EdGCM - Project 1 - Orientation

Download, Install, Test

Goals

• Download the EdGCM codes and documentation
• Install the code on a PC or Mac
• Perform a test run
• Analyze and plot some of the output
• Observe variability year-by-year and, possibly, between different simulations for the same period
• Provide feedback on the advantages and shortcomings of EdGCM

A Note on Performing the Simulations

Instructions

• PC running Windows 7
• Macintosh running OS 10.6 or higher

It appears that the test run performed here (10-yr simulation) uses less than 250 mb for its output. The volume needed increases with the number of fields analyzed and the frequency of the analyzed output (e.g., monthly, seasonal or annual average time series).

Various types of information, including video tutorials and a FAQ are on the Documentation Page for EdGCM.

2. Download The EdGCM Manual (using this link). You may want to print parts of it, but note that the whole document is about 60 pages long. For now, read Chapter 1. Introduction to EdGCM.

You may also want to download the EdGCM Quick Start Guide

3. Obtain a copy of EdGCM. One way is to download EdGCM itself, where you can purchase a student copy for $29. Go to the EdGCM web site. There is lots of information on this page that may become useful. To purchase and download the software, find the "Purchase" link along the top of the window and follow instructions there. You may need to follow the "contact us" link under Student Pricing.

4. Once the software has downloaded, follow the installation instructions on pages 4-6 of the EdGCM Manual as appropriate for your computer. Please make sure you read Section 1.2! Note especially the instructions to avoid letting your computer "sleep" while running a simulation. Note also that the current EdGCM release will work on Macintosh computers using Intel processors, but it may run slower than on a PowerPC Mac. You can also run EdGCM on an Intel Mac running a Windows environment.

Work with EdGCM in this project will follow 4 steps:

1. Run the model
2. Post-process model output (that is perform some computations like averaging on the output after the model has run)
3. View the output
4. Reflect on what you learned about the model.

6. Run the model for a test case.
(WARNING! EdGCM developers keep upgrading the code, so the instructions that follow may not be correct in all details. When setting up the code, it is vital that you follow the instructions in the exact order they are given. More specifically (and note the key point 4 below):

1. Download EdGCM's installation file (version 3.2),
2. then download the EVA update.
3. Make sure the new EVA is copied into the folder where the previous EVA resided, so that the previous version is explicitly overridden.
4. *Only* at this point, do you download the EdGCM update
5. After downloading the EdGCM update, go into the EdgCM folder then into the "database" folder, and copy the updated EdGCM there, overriding the old one.

• (a) Start EdGCM by clicking on the "EdGCM 4D" icon that installation placed on your desktop.

• (b) A bar appears on the left side of your screen. At the top, it should say "EdGCM3.0 Toolbar" or something like that. Immediately below that, you should see a small window with "Simulation Controls", including pause and play buttons. Below that is another window with a "Run List". We will run a version of the "Sample_Control_Run"

• (c) To set up the run and see its features, go to the EdGCM pull-down menue under "Window" and choose "Setup Simulations".

• (d) You should run a short version of the Sample_Control_Run. To do this, choose Sample_Control_Run from the Run List by clicking on it. Then in the window below the Run List, push the "Duplicate" bar to make a copy of the Sample_Control_Run that you will be allowed to alter.

• (e) In the Setup Simulation window, note the top line of "General Info". Change the Run ID to something else, like "Test_Run_1". You can keep the start date as is, but we don't want to run 100 years just now, so change the year after "End on Dec. 31" to "1959". This will run the model 2 years. The model will actually start on December 1 of 1957. This appears to be a spin-up period used by the model whenever starting a new run (as opposed to re-starting (extending an existing run). Keep everything else unchanged for now.

• (f) Push the "Play" button (the one with the ">" arrow) to start the run. A new window with "EdGCM" drawn with letters appears. It shows some basic characteristics of the run you've set up. After the first hour of simulation ends, you see a "STOP 13". As the manual explains, that is a normal pause to allow you and the computer to make sure the setup is working (and correct).

• (g) Push the play button again to restart the simulation. Now the EdGCM window will slowly but surely start to run further, with a new line appearing at the end of each day, telling you how far the model has run.

• (h)Make a note of how long in real time the model needs to complete one month. This will give a rough sense of the rate simulations advance and can be used to estimate the length of longer (e.g., multi-year) runs.

• (a) In the "Window" pull-down menu, select "Simulation Library".

• (b) Highlight the run you just created, by clicking on it.

• (c) Then in the "Simulation Controls" window, push the button on the far right (">|"). You will see a new window asking you how far you want to extend the run. You are free to go as far as you wish (or dare!), but keep in mind that disk space gets used for storage. You should extend at least to the end of 1967.

• (d) Then push the Play button to set the run going again.

  Sometimes, this procedure does not seem to work. Asl an alternative, you can go to "Setup Simulations" in the "Window" pull-down menu and change the end date. Make sure the "lock" icon is unlocked (clicking on it locks or unlocks the page). Change the end date, then push the "play" button (">").

• (e) It appears that simulating just one additional year does not generate enough output for the analysis code to form its averages. Also, the producers of EdGCM recommend running for about 10 years to get plausible (statistically reliable) averages.

8. Analyze the output you created, using EdGCM tools. This step creates the post-processed files that we can use for plots and diagnosis. The User's Guide gives help on this in sections 3.6 and 3.7, but the presentation is rather terse. I suggest looking at it while going through these steps.

• (a) Start the analyzing software by entering ctrl+4 (Windows) or cmd+4 (Mac), or by going to the "4th Dimension" window and selecting "Analyze Output" from the "Window" pull-down menu.

• (b) You will get a window with "Analyze Output" at the top and a set of 5 tabs belwo the title.

• (c) The first tab is "maps". There are a set of variables with boxes next to them that you use to select the variables you want for maps. Alternatively, you can click on the "Label" box and select all of them. For a short, 10-year run, you can do that.

  Along the bottom is a set of time periods. Select "Monthly", "Seasonal" and "Annual" if they are not already chosen. These will produce averages for the periods selected, where "Seasonal" refers to the meteorological seasons (Dec-Jan-Feb, Mar-Apr-May, etc.).

• (d) Select the period to average. If you have run the full 10 years, then you can push the "Last 10" (years) button. Or, you can select some other set of years, but I suggest the last 10.

• (e) Push the "Average" button at the bottom left corner. A new window pops up and scrolls past many lines that indicate the softward is running through the averaging process for the output.

• (f) When completed successfully, the "Extract" button at the middle bottom can become active. First, highlight the time period in the "Averages" window at the bottom left that you want to use. This activates the "Extract" button. Push it. This puts the output into a format that other software can use, including the EVA software that comes with EdGCM.

• (g) After "Extract" is done, the "View" button becomes active. You can select one of the files in the "View Images" window, such as one with the time span in its name (which might be the only one). Select it by clicking on it, then push "View". This will start EVA and open the EVA software's window. You can also start EVA after input for it has been created by clicking on the EVA icon on your desktop.

9. View output - EVA

The menu windows for EVA are generally self-explanatory, but you may still find it useful to look at the EVA documentation page or simply download the EVA documentation. for further guidance. (NOTE: A username and password are needed to go to the documentation page, so this only works if you have purchased the software.)

• (a) EVA has 3 basic windows: data browser, plot window and a tool-bar window. All 3 might have started. If not, select a file, a variable and time period in the browser winder and push the "Plot" button in the lower right. This should open a plot window and accompanying tool bar. The plot should be a latitude-height cross-section of the atmosphere, according to the variable selected for plotting. Note the units, which may be non-standard!

  You can change between plots using the ToolBar. The ToolBar also lets you change features of the plot. Select "Grid" and then "Contour" to see what each produces. "Grid" shows the actual grid values (color-coded) from the model and thus gives a sense of the model's resolution. "Contour" may be easier to read, however. Select different variables and explore the different menus in the tool bar to see how you can make plots easier (or harder) to read.

  Look at the land coverage map with the "Grid" option. Most grid boxes in the model are 100% ocean or 100% land. Along coastal areas, however, the model treats grid boxes as partially land and partially ocean.

  Look at the topography map, using the "Grid" option. Note how the model's relatively coarse grid spacing (8 deg lat x 10 deg lon) smooths the topography. For example, in the real world, the Andes extend to over 6,000 m (19,500 feet) in many places, but the highest grid value is only about 2,000 m. Similarly, the Himalyas go over 8,000 m (~ 26,000 ft) in several places, but the highest grid is only about 4,000 m. Save a copy of your topography map by using the "Save Image" option in the "File" menu (upper left corner of the screen. You will hand it in to me.

• (b) Go back to the EdGCM analyzer window, by bringing the "4th Dimension" application to the foreground. Now select the "Vertical Slices" tab. Because you have already averaged over the time periods, you can simply select all the variables (using the "Label" selection), push the "Extract" button, choose the file you want under "View Images", and then "View". Then you will be put back in EVA. Now there are two files in the "File" list. Choosing the second gives you access to the vertical slices (latitude-height) plots you created. You can bring these up as you wish. Again, switching between "Grid" and "Contour" gives slightly different perspectives on the samee field, with "Grid" showing the field on the actual horizontal-vertical grid of the model.

• (c) Similarly, back in the "Analyze Output" window, you can get "Zonal Averages" and "Time Series". The latter will require you to push the "Time Series" button in the lower left corner to extract them.

  The former are longitudinal averages of fields at a single level, so the plots are simple lines that are functions of latitude. EVA will plot zonal averages over land only, ocean only or land/ocean together (GLOBAL). I do not know what the "Time or Region" choices mean for these fields.

  The time series are time series of annual averages for the variables you select. Surface air temperature (file name = SRFAIRTMP.xls) will be of initial interest. The default plot, showing averages for the globe, land, all ocean, open ocean or ocean ice will likely look like a set of straight lines. Select only the global time series (by deselecting the others). Also change the Min and Max values that set the range of the y axis to better reveal the variability of temperature from one year to the next. (I find that MIN = 13.0 and MAX = 13.4 works well - but tailor them to your runs.)

  Note the range of values for annual, global surface air-temperature. This is a measure of the internal, "natural" variability of the climate system, as simulated by the model. This plot...
  
  ... shows results from several runs. All were done on a PC except the two marked "Mac" in the key. Note that the two Mac runs produced identical output, but the PC runs did not. Why these differences occur, even in the first year, is not clear.

10. Form ensemble of results

Now all members of a Science Team need to combine results to produce a time series plot with multiple runs, like the figure above. Each team member should produce at least one time series to contribute to the figure. More time series are fine, but make sure you start each one from the same starting point and label each one so you can distinguish them.

You can extract the time series from the "SRFAIRTMP" file, which is simply an Excel spreadsheet file or a text file readily imported by Excel (depending on your operating system). A question arises: Where is this file located on your disk? You can find it by looking in the folders under the EdGCM simulation. Contact me or your team members if you cannot, but first note the guidance on EdGCM folders in Chapter 2 of the Basic Guide.

Tabulate the range of temperatures in each separate run and the spread between highest and lowest temperatures in all runs. Are they simiilar? Does your set of simulations have time series that cross over each other, or do some runs not overlap at all with others? Do they show equal amounts of interannual variability?

Science Team report - Send to me:

• Answers to the questions above
• A latitude-height cross-section plot from each team member and one topography map.

11. Reflection

Here I want you to think about what you did, what capabilities EdGCM displayed and what limitations EdGCM displayed. These issues are important for thinking about how EdGCM might be used.

Individual-member report - Send to me: a brief document (about a page or so) that addresses the following questions:

1. If another person asked you how to install and run EdGCM, what instructions would you give and why?
2. How easy or hard was running EdGCM? What features contributed to your answer? (This can include comments on the instructions as well as the code itself.)
3. In what ways does EdGCM promote or inhibit group interaction?
4. Consider some of the climate-physics issues discussed in the lectures.
   • Which ones might be difficult to explore using EdGCM as it is currently set up?
   • How might you alter it so you could do relevant simulations for one of these issues without making the user interface highly complex?

Important! There are no correct or incorrect answers to these questions, though there can be thorough or thoughtless responses. Reflections with higher grades will

• answer all questions,
• give specific features or examples supporting answers, and
• not ramble on with lengthy responses that avoid the first two points!

12. Futher review. From the Basic Guide, read

• the Overview pages (1-2)
• the brief description of the GISS GCM Model II in Chapter 2.

Go to the Computer-Lab Overview page.

Go to the main web page for Climate Modeling.