Low (web) Quality - BALTEX

43
Settled and remaining issues in regional climate modelling with limited-area
nested models
René Laprise 1,4,5 , Ramón de Elía 2,4,5 , Daniel Caya 2,4,5 , Sébastien Biner 2 , Philippe Lucas-Picher 3 , Emilia
Diaconescu 1,4,5 , Martin Leduc 1,4,5 , Adelina Alexandru 1,4,5 , Leo Separovic 1,4,5 and Alejandro di Luca 1,4,5
1 Université du Québec à Montréal (UQÀM), Montréal (Québec), Canada, laprise.rene@uqam.ca; 2 Consortium Ouranos,
Montréal (Québec), Canada; 3 Danish Meteorological Institute (DMI), Copenhagen, Denmark; 4 Canadian Network for
Regional Climate Modelling and Diagnostics (CRCMD), Montréal (Québec), Canada; 5 Centre pour l’Étude et la Simulation
du Climat à l’Échelle Régionale (ESCER), Montréal (Québec), Canada
1. Introduction
A pragmatic approach to reduce the computing cost of highresolution
global climate models is to apply the high
resolution over only a subset of the globe, a technique
known as nested Regional Climate Modelling. Lower
resolution data, either simulated by Coupled Global Climate
Models (CGCM) or from reanalyses of observations, are
interpolated in time and in space on the high-resolution grid
of a limited-area Regional Climate Model (RCM) and serve
to define the lateral (and often the ocean surface) boundary
conditions (LBC). Like all models, RCMs are subject to a
variety of sources of errors and uncertainty, as well as to
scepticism about their potential usefulness. But in addition,
there are specific issues that arise with RCMs relating to
their geographically limited computational domain, the
nesting technique, the resolution jump between the RCM
and the driving data, the update frequency of the LBC, and
the imperfections in LBC data.
This presentation will summarise the work presented in
Laprise (2008), Laprise et al. (2008) and references therein.
2. Dynamical Downscaling Hypothesis
The ansatz behind the “dynamical downscaling” technique
is that an RCM, driven by large-scale fields at its LBC,
generates fine scales that are dynamically consistent with,
and somehow preconditioned by the LBC. Initialised and
driven by data without small-scale information, nested
models develop small-scale variance even in the absence of
strong surface forcing. Hence RCM are expected to act as a
kind of magnifying glass that reveals details that cannot be
resolved on a coarse mesh. The small scales represent the
main potential added value of high-resolution RCM. A more
controversial issue concerns the potential improvement of
the large scales in the case of driving by low-resolution
CGCM data.
The spin-up of fine scales proceeds fairly rapidly. Within a
few days in the simulation, atmospheric variance spectra and
forecast error spectra become independent of the time since
the simulation’s initiation and of the presence or absence of
small scales in the initial and LBC. The hypothesised
development mechanisms include fine-scale surface forcing,
hydrodynamic instabilities, mesoscale processes, and
nonlinear interactions cascading information from the large
to the small scales.
3. Deterministic Forecast Vs Climate Skill
In idealised forecast experiments for mid-latitude domains
with 100 by 100 grid points on a 45-km mesh, scales larger
than about 800 km appear to retain extended deterministic
predictability, while scales smaller than about 400 km loose
predictability in a fashion similar to global models. For the
small scales, the shorter the length scale, the shorter the
predictable time. Hence even when fine scales are present
initially and in the LBC, they do not retain deterministic
temporal coherence (at the right place at the right time)
beyond a day or so. This implies that part of the
downscaling is not deterministic, i.e. not entirely
determined or preconditioned by LBC. A nondeterministic,
“free” component exists (free in the sense of
its stochastic-like character) which is also reflected in the
presence of some level of internal variability (IV) in
nested models.
Hence small scales are generated by nested models, but
not with deterministic skill. Climate statistics of small
scales appear to be skilful though, lending some
confidence in the potential usefulness of RCMs in climate
simulations and climate-change projections. IV puts
however severe limitations on the usefulness of singleseason,
single-simulation experiments. While IV may not
be a major problem for climate applications as far as loworder
statistics are concerned (when the domain size is not
too big; see next section), as long as the user of an RCM is
aware of its presence in the interpretation of results.
4. Regional Domain Size
In idealised experiments, it is seen that fine scales develop
in RCM within a few days, and they have the right
amplitude and the right statistics. It has been shown
however that the full spin-up of small scales within the
regional domain requires rather large domains, particularly
in the upper troposphere in mid-latitudes where the flow is
strong. Due to the continuous transport of low-resolution
information from the lateral boundary, some distance is
required for the spin-up process to proceed. The spin-up
distance on the inflow side of domain is larger for stronger
ventilation flow through the domain. For mid-latitude
domains, this distance is thus larger in winter and at upper
levels due to stronger winds. The physically meaningful
portion of the limited-area domain must exclude the spinup
distance, in addition to any sponge or buffer zone used
as part of the lateral boundary nesting.
The downscaling skill has been documented through a set
of idealised “perfect prog” tests. Over small domains (e.g.
70 linear grid points), the small-scale transient eddies are
amplitude deficient, especially at upper levels. With larger
domain (e.g. of the order of 200 linear grid points), smallscale
transient eddies are simulated with the correct
amplitude, but with very little time correlation with the
reference. Lack of time synchronicity is of secondary
importance for climate applications, but must be taken into
consideration for process studies. Failure to reproduce the
correct monthly or daily anomalies may however have
important consequences for the downscaling skill in
seasonal prediction and even for some climate
applications.
By their nature, nested RCM require externally provided
data to drive them. Over mid-latitude domain, the
application of LBC is usually sufficient to control the
large-scale circulation through the RCM domain. There