AR4 concluded global land-surface air temperature (LSAT) had increased over the instrumental period of record, with the warming rate approximately double that reported over the oceans since 1979. Since AR4, substantial developments have occurred including the production of revised data sets, more digital data records, and new data set efforts. These innovations have improved understanding of data issues and uncertainties, allowing better quantification of regional changes. This reinforces confidence in the reported globally averaged LSAT time series behaviour.
Global Historical Climatology Network Version 3 (GHCNv3) incorporates many improvements (Lawrimore et al., 2011) but was found to be virtually indistinguishable at the global mean from version 2 (used in AR4). Goddard Institute of Space Studies (GISS) continues to provide an estimate based upon primarily GHCN, accounting for urban impacts through nightlights adjustments (Hansen et al., 2010). CRUTEM4 (Jones et al., 2012) incorporates additional station series and also newly homogenized versions of many individual station records. A new data product from a group based predominantly at Berkeley (Rohde et al., 2013a) uses a method that is substantially distinct from earlier efforts (further details on all the data sets and data availability are given in Supplementary Material 2.SM.4). Despite the range of approaches, the long-term variations and trends broadly agree among these various LSAT estimates, particularly after 1900. Global LSAT has increased (Figure 2.14, Table 2.4).
Since AR4, various theoretical challenges have been raised over the verity of global LSAT records (Pielke et al., 2007). Globally, sampling and methodological independence has been assessed through sub-sampling (Parker et al., 2009; Jones et al., 2012), creation of an entirely new and structurally distinct product (Rohde et al., 2013b) and a complete reprocessing of GHCN (Lawrimore et al., 2011). None of these yielded more than minor perturbations to the global LSAT records since 1900. Willett et al. (2008) and Peterson et al. (2011) explicitly showed that changes in specific and relative humidity (Section 2.5.5) were physically consistent with reported temperature trends, a result replicated in the ERA reanalyses (Simmons et al., 2010). Various investigators (Onogi et al., 2007; Simmons et al., 2010; Parker, 2011; Vose et al., 2012a) showed that LSAT estimates from modern reanalyses were in quantitative agreement with observed products.
Particular controversy since AR4 has surrounded the LSAT record over the United States, focussed on siting quality of stations in the US Historical Climatology Network (USHCN) and implications for long-term trends. Most sites exhibit poor current siting as assessed against official WMO siting guidance, and may be expected to suffer potentially large siting-induced absolute biases (Fall et al., 2011). However, overall biases for the network since the 1980s are likely dominated by instrument type (owing to replacement of Stevenson screens with maximum minimum temperature systems (MMTS) in the 1980s at the majority of sites), rather than siting biases (Menne et al., 2010; Williams et al., 2012). A new automated homogeneity assessment approach (also used in GHCNv3, Menne and Williams, 2009) was developed that has been shown to perform as well or better than other contemporary approaches (Venema et al., 2012). This homogenization procedure likely removes much of the bias related to the network-wide changes in the 1980s (Menne et al., 2010; Fall et al., 2011; Williams et al., 2012). Williams et al. (2012) produced an ensemble of data set realizations using perturbed settings of this procedure and concluded through assessment against plausible test cases that there existed a propensity to under-estimate adjustments. This propensity is critically dependent upon the (unknown) nature of the inhomogeneities in the raw data records. Their homogenization increases both minimum temperature and maximum temperature centennial-time-scale USA average LSAT trends. Since 1979 these adjusted data agree with a range of reanalysis products whereas the raw records do not (Fall et al., 2010; Vose et al., 2012a).
Regional analyses of LSAT have not been limited to the United States. Various national and regional studies have undertaken assessments for Europe (Winkler, 2009; Bohm et al., 2010; Tietavainen et al., 2010; van der Schrier et al., 2011), China (Li et al., 2009; Zhen and Zhong-Wei,2009; Li et al., 2010a; Tang et al., 2010), India (Jain and Kumar, 2012), Australia (Trewin, 2012), Canada (Vincent et al., 2012), South America, (Falvey and Garreaud, 2009) and East Africa (Christy et al., 2009). These analyses have used a range of methodologies and, in many cases, more data and metadata than available to the global analyses. Despite the range of analysis techniques they are generally in broad agreement with the global products in characterizing the long-term changes in mean temperatures. This includes some regions, such as the Pacific coast of South America, that have exhibited recent cooling (Falvey and Garreaud, 2009). Of specific importance for the early global records, large (>1°C) summer time warm bias adjustments for many European 19th century and early 20th century records were revisited and broadly confirmed by a range of approaches (Bohm et al., 2010; Brunet et al., 2011).
Since AR4 efforts have also been made to interpolate Antarctic records from the sparse, predominantly coastal ground-based network (Chapman and Walsh, 2007; Monaghan et al., 2008; Steig et al., 2009; O’Donnell et al., 2011). Although these agree that Antarctica as a whole has warmed since the late 1950s, substantial multi-annual to multi-decadal variability and uncertainties in reconstructed magnitude and spatial trend structure yield only low confidence in the details of pan-Antarctic regional LSAT changes. In summary, it is certain that globally averaged LSAT has risen since the late 19th century and that this warming has been particularly marked since the 1970s. Several independently analyzed global and regional LSAT data products support this conclusion. There is low confidence in changes prior to 1880 owing to the reduced number of estimates, non-standardized measurement techniques, the greater spread among the estimates and particularly the greatly reduced observational sampling. Confidence is also low in the spatial detail and magnitude of LSAT trends in sparsely sampled regions such as Antarctica. Since AR4 significant efforts have been undertaken to identify and adjust for data issues and new estimates have been produced. These innovations have further strengthened overall understanding of the global LSAT records.