Temperature Trends in the Lower Atmosphere - Understanding and Reconciling Differences

Surface It is likely that errors in the homogenized surface air temperature data do not contribute substantially to the large-scale differences between trends for different levels because these errors are very likely to be smaller than those for the upper air data. • Systematic local biases in surface trends may exist due to changes in station exposure or instrumentation over land, and due to the small number of measurements over a number of regions of the Earth, including parts of the oceans, sea ice areas, and some land areas. Such biases have been documented at the local and regional scale, but no such effect (including any significant urban bias) has been identified in the zonal and global averages presented in this Report. On large spatial scales, sampling studies suggest that these local biases in trends are likely to mostly cancel through the use of many observations with differing instrumentation. • Since all known bias adjustments have not yet been applied to sea surface temperature data, it is likely that errors remain in these data, though it is generally agreed that these errors are likely to be small compared to errors in radiosonde and satellite measurements of the upper air, especially for the satellite era.