simpler model represents the same physical processes in the clouds and rain, but replaces the vegetated land surface with a homogeneous landscape that is the same everywhere across the region. The third is a much simpler model that is commonly used to represents clouds and rainfall in global climate models. Each of these models will simulate the weather of the whole summer, and can be compared in detail to the real weather that is observed by the DOE instruments. The point is to understand exactly what we gain by each level of additional complexity and small-scale variation in the interaction between the landscape and the atmosphere as they influence clouds and rainfall, compared to reality. In other words, we will use these experimemnts to measure what we lose by making the kinds of simplifying assumptions used in global climate models.

Finally, we will simulate the current global climate with these very same three models (lots of small-scale variations; variations in the air but not the land; and just big grid cells for both land and atmosphere without the fine-scale variations). Using what we learn at the DOE Oklahoma site, we will then interpret the three different versions of today’s global climate around the world in terms of the impacts of fine-scale variations on clouds and rainfall. We will pay special attention to the interactions between variable landscapes and fine-scale clouds over grasslands and croplands where conditions are similar to those on the Southern Great Plains of the USA.

We expect that what we learn about the role of fine-scale interactions between landscapes and clouds in these experiments can be used to build better models of the global climate. We also hope that the improved fundamental understanding of these processes will also lead to better weather forecasts, especially for heavy summer rainfall and severe weather.