Models(2) Forgotten Layers:

Models(2) Forgotten Layers:

The comments on the blog have been pretty intense lately. Be good to each other. I’ll talk about the 1930s, when I know a little bit more. Maybe we can figure out some good model experiment. I want to continue with a thread I started a couple of blogs ago; the one on models. In the previous entries I wrote about the different types of models: heuristic, statistical, and physical. I wrote about what is assumed in building a model, and those things which are known with significant certainty. In this blog I want to talk more about the different types of models and this idea that keeps coming up that there things that scientists don't include because they are not "on message"

Here are the previous blogs on models.
Uncertainty and Types of Models
Models (1) Assumptions

Layers of Models: Climate models and weather forecasting models are closely related. Originally, climate models were “atmospheric” models. Since the late 1960’s climate models have evolved to include the atmosphere, the ocean, the land and soil hydrology, and sea ice. These are the primary component models of a physical climate model. Each of these components was initially developed in their own discipline of study; that is, ocean models were developed by oceanographers to study the ocean. Using this as an example, the ocean models view the atmosphere as a boundary condition, which provides temperature, wind stress and other parameters needed to “force” the ocean circulation. When these models are put together, often called coupling, each component provides interactive information for the other components. When the model is coupled it can simulate – or not – modes of variability that require the interaction of, say, the atmosphere and ocean (like El Nino). I have introduced two ideas here that will come back in the future; they are: motions that are forced and motions that are internal variability. Both are important in the climate problem.






Figure 1: "Components" of the physical climate system. These are the major component models of a climate model.


A couple of paths to go down:

First Path: The computational challenges that confront modelers are large. Each component model can expand in complexity and resolution to consume available computing resources. Resolution is how coarse or fine the model represents the Earth. When the component models are put together, they are generally simpler components than would be used in stand-alone studies. Often the simplification is a more coarse resolution.

Once these models are put together, however, that is not the end of the story. The component models still exist, and they can be used for experiments that require more complexity and resolution to understand the observations. As component models they can have as boundary conditions (forcing) either output from other models or observations. The purpose of these experiments is to investigate the processes that determine cause and effect. This allows evaluation of how well the process is represented in the coupled model.

It is this ability to perform focused experiments at the process level that tells me the confidence level that I can have in the coupled model. The Earth supports many scales of variability, and models are multi-scale. The robustness of climate predictions is based not just on the results of coupled climate models, but on analysis of this multi-scale system and the ability – or not – of the models to represent observations. Models are far from perfect, but as I have discussed earlier, we make decisions all the time with imperfect information.

Here is a link to the Community Climate System Model. There is data here from model simulations that you can investigate on your own. (Question: If you had the ability to be running a IPCC climate model in the background on your PC would you?)

Second Path: The components described above are the physical climate system. The biology in the ocean and on the land needs to be included. Chemistry in the atmosphere and the ocean and on land needs to be included. Aerosols, particulates in the atmosphere, need to included. There are efforts on all of these. Atmospheric chemistry is especially well developed, and there are climate models with coupled chemistry. Chemistry, however, requires inclusion of many species – a source of significant expense. Aerosols are important to both chemistry and climate; they are the focus of much of the current model development.

Biological modeling is less well developed than the atmosphere, ocean, land, and sea ice. The strategies for looking at the biology require taking a “big picture” view of the biology; for instance, what is the flux of carbon dioxide or sulfur species. There are also models and measurements of the Sun and its variability.

There are assumptions in all of these model components. Some are well founded, and some are crude. In no case is “nothing known.”

The Forgotten and that not included: Occasionally in the comments the statement is made that the climate scientists have ignored some important process. I frequently sit in meetings where I am learning a new subject, and I wonder whether or not something has been considered or forgotten. For the climate problem, I have yet to find something that has been truly forgotten. Every suggestion I have seen, I have followed up, and there is a substantial literature. That does not mean that the process should not be revisited, but I can’t find the forgotten. The credibility, the stubbornness, the obsessions, and the vanity of scientists doesn’t really let that happen.

r

Chapter 16: Fundamentals of Modeling ....