Geoscience Reference
In-Depth Information
concern for virtually all aspects of climate modelling and climate change research. In a
context where the climate modelling community is increasingly focussing its efforts on
regional climate change impacts and biogeochemical (e.g., carbon and aerosols) climate
feedbacks, improving our understanding of cloud-climate interactions and assessing our
confidence in the simulation of cloud processes and feedbacks in climate models is
imperative.
In the following sub-subsections, we discuss current research on the parametrisation of
cloud and precipitation processes, we highlight common systematic errors such as failure
to reproduce the diurnal cycle of clouds and precipitation over land, and we discuss issues
surrounding the representation of hydrological processes in the mid-latitudes.
2.3.1.1 Model Representations of Cloud and Precipitation and Their Evaluation The
formulation of cloud microphysical parametrisations is very important for simulation of the
hydrological cycle and for model evolution because they modify the three-dimensional
structure of temperature and humidity directly (e.g., condensation/evaporation) or indi-
rectly by interacting with other parametrisations (e.g., radiation) and the large-scale
dynamics. Therefore, the evaluation and improvement of these parametrisations is crucial
to improving our weather forecasts or increasing our confidence in climate projections.
Improvements to the representation of clouds, humidity and radiation in models have
been a focus in several modelling groups in recent years (e.g., Collins et al.
2006
; Wilson
et al.
2008
; Salzmann et al.
2010
). For example, in the MetUM, the new PC2 cloud scheme
(with prognostic cloud and condensate) improves cloud and humidity distributions and, in
combination with more advanced aerosol schemes, results in improved radiation balance
(Walters et al.
2011
).
Model Intercomparison projects, including the Third Climate Model Intercomparison
Project (CMIP3), have always exhibited a large range of cloud-climate feedbacks (Webb
et al.
2006
; Dufresne and Bony
2008
). There are so many factors or physical processes that
may potentially contribute to this spread, that interpreting the origin of inter-model dif-
ferences has turned out to be difficult, and that designing specific observational tests to
assess the different feedbacks has remained elusive.
Satellites have proven to be very helpful tools for model evaluation because they
provide global or near-global coverage, thereby giving a representative sample of all
meteorological conditions. However, satellites do not measure directly those geophysical
quantities of interest, such as the amount or phase of cloud condensate. They measure the
intensity of radiation coming from a particular area and direction in a particular wave-
length range (Bodas-Salcedo et al.
2011
). A great deal of research has been conducted into
producing satellite retrievals of many different geophysical variables, such as water
vapour, atmospheric temperature, cloud properties and land surface products (e.g., Chahine
et al.
2006
; Wylie et al.
2005
; Schaaf et al.
2002
). Satellite retrievals have been used in
numerous studies to analyse the performance of NWP and climate models (e.g., Allan et al.
2007
; Pincus et al.
2008
).
In the last two decades, a different avenue has been followed to exploit satellite data in
model evaluation: the use of forward modelling of basic satellite measurements from
model fields (Bodas-Salcedo et al.
2011
). Simulators have been developed that mimic the
observational process and essentially acknowledge the issue that a retrieval produced by a
satellite might not be directly comparable to a model variable, giving rise to multiple
values of geophysical quantities from different sensors and retrieval algorithms. A pilot
model
intercomparison
using
the
CFMIP
Observation
Simulator
Package
(COSP;
