My original intent was to gain a basic understanding of the global warming issue and the IPCC's AR4 document seemed to be a reasonable starting point. The overall AR4 document is more than I wanted to deal with so I poked through the Summary for Policymakers, using the main document for reference. My goal was to understand whether climate model temperature predictions can be trusted to agree with reality.
Given the controversy over data from surface stations, I decided to look at data from satellites with the idea that these more modern measurements should be reliable with less controversy. Per Feynman, I tried to look at what they started with and where they ended up, then tried to judge whether the result was reasonable in a broad-brush way. That is, I didn't try to follow the details but instead looked at the pictures summarizing the analysis. The pictures chosen present an enormous amount of information in an easy to digest form but require some ability to visualize sections through the information and how they should look from another perspective.
The following picture from AR4 (Chapter 9 pg 675) shows the signature associated with global warming due to CO2 and other greenhouse gasses.
My interpretation of the above picture is that according to the climate models the temperature in the troposphere should be rising more rapidly than at the surface. Also, temperature in the troposphere should be rising more rapidly at the equator than toward the poles.
The following picture, also from AR4 (FAQ pg 104), shows observed tropospheric temperature trends in the right panel as determined by satellite measurements. This shows that the tropospheric warming trend is lower in a band centered at the equator. This is opposite to the model results shown above. In addition, the warming trend near the equator for the surface doesn't look much lower than the warming trend for the troposphere.
That is, the model doesn't seem to agree with observations in two separate ways.
I posed the above as a question (off topic) on RealClimate and received a helpful response with a reference. My observation is old news, climate scientists have investigated this discrepancy between model and observations and written a report.
As an engineer - not a climate scientist - I read the "Executive Summary" and extracted the following:
Graph G in this extract supports my earlier observation from AR4 of less tropospheric warming in a belt at the equator and the fact that this conflicts with model based predictions of a CO2 forcing signature (see AR4 pg 675).
The first paragraph of the abstract in the Executive Summary:
Reading this, one would think that discrepancies between model and observations have been reconciled. This is contradicted by the last paragraph on the next page in Executive Summary (an interpretation of the observed discrepancy between predictions and observations in the tropics):
It all depends on what the meaning of "significant" is.... In concert with the disclaimer for Graph G in the text at the bottom of Fig. 4 these items raise my concern for objectivity in this report. The last sentence should not be in the Executive Summary because it clearly shows a bias that the writer can't back up with facts. To me, this writing conflicts with Feynman's thoughts on an objectively written paper where one takes pains to present both sides of an issue fairly.
One of the problems with a newbie poking around at a subject like global warming is that one is likely to plow ground that has already been plowed. And so it is with my poking above; starting with the greenhouse gas signature noted above Monckton's paper does a similar analysis. He doesn't use information from AR4 to show a conflict within the document (as I do above) but uses similar (and more detailed) info from other sources. His information (based on observations) shows the troposphere as slightly cooler at the equator so it agrees with my observation about the troposphere temperatures shown in AR4 FAQ Fig. 3.1 (above). Unfortunately, the tone of his paper is confrontational so it is unlikely to be taken seriously.
The AR4 Technical Summary has an interesting chart showing projected global temperature vs various scenarios:
From the text below the chart I noted that model results for the period prior to 2000 were not shown for the various scenarios; this prevents evaluating the model for most of the time for which observation data is available. The chart shows the scenarios are well above the observed data at the time of publication (seems like the models should match existing observations to be credible). So, I extracted a section of the chart showing the area where both model results and observations are available:
The best fit of observed values to any model output is provided by the "Committed CO2" line, where CO2 levels are assumed to be stable (since 2000) rather than rising; the other scenarios modeled assume various rates of CO2 rise. I took the liberty of adding (in purple) an eyeball approximation to the observed global temperature data (from HADCRU through April 2008) since publication of AR4. The smoothed global temperature per HADCRU's graph is about the same in 2000 and 2008 so that is what my addition tries to show. Crude as it is, this update makes the departure from the AR4 model scenarios glaringly obvious.
The long blue bar at the end of HADCRU's April chart indicates that 2008 is continuing the down trend in temperatures (and it may be accellerating). Looking at the chart, in the past similar shape maximums generally trend down for a while. If this holds true it suggests that the models aren't accounting for a major input, an input with a much larger effect than greenhouse gasses.
HADCRU is pro-AGW so they would not show a decline in global temperature unless it were real. For example, they revised their smoothing algorithm because it made the recent temperature decline appear more dramatic than it really is - but they didn't notice when the same processing error made the temperature rise appear more dramatic a couple years earlier. Looking at the green bar for 2008, the corrected value, it is still showing a major drop in global temp. Scroll down their page to see how filtering changes things - the recent temperature drop is rapid so it is masked by heavier filtering.
Since HADCRU is pro-AGW they spin this by using a mid-2007 chart and talking about long term trends rather than the obvious downturn in the data since 2002. This type of spin may work on readers in the short term but you can't fool Mother Nature :-)
In the prediction (scenario) business, if an investment advisor provided projections similar to those from AR4 and the investments performed as shown by the observed temperature chart one would likely be looking for a better advisor...