Identifying human influences on atmospheric temperature

Footnotes

• ↵¹To whom correspondence should be addressed. E-mail: santer1{at}llnl.gov.

• This article is part of the special series of Inaugural Articles by members of the National Academy of Sciences elected in 2011.

• Author contributions: B.D.S., C.A.M., S.S., K.E.T., L.T., F.J.W., and T.M.L.W. designed research; B.D.S., J.F.P., C.A.M., C.D., P.C., J.P., and L.J.W. performed research; B.D.S., J.F.P., C.A.M., C.D., P.C., P.J.C.-S., P.J.G., J.P., S.S., L.T., and L.J.W. analyzed data; C.A.M., F.J.W., and C.-Z.Z. contributed key observational datasets; and B.D.S., J.F.P., C.A.M., C.D., P.C., J.A., P.J.C.-S., N.P.G., P.J.G., J.L., J.P., S.S., P.A.S., K.E.T., L.T., P.W.T., M.F.W., F.J.W., T.M.L.W., L.J.W., and C.-Z.Z. wrote the paper.

• The authors declare no conflict of interest.

• This article is a PNAS Direct Submission.

• This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1210514109/-/DCSupplemental.

• *Other CHEM models (such as GISS-E2-R [p2] and GFDL-CM3) substantially overestimate observed ozone loss in certain regions and at certain times of year. The fact that some CHEM models underpredict observed ozone loss and others overestimate observed ozone trends helps to explain why we do not find even larger TLS trend differences between the O3+V case (which excludes CHEM models) and the BASE case (which includes CHEM results).

• ^†Previous multimodel studies have found either small (45) or large (47) impacts of stratospheric ozone changes on tropospheric temperature. The ozone-induced tropospheric temperature signals inferred from such multimodel analyses can be obscured by intermodel differences in other applied external forcings and model differences in climate sensitivity (48).

• ^‡Model temperature fields are spatially complete and sampled at uniform time intervals, whereas MSU-based temperature measurements are not spatially complete and not sampled at uniform time intervals. These sampling differences tend to inflate the high-frequency variance of the observations. The RSS percentile realizations attempt to account for this variance inflation (26).

• ^§For each of the four atmospheric layers except TLT, the O3+V fingerprint is searched for in 14 individual observational datasets (RSS v3.3, UAH v5.4, STAR v2.0, and 11 RSS percentile realizations). For TLT, there are only 13 model vs. observed comparisons, because STAR does not provide TLT information. There are, therefore, a total of (3 × 14) + 13 comparisons.