through the National Research Council. from irregular and heterogeneous obser- model results to the much finer levels of
ASA recommends that there should be vations. Evaluating the advantages and detail required for policy makers.
greater involvement by statisticians in disadvantages of different interpolation High dimensional data analysis.
future reviews of the state of climate approaches (referred to as infilling in The results of climate models and cur-
science conducted by the CCSP. climate applications) could be very help- rent observational data sets are extreme-
Although there are numerous oppor- ful. This research area contains many ly multi-dimensional and difficult to
tunities for increasing the participa- opportunities for the development and visualize and analyze. A commonly-used
tion of the statistical community in the fitting of sophisticated space-time models technique is principal components analy-
IPCC, CCSP, and other assessment pro- to sparse data. sis (often known as empirical orthogonal
cesses, the ASA notes that there is already Climate models. Complex computer functions analysis in the geophysical sci-
extensive and healthy collaboration models based on physical laws are used ences). This standard method can miss
between statisticians and climate scien- to simulate the dynamics of the Earth’s the nonlinear and non-Gaussian attri-
tists in basic research on climate change. atmosphere, ocean, and sea ice. These butes often associated with geophysical
Furthermore, climate science contin- models provide a basis for exploring the processes. Statisticians have the opportu-
ues to offer many statistical challenges physical relationship among different nity to contribute improved analytic tech-
that are currently not being tackled and components of the climate system and niques for interpreting geophysical data.
many opportunities for collaboration also for making projections of future It is very difficult to present all of the
with geoscientists. The ASA strongly climate states. The design and analysis of information concisely in a manner that
urges statisticians to collaborate with computer experiments is an area of sta- can be understood by decision makers.
other scientists in order to advance our tistics that is appropriate for aiding the Dimension reduction and data presenta-
understanding of the nature, causes, development and use of climate models. tion techniques are needed for comparing
and impacts of climate change. Statistically based experimental designs, spatial maps, explaining what is being
The workshop convened by ASA not currently used in this field, could presented, and determining how to
identified several specific areas where be more powerful. It is also important describe the confidence levels associated
statistical science can make a contribu- to understand how to combine the with projections obtained from noisy
tion. Besides the obvious benefit to the results of experiments performed with and spatially incomplete data.
geosciences these topics may well push different climate models. Despite their Human health effects of climate
the boundaries of statistics and suggest sophistication, climate models remain change. The available evidence sug-
new methods, algorithms, and theory. approximations of a very complex sys- gests that certain extreme events with the
Interpreting and synthesizing cli- tem and systematic model errors must potential to impact human health may
mate observations. Observational be identified and characterized. Model be increasing in frequency as a result
data from different measurement plat- evaluation is an area of active research, of global warming. For example, the
forms and sensors, such as satellites, with many opportunities for informed IPCC concluded that there have been
weather balloons, surface stations, or statistical input. Finally, assessing the more intense and longer droughts, and
ocean drifter buoys often represent cli- many sources of uncertainty in climate an increase in the frequency of hot days,
mate processes at very different spatial projections requires innovative tech- hot nights, heat waves, and heavy pre-
or temporal scales. Moreover, observa- niques for better quantifying and, where cipitation events. Climate change can
tional records from earlier parts of the possible, reducing these uncertainties. also impact human health through its
20th century are sparse, particularly in Quantifying uncertainty and formal effects on the vectors carrying diseases, or
southern oceans and in the developing assessment of confidence intervals on through the complex interplay between
parts of the world. Even in the satellite observations and model projections are large-scale warming and local air pollu-
era – the best observed period in Earth’s core activities of statistical science, and tion. Links between local air pollution
climate history – there are significant become particularly appropriate when and human mortality are already well-
uncertainties in key observational datas- climate models are used to identify established. One issue that has emerged
ets. Reduction of these uncertainties will human effects on climate or to estimate in recent research is the extent to which
be crucial for evaluating and better con- climate-change impacts. individual extreme events, such as the
straining climate models. Statisticians Regional and local effects of climate 2003 European heat wave, can be attrib-
can advise on how best to combine data change. There is great need for taking uted to global warming as opposed to
from different sources, how to identify coarse-resolution projections from glob- other possible explanations, including
and adjust for biases in different mea- al and regional climate models down to natural causes. A fruitful line of research
surement systems, and how to deal with estimates for small areas. Indeed, translat- is to explore how concepts borrowed
changes in the spatial and temporal cover- ing the large scale understanding of cli- from epidemiology, such as relative risk,
age of measurements. The climate science mate processes to changes at a local level are potentially valuable in this context.
community often requires regular fields is a grand challenge in climate research. Papers along these lines have started to
of geophysical variables, such as surface Statisticians can provide valuable input appear in the climate literature, but there
temperature, which must be derived to this problem of downscaling climate- is much scope for further development.
JANUARY 2008 AMSTAT NEWS 7
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84