An early online release of GLERL researcher Brent Lofgren’s paper entitled “Physically Plausible Methods for Projecting Changes in Great Lakes Water Levels Under Climate Change Scenarios” can be found on the American Meteorological Society’s Journal of Hydrometeorology website.

In the paper, Dr. Lofgren and his co-author, Jonathan Rouhana, explore two different ways to model the effects of climate change on evapotranspiration (the movement of water from the land to the atmosphere as the combined result of evaporation and transpiration), and, subsequently, on the water levels of the Great Lakes.

Predicting how climate change will affect the water levels of the Great Lakes is a tricky business. To answer questions like this, it is often best to use models. Modeling is central to what scientists do, both in their research as well as when communicating their explanations. Within their models, scientists study relationships between variables in nature and then apply those relationships to possible future scenarios with one or more tweaked variables.

However, earth systems are so complex and have so many moving parts, that it’s almost impossible to capture them completely in an equation or series of equations. The beauty of modeling, is that it allows scientists to start with a small amount of data and, as time goes on, to build up a better and better representation of the phenomenon they are explaining or using for prediction.

Sometimes, particularly when modeling climate change, problems arise with so-called empirically-based models. Empirically-based models are created by making observations about two or more variables over a certain time period and under certain conditions, and inferring relationships from those observations. Often, those models don’t hold up when conditions change.

An alternative is physically-based models, which use the laws of physics (like conservation of mass, energy, etc.) to make predictions. Complexity is still a hurdle, but the laws of physics hold up no matter what—even when the climate changes.

Dr. Lofgren’s paper details issues with an empirically-based model widely used in Great Lakes research, the Large Basin Runoff Model (LBRM). From the abstract:

This model uses near-surface air temperature as a primary predictor of evapotranspiration (ET); as in previous published work, we show here that its very high sensitivity to temperature makes it overestimate ET in a way that is greatly at variance with the fundamental principle of conservation of energy at the land surface. The traditional formulation is characterized here as being equivalent to having several suns in the virtual sky created by LBRM.

Several suns in the sky – wow! In the most extreme case, this method of calculating evapotranspiration behaves as though there were 565 suns. 

In the context of climate modeling, “The LBRM oversimplifies the physics of the interaction between the earth and the atmosphere,” says Dr. Lofgren.

This doesn’t mean the LBRM isn’t useful in specific instances (e.g. short-term forecasting), or that you shouldn’t ever trust empirically-based models. It just means that different types of models have their place in different circumstances, and that the LBRM probably isn’t the best choice for modeling hydrologic response under climate change conditions.

Scientists often argue about the rightness of their model, and in the process, the model can evolve or even be rejected. Consequently, models are central to the process of knowledge-building.

Scientists who dare to create models know that their models will be scrutinized and tested. Research like Dr. Lofgren’s ensures not only that models are used appropriately with an acknowledgment of their limitations, but that they are continually improved upon.