The "hiatus" doesn't exist.

That is the conclusion of a new study published in Science yesterday.  Tom Karl and his co-authors used a newly available database of weather station data that combined the Global Historical Climatology Network (GHCN) used by NOAA and NASA with over 40 other historical data sources, effectively doubling the size of the available land temperature data set.  They applied the same corrections for changes in location, urban heat island effect, etc as with the GHCN-only data set and used the same algorithms to calculate the global average over land.  To get a global average, they used the Extended Reconstructed Sea Surface Temperature dataset version 4 (ERSST4), which better integrates ship-based temperature data and buoy-based temperature data, and merged it with their new land data.  Karl et al. then created a third global temperature average that also fills in the gaps between weather stations in the polar regions.

Their new data set shows much higher trends than the GHCN-only data, especially since 1998.  Their analysis shows >2x the warming since 1998 (0.086ºC/decade) as the GHCN-only data (0.039ºC/decade).

Figure 1 from Karl et al. 2015 showing the trends over selected time periods as calculated using GHCN-only data (circles), their new data (squares), and their new data combined with polar interpolation (triangles).

This is mostly due to a much higher sea surface average (0.075ºC/decade versus 0.014ºC/decade) since 1998.  The upshot is that there is no statistically difference between the 1951-2014 warming rate and the 1998-2014 warming rate, meaning that the so-called "hiatus" does not exist.

The main conclusions from Karl et al. is that the appearance of a pause was due to a) a short time period (changing the period by just two years had a significant impact on the calculated trend) and b) a cherry-picked start year (1998, the warmest El Niño currently on record), and c) artifacts due to incomplete data.  This matches the main thrust of other research into global temperature data (i.e. Foster and Rahmstorf 2011, Rahmstorf et al. 2012, Rohde et al. 2013, Cowtan and Way 2014, the various Berkeley Earth papers) and various blog posts (i.e. Tamino and my own writings).

The "pause since 1998" is dead.  Long live the "pause since ______" that deniers come up with next.