UK tourism travel, accommodation

FLIGHTS ONLY:    Avro   Excel Airways   FlyBe    Freedom Flights   Virgin Atlantic   Lastminute.com    Wegelo
FLIGHTS+ACCOM:   Alpha Rooms   Aspro   Cosmos   EBookers    Going Places   Henoo   LowCostHols    MyTravel   Superbreak Thomas Cook
ACCOM ONLY:    HotelPronto.com   Hostelbookers   Hotel.com   KeyCamp Holidays   Laterooms   Somewheretostay.com   Quickrooms  

Multiple Regression Modelling Multiple Regression Modelling of the Global Surface Temperature Record and the Satellite Temperature Record Multiple Regression Modelling of the Global Surface Temperature Record and the Satellite Temperature Record Multiple Regression Modelling of the Global Surface Temperature Record and the Satellite Temperature Record

Multiple Regression Modelling of the Global Surface Temperature and the Satellite Temperature Records

In the course of studying regional temperature changes during the twentieth century (see 1990s: the warmest decade of the twentieth century?), two interesting conclusions were reached:

1) the 1990s was not the warmest decade of the twentieth century in all regions: only in six of 11 regions studied to date.

2) in some cases, highly significant correlations were found between regional temperature anomalies 1900 to 2001 and regional climate indicators like El Nino Southern Oscillation (SOI), Pacific Decadal Oscillation (PDO), the North Atlantic Oscillation (NAO) and the Arctic Oscillation (AO). These are not new discoveries but nevertheless are major contributory factors as to why the 1990s was the warmest decade in some regions. Table 1 shows the strength of the relationship between these indicators and regional temperatures with graphs in most cases.

Table 1   Regions in which significant correlations were found between annual temperature anomalies 1900-2001 and regional climate indicators (click on hyperlinks to see graph showing the relationship): North Atlantic Oscillation (NAO); Arctic Oscillation (AO); Pacific Decadal Oscillation (PDO); El Nino Southern Oscillation (SOI).

NAO                        r AO                             r PDO                  r SOI                     r
_______________________________________ ________________________________________
       
Europe                0.562   Central Russ. Fed.    0.493    West USA      0.284 West Canada  -0.443   
West  Russ. Fed.   0.579  East Russ. Fed         0.587 West Canada   0.614  
Eastern Canada    -0.379 Arctic Rim            0.208*    
________________________________________________________________________________

*  denotes correlation coefficient (r) significant at 5% level.  All other correlations significant at 0.1% level.

Given the strength of the correlations shown in Table 1, it logically follows that the global surface temperature record and the satellite record of temperatures in the lower atmosphere may also be correlated with these regional climate indicators. Indeed, the El Nino Southern Oscillation is widely acknowledged as a major factor in global temperatures: the warmest year in both the surface temperature record and the satellite record is 1998 which was characterised by an unusually intense El Nino. This article states "Given that a moderate El Niño event is taking place, there is a good chance that 2003 will be warmer than 2002" again acknowledging the particular importance of the El Nino to global temperatures. El Nino & Global Temperature  shows a strong relationship between El Nino and the satellite record of the temperatures of the lower atmosphere.

Sunspots are known to vary in a 11 year cycle (graph).  Clearly, there has been an increase in sunspot numbers during the twentieth century and this could be related to the warming during the century. Those who consider sunspot numbers to be an important climate factor argue that extra energy is stored in the oceans during periods of high sunspot numbers as over land, any warming in summer is soon negated by cooling in the winter months. The IPCC consider variations in solar activity as a minor factor (Climate Change 2001: the scientific basis) which is somewhat surprising considering the the huge amounts of energy entering the earth's atmosphere from the sun every second of every day. Their view is "Mechanisms for the amplification of solar effects on climate have been proposed, but currently lack a rigorous theoretical or observational basis."  Here are a number of articles on solar activity and climate: Solar Activity: a dominant factor in climate dynamics; Solar Activity controls El Nino & La Nina; Sunspots & Climate; Little Ice Age a Global Event; Sunspots & the Solar Cycle. Solar activity is just one of many aspects of climate that is uncertain and controversial but annual sunspot numbers are included in the analysis because they clearly are a factor that has increased during the twentieth century.

In this multiple regression analysis of the global surface temperature record (1900 to 2001) and the satellite record of the temperature of the lower atmosphere (1979 to 2001), the following factors were tested as being major explanatory variables: atmospheric concentrations of CO2; El Nino Southern Oscillation (SOI); Pacific Decadal Oscillation (PDO); North Atlantic Oscillation (NAO); and sunspot counts.

Modelling the Global Surface Temperature Record

Modelling the Satellite Record of Atmospheric Temperatures

Arguably, as the surface temperature record is highly imperfect as a measure of global temperature, the model described above is not necessarily of great significance.  The satellite record has 100% global coverage and is technically superior to the surface temperature record.  A model that can successfully predict variations in the satellite record would have greater scientific credibility.

Both models convincingly explained the variations in the two global temperature records. The size of the coefficients in the two equations may be used to rank the importance of the explanatory variables. For the surface temperature record, in descending order, the ranking of importance is: PDO; CO2; SOI & sunspots.  The comparative figures for the satellite data are: PDO; SOI; CO2 & sunspots. As already mentioned, the satellite record has to be considered the better of the two temperature records so logically the satellite temperature model is the better model.  The fact that SOI had the greatest single correlation with satellite temperature is consistent with scientific knowledge of the overriding importance of SOI to global temperatures. PDO was the most important factor in both the surface and satellite temperature model and the implication that the temperature of the Earth's largest ocean has a big influence on global temperatures is not only logical but also unsurprising. In the satellite model, SOI was ranked second in importance while  CO2 was ranked third  but as CO2  increments by a regular amount each year, it is proof that there is an underlying warming trend in the data but not proof that CO2 is the cause of the underlying warming trend.  The $64,000 question is whether or not rising atmospheric concentrations of CO2 is the cause of the warming trend of 0.08o C per decade 1979-2002 or is there another explanation?  This graph suggests that most of the warming owed to the 1998 El Nino alone but 2002 was the second warmest year on record for both the surface temperature record and the satellite record. As the 2002/2003 El Nino commenced late in 2002 and is a much weaker event than the 1997/1998 episode, the fact that 2002 was so warm is perhaps surprising. While this fact may lend support to the anthropogenic global warming theory, the re-occurrence of low temperatures (-30 to -50oC) in some countries such as Finland, Russian Federation and Canada during winter 2002/2003 in the northern hemisphere after a run of years when such low temperatures have not been widespread argues against the theory as it is in winter in high latitudes where the warming effect of more CO2 in the atmosphere is expected to be greatest.  These conflicting signals are symptomatic of the complexity of the Earth's climate and show that the global warming debate is far from settled.