I was writing the summary post for my recent series that compare observed Global Surface Temperatures to the IPCC and CMIP3 Climate Model outputs when I realized I had not presented Land Surface Temperature Anomaly comparisons.
This post illustrates observed Global Land Surface Temperature anomalies using CRUTEM3 data. It is the Land Surface Temperature portion of the HADCRUT3 combined Land-Plus-Sea Surface Temperature anomaly data that the IPCC used in their AR4 Chapter 9, Figure 9.5 comparisons to Multi-Model Ensemble Mean data. In this post, the CRUTEM3 observations are compared to the Multi-Model Mean of the CMIP3 archive, which contain the models the IPCC used as the source for AR4. (The “land only points” option was used at the KNMI Climate Explorer to mask the marine air temperature portion.) The base years for anomalies are the same as the IPCC used in their Figure 9.5 (1901-1950), which is why the data in the comparisons of the model mean and observations appear so awkward during the late warming period, Figure 1. But as we can see the trend of the Multi-Model Mean aligns itself well with the observed rate at which surface temperature rose during the late warming period.
Based on the comparisons in the earlier posts, see here, here, here, and here, we would expect the modeled and observed trends during the late warming to agree well. Likewise, we would expect trends during the early 20th Century “flat temperature” period and early warming period to agree quite poorly, since they didn’t agree well with Sea Surface Temperature or Combined Land+Sea Surface Temperature datasets. That leaves the question, how well does the trend of the models match the trend of the Land Surface Temperature anomaly observations during the mid-20thCentury “flat temperature” period.
The answer: Not so good. More realistically, the match is terrible. See Figure 2.
Note: The early warming period peaked (or ended) in 1938 with the Land Surface Temperature observations. This is different than the Sea Surface Temperature data and Combined Land+Sea Surface Temperature anomaly datasets, which both peaked in 1944.
Figures 3 and 4 show, as expected, how poorly the trends of the Multi-Model Mean match the observed Land Surface Temperature anomaly trends during the early warming period (Figure 3), and the early “flat temperature” period (Figure 4).
JUST IN CASE…
…someone thinks NCDC or GISTEMP (dTs) land surface temperature data would offer significantly different results, as shown in Figure 5, there is little difference between the three datasets.
The climate models used by the IPCC show little to no skill at being able the reproduce the Land Surface Temperatures of the 20th Century, especially during the early “flat temperature” and early warming periods. These failures are similar to the other datasets examined in this series of posts. The failure of the models to match the decline in Land Surface Temperature during the mid-20th Century “flat temperature” period is similar to the results using the recently released updated version of the Hadley Centre’s Sea Surface Temperature data, HADSST3.
The CRUTEM3, NCDC, and GISS dTs Land Surface Temperature data is available through the KNMI Climate Explorer, specifically at the Monthly observations webpage, and the model mean data for land surface temperature anomalies is found at the Monthly CMIP3+ scenario runs webpage by selecting the “only land points” option on the “Field” page.