{"id":298186,"date":"2024-01-23T09:25:00","date_gmt":"2024-01-23T08:25:00","guid":{"rendered":"https:\/\/climatescience.press\/?p=298186"},"modified":"2024-01-23T09:25:03","modified_gmt":"2024-01-23T08:25:03","slug":"a-test-of-skill-for-climate-models","status":"publish","type":"post","link":"https:\/\/climatescience.press\/?p=298186","title":{"rendered":"A Test of Skill for Climate Models"},"content":{"rendered":"\n
\"\"<\/figure>\n\n\n\n

<\/p>\n\n\n\n

From Watts Up With That?<\/a><\/p>\n\n\n\n

Richard Willoughby<\/p>\n\n\n\n

The temperature measured across the globe in October 2023 has set a new record high anomaly.  With that as the backdrop, this essay looks at the predictive skill of climate models in forecasting the record.<\/p>\n\n\n\n

Anomalies \u2013 A Brilliant Deception<\/strong><\/p>\n\n\n\n

Climate botherers are fixated on temperature anomalies and have convinced sufficient number of voters that 1.5C rise in global average surface temperature (GAST) is a critical threshold for the globe.  October 2023 came perilously close to breaching the threshold so it deserves closer analysis as to what it means and how well various climate models have managed to forecast this outcome.<\/p>\n\n\n\n

Temperature of the Lower Troposphere<\/strong><\/p>\n\n\n\n

This analysis begins with the examination of the temperature of the lower troposphere produced by the satellite remote sensing system (RSS) and displayed below in Chart 1.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The data is available at\u00a0KNMI<\/a>\u00a0but was last updated in July 2023, which is usually the yearly peak; reaching minus 0.16C in that month.\u00a0 October is typically 2C lower that July.\u00a0 A key feature of Chart 1 is the trend; rising at 2.1C\/century.\u00a0 The regression coefficient is only 6.29% due to the significant annual swing in the range 3 to 4 Centigrade degrees.\u00a0 However, if this data is reduced to an anomaly relative to the same month over a 30 year period to iron out the annual variation, the result gets closer to a trend line as shown in Chart 2.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The trend line has almost the same slope; increasing at 2.1C\/century as in Chart 1 but now the regression coefficient improves dramatically to 71.4%.  Given the high correlation, it is reasonable to expect this trend to continue for the next 76 years such that the RSS TLT will reach 1.1C in July 2100.  And there are certain to be many more months prior to that year when the 1.5C threshold will be exceeded.  In fact, October 2023 probably exceeded the 1.5C threshold.  The RSS TLT only needed to reach minus 2.85C in October 2023 to exceed the threshold.<\/p>\n\n\n\n

Taking A Global Perspective<\/strong><\/p>\n\n\n\n

Reducing the monthly GAST to a single anomalous value is somewhat deceptive.\u00a0 It is as if all surfaces are created equal so it is OK to average them over a whole range of different surface.\u00a0 But this is not based on sound thermodynamic principals.\u00a0 A first step in getting a better appreciation of the significance of record temperatures in 2023 is to examine the warmest month in 2023 with the same month in 1982, the early days of remote satellite sensing to observe how the entire global temperature has responded regionally per Image 1.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The first observation is that the entire globe has not warmed.  Some locations are up to 7.1C cooler than in July 1982 and some are 7.4C warmer.  The warmest regions are in the Northern Hemisphere while the coolest regions are in the Southern Hemisphere.  The area average increase in temperature is 1.3C corresponding to 3.2C\/century so slightly above the trend.<\/p>\n\n\n\n

Note also that even this satellite determined temperature is not global.  The coldest regions near the poles are excluded from the measurement.<\/p>\n\n\n\n

Global Land<\/strong><\/p>\n\n\n\n

It is evident that the thermodynamic response of land surfaces and water surfaces are different.\u00a0 Land responds faster and over a wider temperature range for a given thermal energy input than water.\u00a0 Image 2 indicates the temperature difference from October 1982 to October 2023, the month of highest anomalous GAST, for just land surface, excluding Antarctica, based on the Global Historic Climatology Network (GHCN) measurement.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The area average temperature increase over the land is 2.5C, corresponding to 6.1C per century.  Some land surface has cooled by as much as 4.3C while some warmed by up to 14.1C.  The highest temperature increase is in the high latitudes in the Northern Hemisphere likely to have been permafrost back in 1982.  <\/p>\n\n\n\n

Global Ocean<\/strong><\/p>\n\n\n\n

The Reynolds optimally interpolated ocean surface temperature data set is not yet available for the NH summer and fall of 2023 so the ocean surface analysis is stepped back a year by comparing August 2022 to August 1982 as displayed in Image 3.<\/p>\n\n\n\n

\"\"<\/figure>\n\n\n\n

The increase in area average temperature is 0.5 Centigrade degrees with a range in difference of minus 3.6 to plus 13.1 Centigrade degrees.  The region with the most warming is near the Russian coastline in the Arctic and likely due to earlier annual loss of sea ice in 2022 relative to 1982. <\/p>\n\n\n\n

There are large regions of the oceans in the SH that are now cooler than 1982.  The cool zone along the tropical Pacific was likely warmer in 2023 than 2022 due to El Nino conditions prevailing.  Combining the 0.5C for ocean with the 2.5C for land, gives an area average increase of 1.13C, which is slightly lower than the 1.3C over the same period for the RSS TLT.  It is more than likely that the October 2023 ocean temperature increase is slightly more than 0.5C due to El Nino conditions prevailing in 2023. <\/p>\n\n\n\n

Predictions from Selected Climate Models<\/strong><\/p>\n\n\n\n

The following four images, numbered 4 to 7, compare the predicted results for global temperature based on CMIP6 SSP85 input from various climate prognosticating groups based in regions as follows:<\/p>\n\n\n\n