Bulletin of the American Meteorological Society, 89, 459-470
Towards "seamless" prediction: Calibration of climate-change projections using seasonal forecasts
Tim N. Palmer, Francisco J. Doblas-Reyes, Antje Weisheimer and Mark Rodwell
ECMWF, Shinfield Park
RG2 9AX, Reading, UK
Reliable climate predictions of regional precipitation are
needed to guide decisions on infrastructure investment to
adapt to anthropogenic climate change (ACC). For example,
following two consecutive dry winters in Southern England,
is a national water grid required to prevent prolonged water
shortages? Critical to answering these questions are the
latest ACC projections of precipitation change from the
ensemble of global climate models. However, are these
projections reliable; in particular is agreement between
models a reliable indication of forecast accuracy? Motivated
by recent proposals to explore the utility of "seamless"
prediction methods across weather and climate timescales,
we propose a quality-control criterion for multi-model ensemble
projections of climate-change, based on corresponding
seasonal-forecast skill estimates and related attributes-diagram
analysis. From an analysis of the DEMETER seasonal-forecast
database, we use our proposed criterion to cast doubt on the
reliability of consensus in the latest ACC projections of
precipitation change over Europe. By analysing additional
simulations, we conclude that improved reliability may require
global climate models to be integrated at substantially higher
resolutions than is currently possible.