Global Warming - GCM Model-E Overview

I have tried to understand the physics behind Anthropogenic Global Warming for some time. It is surprising how little real information is available on the internet. Don't get me wrong - there is a lot of raw data, but not much on how it is used. So, I was happy to stumble onto the source code for GCM Model-E, the model used by IPCC AR4 (2004/2007).

This model is maintained and provided for free by the NASA Goddard Institute for Space Studies (GISS), at Columbia University in New York City. While I was excited to find something this useful, I was disappointed to discover that the model's source is a combination of Fortran, Perl, and Unix shell scripts.

This is one of the models used by James Hansen and his associates.

Notes: On their pages, the model name is written without the dash - ModelE - which places the L and E close together. Because I think that this is difficult to read, I added the dash.
GCM - General Circulation Model

Running the Model

GCM Model-E was written to run on Unix systems. I made a few a few modifications so it would compile on Windows XP and a couple more so it would read the data files.

Unfortunately, this model requires about 60 minutes (1.5 hours without the -O2 optimization flag) for each month simulated. (This is way too slow to be useful to me.)

After running 10.5 hours (7 months of simulation), the model crashed on some bad data.

After I recompiled with the -O2 switch, the model ran for 57 minutes (simulating about 28 days) before it crashed with the following

  Program terminated due to the following reason:
  >>  In Radia: Q<0  <<

Interpreting the results

The model outputs 2 files per simulated month (about 25 megabytes combined). There are also a couple of 29 megabyte files that are periodically updated - I think that these are simply restart files.

I have not found any descriptions of either the input or output data formats, nor have I found software to analyze the results.

I am open to suggestions.

Discussion

This model runs very slow on Windows XP ... about 1 hour per month simulated. I don't see how anyone can use it like this. (Except, of course, those that have very fast processors.) I spent several days trying to analyze the code and make it faster. I found nothing specific wrong with the implementation. There is still a lot of code to look at, but I can't dedicate more time to this.

The model still crashes using the provided data - I don't know why.

I had wanted to try a few ideas and see what effect they might have.

What happens if CO2 is held constant or decreased
What happens if the Arctic ice pack completely melts
Is it better to plant trees or cut them down
What effect does the number of hurricanes have

Author: Robert Clemenzi