-
Annual Progress Report provided by theCanadian Regional Climate
Modelling Network
to CFCAS for the period of July 1 2005 – June 30 2006August 11,
2006
1.
Progress.....................................................................................................................................................
1THEME 1. Diagnostic and budget
studies..................................................................................................
1THEME 2. Optimisation of dynamical downscaling
................................................................................
12THEME 3. Development and improvements in the
CRCM......................................................................
17
2. Impact
.....................................................................................................................................................
263. Level of support
......................................................................................................................................
274.
Dissemination..........................................................................................................................................
275.
Training...................................................................................................................................................
33
1. ProgressThis section presents the activities of the Canadian
Regional Climate Modelling (CRCM) Network for theperiod from 1 July
2005 to 30 June 2006, according to the definition given in the
detailed proposal presentedto CFCAS on September 9 2003, entitled
« The Canadian Regional Climate Modelling Network », for
the 3-year period, 2003-2006, and in revised form on February 4,
2005.To summarise, the current Network includes 15 Co-Is, 17
Graduate Students, 1 Postdoctoral Fellow and 8Research Associates,
and the Network programme is distributed in 13 Research
Projects.The overall research goals of the Network remain under
three themes: Theme 1 – Diagnostics and budgetstudies, Theme 2 –
Dynamical downscaling approaches, and Theme 3 – Development of the
CRCM system.The scientific progress report by subprojects is given
hereinafter; it describes the progress made by each ofthe
subprojects towards achieving their objectives and milestones.
THEME 1. Diagnostic and budget studies
Sub-project 4.1.1 "Scale-selective diagnostic budget studies";
Laprise, Boer and Caya;RA, Soline Bielli, UQAM; PhD student,
Leticia Hernandez-Diaz, UQAM.
2005 – 2006 : • Generalize the scale-selective regional
diagnostic approach to variables such as energy,momentum or
enstrophy;
• Complete regional-scale Lorenz energy cycle budgets.This
subproject on diagnostic and budget studies aims at a better
undestanding of the added value of RCMover lower resolution
objective analyses or GCM used to drive the high-resolution
simulations. As a keyfactor in the energetics of earth’s climate,
the atmospheric water budget is used to apply the
scaledecomposition method developed by Bielli and Laprise (2006).
At first, a spectral Fourier analysis isperformed individually on
each term of the vertically integrated atmospheric water budget,
i.e. divergence ofthe vertically integrated moisture flux,
vertically integrated water vapour tendency, evapotranspiration
rateand precipitation rate. In order to study the regional
implication of scale interactions, each term isdecomposed
spectrally into three spatial scales: the first spectral band
represents the very large scales that arenot resolved by the RCM,
the second includes the scales that are resolved by the RCM and its
global large-
-
Progress Report – CRCM Network – 2005-2006 2 / 34
scale driving data, and the third band accounts for the small
scales that are only resolved by the RCM. Toisolate the
contribution of different spatial scales, spectral decomposition is
applied on each term on pressurelevels. The vertically integrated
moisture flux is then calculated during the integration of the
model for eachfiltered simulated atmospheric field, at every time
step of the CRCM and cumulated as time integralsbetween 6-hour
archival periods.Before diagnosing long climate simulation, the
scale-decomposed methodology was tested on a singlesimulated winter
(Bielli and Laprise 2006). Afterwards, the method has been
exploited to investigate theatmospheric water budget for different
seasons through analysis of 25 years of simulation by the
CanadianRegional Climate Model, for the period 1975-99 over North
America. Seasonal and monthly analysis ofvariances of the
vertically integrated moisture flux divergence and its large-scale
and small-scale parts werecalculated and analyzed (Figs. 4.1.1.1
and 4.1.1.2).
Fig. 4.1.1.1: Intra-seasonal climatological standard deviation
of the large-scale (left panel) and small-scale (right panel) part
of thedivergence of the vertically integrated moisture flux for the
winter season (December, January and February) from 1975 to
1999simulated by the CRCM.
Fig. 4.1.1.2: Same as Fig. 4.1.1.1 but for the summer season
(June, July and August).Theses figures show the intra-seasonal
climatological variability of the large-scale part and the small
scale-part of the vertically integrated moisture flux divergence
for winter (Fig. 4.1.1.1) and summer (Fig. 4.1.1.2)seasons over the
25 years of simulation. The small-scale part represents the added
value of the CRCM. Thestructures are quite different in winter from
those in summer. Moreover, the added value in winter is as largeas
the variability of the large-scale part while the small scales
dominate the variability over the continent insummer (Fig.
4.1.1.2). A paper focusing on winter and summer seasons will be
submitted to ClimateDynamics shortly.
-
Progress Report – CRCM Network – 2005-2006 3 / 34
At the same time, Leticia Hernandez-Diaz, PhD student, is
working on a paper describing the formalism ofenergy conversions at
the regional scale. The formulation is cast for a domain limited
horizontally andvertically, on pressure levels, and includes the
presence of topography through a Boer’s mask. Three forms ofenergy
emerge from the analysis: kinetic, enthalpy and potential
gravitational energy corresponding to themass in the domain. Only
kinetic energy is separated in zonal means and deviation from the
mean. Contraryto the Lorenz approach used extensively elsewhere,
the formulation does not include an arbitrary basic stateused to
separate thermodynamic energy into available and unavailable
portions. This is exact formally and isapplicable to the regional
scale.
The figure above shows the energy cycle for a domain limited
horizontally and vertically, in pressurecoordinates respecting
topography. The boxes represent the different types of energy:
enthalpy, potentialgravitational and kinetic. The arrows linking
the boxes represent the conversion terms or energy
transfers,whereas those leaving or entering the boxes represent the
energy fluxes through the domain boundaries,dissipation of energy
due to friction and the diabatic generation of energy. The vertical
integration isrepresented by the symbol |p. The square brackets
identify the zonal mean, which is in this case an averagealong the
X axis of the polar-stereographic grid, and the oblique brackets
identify the average throughout thedomain or global average.The
computer programming of these energy balance equations is well
under way and will soon be applied tothe specific case of African
easterly waves.
Sub-project 4.1.2 "A decadal-scale Canadian experiment"; Caya
and Laprise;PhD student, Biljana Music, UQAM; RA, Zav Kothavala,
UQAM.
2005 – 2006 : • Complete the validation of CRCM water and energy
budget over other Canadiancatchment areas with available
observations.
Climate models require the fluxes of radiation, momentum,
sensible and latent heat across the soil-vegetation-atmosphere
interface. These fluxes are provided by land-surface
parameterization schemes. Theappropriate level of complexity of
land-surface schemes for use in climate models is still an
unresolved issue.Simple Manabe-based parameterizations of land
surface processes (e.g. the ‘bucket’ representation of soil),as
well as very complex formulations (e.g. CLASS [1]), where exchange
between the atmosphere and the land
1 Verseghy, D.L., 1991: CLASS – A Canadian land surface scheme
for GCMs. Part I: Soil model. Int. J. Climatol. 11,
111–133.Verseghy, D.L., et al., 1993: CLASS – A Canadian land
surface scheme for GCMs. Part II: Vegetation model and coupledruns.
Int. J. Climatol. 3,347–370.
-
Progress Report – CRCM Network – 2005-2006 4 / 34
surface is controlled by plant physiology, continue to coexist.
Numerous investigations performed withdifferent climate models have
shown that the simulated climate is sensitive to the formulation of
land-surfaceprocesses.In her doctoral project, Biljana Music
investigates the sensitivity of hydrological cycle simulated by
twoCRCM versions to the formulation of land-surface processes.
Specifically, CRCM_3.7 and CRCM_4.0 differonly in the land surface
parameterization scheme. The CRCM_3.7 uses the ‘bucket’
representation of soil,while the CRCM_4.0 is coupled to CLASS. The
sensitivity study is carried out through a comparison of timeand
spatial averages of the water budget components, over the period
1961-99, for six river basins, calculatedfrom two CRCM simulations.
The studied river basins, Mississippi, Nelson, Churchill, Fraser,
Mackenzieand Yukon, cover the major climate regions of North
America and differ in size and topography.Then, in order to
evaluate the CRCM ability to simulate all water cycle components at
the scale of a largeriver basin, an integrative approach linking
both the atmospheric and terrestrial branches of the
hydrologicalcycle (Music and Caya 2006) is applied over the
Mississippi River Basin (Fig. 4.1.2.1).
Fig. 4.1.2.1. Annual mean of the water budget components in
mm/day. W (kg/m2) represents the storage of atmospheric
water(precipitable water), (M+S) (kg/m2) is the storage of water
soil moisture (M) and the accumulated snowpack (S), E (kg/m2) is
theevapotranspiration, P (kg/m2) is the precipitation, C (kg/m2) is
the water vapor flux convergence, and finally R (kg/m2) is the
totalrunoff.
The results show that the simulated hydrological cycle at the
scale of the Mississipi Basin is more realisticwhen CLASS is used.
The simulated hydrological cycle is sensitive to the choice of the
land-surface scheme;
-
Progress Report – CRCM Network – 2005-2006 5 / 34
size, location and topography differences of the studied river
basins are responsible for their differenthydrological responses to
the change of land-surface scheme. Finally, the implementation of
CLASS inCRCM_4.0 causes an increase of annual mean precipitation
for Mississippi, Nelson and Churchill, which isconsistent with the
increase of annual-mean evapotranspiration (Mississippi) and
annual-mean convergenceand runoff (Nelson and Churchill). In
contrast, Mackenzie, Fraser and Yukon basins indicate a decrease
ofthe annual-mean precipitation that is related to a decrease of
the annual-mean evapotranspiration (Mackenzie,Fraser and Yukon) and
water vapour convergence (Fraser).In the meantime, Dr. Zav
Kothavala has been evaluating the overall performance of the GEM
Limited-AreaModel (LAM) compared to the newest version of the CRCM
as a step towards migrating to the GEM-LAMmodel as the next model
for regional climate simulations at the CRCM Network. The
CRCM_4.1.1 includingthe land-surface scheme CLASS 2.7, is the
newest operational version of the Climate Simulations Team
ofOuranos which is used to produce high-resolution climate-change
simulations. This model inter-comparisonis being conducted under
the auspices of the Inter-comparison Transferability Study (ICTS).
Eight regionalclimate modelling groups from Europe and North
America, are participating in this GEWEX sub-project. Dr.Kothavala
has been performing a series of diagnostics to evaluate the
performance of these RCMs in sevenregions of the globe (Fig.
4.1.2.2), with a special attention over North-America domain (Fig.
4.1.2.3).
Fig. 4.1.2.2. Illustration of the Inter-Continental Scale
Experiments (ICTS) domains.
Fig. 4.1.2.3 ICTS North-America domain used by CRCM(121x121 grid
points). Red crosses indicate theCoordinated Enhanced Observation
Period(CEOP) stations. The two used sites are locatedin the right
corner: Fort Peck (48.31N-105.1W)and Old Black Spruce forest
(53.98N-105.11W) in southern Saskatchewan.
-
Initial results show that the newest version of the CRCM with
the CLASS land-surface scheme and the GEMmodel with the ISBA
land-surface parameterization produce simulations close to the
observations over theNorth American continent (Fig. 4.1.2.4).
Fig. 4.1.2.4. Comparison of the frequency of observed daily
precipitation to that simulated by the MRCC and GEM models.
FortPeck (top panel). BERMS Old Black Spruce Forest (lower
panel).
The figure above shows the comparison of daily precipitation
simulated by the CRCM (i.e. MRCC) andGEM models compared to the
field observations for two sites: Fort Peck and Old Black Spruce
forest. Boththese points fall near the centre of the model domains
(see Figure 4.1.2.3) and are not likely to be affected bythe
boundary conditions in the sponge zone. The observed data were
collected hourly, from July 1 toSeptember 30 2001, at both sites as
part of GEWEX [Global Energy and Water Cycle Experiment] GAPP,and
the MAGS Boreal Ecosystem Research and Monitoring Sites (BERMS) for
Fort Peck and the Old BlackSpruce forest. The total daily
precipitation amounts are placed in ten bins. At the Fort Peck
site, both modelsshow the same number of days with zero
precipitation. Since it is not practical to measure
precipitationbelow 0.2 millimeters, the observations at both sites
show no amounts in these bins. For both sites, theCRCM model shows
a greater frequency of daily precipitation falling as drizzle
compared to the observationsand the GEM model. The GEM model is
closer to the observations at the BERMS site.This type of analysis
reveals the differences in the convective schemes of the two
models. Dr Kothavala ispreparing a publication to summarise this
model intercomparison shortly.
Sub-project 4.1.3 " Regional-Scale Seasonal Prediction Project
"; Brunet, Jones, Laprise, Caya and Zwiers;MSc student, Etienne
Tourigny, UQAM.
2005 – 2006 : • Evaluate the added value of GEM-VR against
lower, uniform-resolution configurationGEM.
In this project, MSc student Etienne Tourigny is assessing the
feasibility of using a high-resolution RCM tosimulate seasonal
climate anomalies, over central and South America, associated with
ENSO (El Niño –Southern Oscillation) SST variability. Two RCMs are
being used in this assessment: the Rossby Centre
-
Progress Report – CRCM Network – 2005-2006 7 / 34
Regional Atmospheric Model Version 3 (RCA3) and the limited-area
version of GEM (GEM-LAM). TheRCMs have been configured to encompass
Central America and the Eastern tropical Pacific (see
Figure4.1.3.1) and integrated using ECMWF Reanalysis data as
lateral boundary conditions and observed SeaSurface Temperatures
(SSTs).
Fig. 4.1.3.1 Illustration of the domain. The topography contours
are every 500 m.
The RCMs have been run for the period February to November for
each of the years 1970 to 2002. In thismanner a variety of El Niño
and La Niña events are prescribed through either the SST field or
the analysedlateral boundary conditions. Our analysis of these
results will concentrate on 3 questions:1. Given nominally the
correct SSTs and external large-scale forcing, how accurately do
the respectiveRCMs simulate the regional climatology over the
Central and South American regions?2. How accurately do the RCM
physical parameterisation packages respond to anomalous SST forcing
in theEast tropical Pacific associated with ENSO variability?
Furthermore, are the RCMs capable of simulatingseasonal climatic
anomalies over Central and South America under the influence of
ENSO SST variability?3. Does increased RCM resolution allow for
improved depiction of regional climate anomalies over Centraland
South America?Performing these integrations in hindcast mode, with
analysed and observed SSTs is a prerequisite to doingthe same type
of analysis with GCM predicted boundary conditions.To address the
questions listed above, RCA3 and GEM have been integrated with
0.33° grid mesh and thesimulated seasonal climatologies evaluated
against observations. Composite El Niño and La Niña seasonalmeans
have been developed from observations, reanalyses and RCM
simulation results and therepresentation of seasonal timescale,
climatic anomalies over Central America from the respective RCMs
isbeing assessed.
-
Progress Report – CRCM Network – 2005-2006 8 / 34
Fig. 4.1.3.2 Precipitation difference (%) by season.
Figure 4.1.3.2 shows the observed and RCA3-simulated composite
mean precipitation anomalies (El Niño –La Niña), normalised by the
respective climatological values (e.g. model or observed
climatology) andexpressed as a percentage difference. Results are
presented for the summer season (JAS) preceding the peakEl Nino
event (defined as December with the maximum SST anomaly in the East
Pacific) and 3 months after(AMJ +1) the peak El Niño occurrence.
The top panel shows percent precipitation anomalies from the
GPCPsatellite data set [2] and the lower panel from the land-only
CRU data set [3]. The central panels show theequivalent RCA3
results. The large increase in precipitation during El Niño years
over the Central and eastPacific is relatively well captured by
RCA3, as are remote increases in precipitation over the tropical
Atlanticand Amazon regions. Central America is observed to be drier
in El Niño years, with a normalized-percentagereduction in
precipitation relative to La Niña years of ~50%. This reduction is
also simulated in RCA3 but isoverestimated, being closer to a 100%
reduction. The reasons for this signal amplification are currently
underinvestigation.A small RCA3 model domain has also been
configured with 0.15° grid mesh, centred on Central America,and
will use results from RCA3 at 0.33° as lateral boundary conditions
to make equivalent high-resolutionintegrations over Central
America. This will aid in determining the benefits of increased
resolution indownscaling seasonal climatic anomalies forced by ENSO
variability. These runs are presently in progress.
2 Global Precipitation Climatology Project (GPCP)
http://www.gewex.org/gpcpdata.htm3 Mitchell T.D., and P.D. Jones
2005: An improved method of constructing a database of monthly
climate observations and associated high-resolution grids. Int. J.
Climatol. DOI: 10.1002/joc.1181.
-
Progress Report – CRCM Network – 2005-2006 9 / 34
Sub-project 4.1.4 " Atmospheric Models Intercomparison Project
(AMIP-II – SGMIP)"; Zadra, Côté,Jones, Laprise and Caya;RA, Katja
Winger and Dr Zav Kothavala, UQAM, MSc student, Jonathan Mainville,
andInternship Student, Marc Verville, UQAM.
2005 – 2006 : • Continue experimentation with a
variable-resolution version of GEM in SGMIP mode, i.e.multi-annual
climate simulations made with analysed sea surface temperatures and
sea ice.
The Global Environmental Multiscale (GEM) model has been used
operationally by the CanadianMeteorological Centre to produce
short- and medium-range weather forecasts; more recently, the
GEMmodel began to be applied for seasonal forecasts. The GEM model
is routinely run in climate mode as avalidation tool, differences
between the model's climate and that derived from recent
re-analysis experimentsprovide useful indications of GEM's
strengths and deficiencies. The model physics configuration
closelyresembles that of the new high-resolution operational global
model to be implemented shortly.Most of the year has been spent
configuring the new climate version of GEM and using it to complete
thesecond phase of the international Streched-Grid Model
Intercomparison Project (SGMIP2). Two sets of three26-year
high-resolution climate simulations have been completed and their
monthly data has been sent to theUniversity of Maryland to be
compared with other SGMIP2 results. The first set of integrations
used aclimatological annual cycle of sea surface temperatures and
sea-ice fractions, while the second set was forcedby historically
varying monthly mean sea-surface conditions as prescribed by the
Atmospheric ModelIntercomparison Project 2 (AMIP2) experimental
protocol. Two of these six GEM SGMIP simulations wereconfigured
with a 0.5° high-resolution core area over North America and two
with a 0.5° core area overEurope. The last two simulations
contributed to the SGMIP2 effort are the corresponding 1° global
uniform-resolution controls. All of these simulations have roughly
the same large number of degrees of freedom in thehorizontal domain
and the same vertical setup. The model’s set of physical
parameterizations is the same inall simulations. The Limited-Area
Model (LAM) version of GEM is also being used to extend locally
theSGMIP2 experiment. Two 26-year GEM-LAM 0.5° horizontal grid mesh
simulations are nearingcompletion, their domains reproducing the
North-American and European high-resolution core areas of theGEM
SGMIP2 simulations for the same time period. The lateral boundary
conditions for these GEM-LAMexperiments are provided by another
1.5° global uniform-resolution GEM simulation, again retaining
thesame set of physical parameterizations.We are continuing the
comparison of the regional climate simulations produced with two
limited-areamodels, GEM-LAM and the operational version of the CRCM
of the Ouranos consortium (CRCM 4.1) and a45-year GEM-LAM
simulation driven by lateral boundary conditions from the ERA40
re-analysis data willbe started shortly. The domain will
include the North-American continent and the
boundary conditions willbe provided by reanalyses. Several
CRCM runs driven by both reanalyses and CGCM2 atmospheric datahave
been generated with various versions of the CRCM. These simulations
will serve as a referencedatabase for comparison with GEM-LAM
simulations using similar boundary conditions. This
projectrepresents one of the steps in the development of a new
system for regional climate scenarios, incollaboration with
the Climate Simulations Team of the Ouranos Consortium and with the
UQAM CentreESCER.During the first half of 2006, Dr. Kothavala has
collaborated closely with Dr. Bernard Dugas at RPN/MSC todetermine
the scalability of the GEM model to different resolutions. With the
assistance of Dr. Dugas andothers at RPN, Dr. Kothavala conducted a
global simulation of the GEM model with AMIP II specificationsat a
horizontal resolution of 1.5°. The output from this simulation was
used as driving data for a multi-decadal simulation (1978-present)
of GEM-LAM over Europe and North America. The results of
thesesimulations are currently being compared to a similar run with
the CRCM model.In the meantime, MSc student Jonathan Mainville is
evaluating the representation of tropical variabilityassociated
with ENSO SST forcing in the AMIP II integrations made with GEM for
the period 1979-2004.
-
Progress Report – CRCM Network – 2005-2006 10 / 34
Interannual variability associated with ENSO cycles explains a
large percentage of the forced atmosphericvariability over North
America. This variability is a result of anomalous planetary-wave
forcing in theatmosphere, associated with anomalous tropical
deep-convective activity in response to varying SST forcingwithin
the ENSO cycle. The AMIP II integrations utilised prescribed
time-varying SSTs and included withinthe integration period a
number of large-amplitude El Niño and La Niña occurrences. The aim
of this projectis to evaluate the response of the GEM atmosphere to
this varying SST forcing associated with the ENSOphases encompassed
by the integration period. A prerequisite for GEM to successfully
simulate ENSO forcedvariability over North America is that the
response of tropical deep convection to varying SST forcing
isaccurate. GEM has been integrated at a variety of model
resolutions for the period listed above, using theAMIP II protocol.
A suite of observations needed to evaluate simulated tropical
variability have also beencollated and archived. Initial work has
concentrated on developing suitable diagnostic techniques
forcomparing the GEM simulations with observations. These
diagnostic techniques are now being applied to theGEM-simulated
results.
Sub-project 4.1.5 "Arctic Model Intercomparison Project (ARCMIP
– GLIMPSE)"; Girard, Jones,Blanchet, McFarlane, Laprise and
Caya;
PDF Dr Ping Du, UQAM.2005 – 2006 : • Resume analysis of the
CRCM_3 and CRCM_4 GLIMPSE simulations. Identify strengths
and weaknesses of the CRCM. Perform CRCM_4.C simulation over the
ARCMIP domainand comparison with the 2 other versions of the model.
Identify the benefits (if applicable)of having prognostic cloud for
the simulation of cloud radiative effects during winter.
The CRCM_3, the operational version of the CRCM, has been used
for the intercomparison of clouds andradiation with other
participating models in ARCMIP. In its current version, the CRCM
have shown someweaknesses, particularly during winter. The boundary
layer is not well reproduced with too few mixingoccurring in the
lowest 1 km (too warm near the surface and too cold at the top of
the boundary layer). Thisbias causes important problems for the
simulation of cloud cover, particularly at the top of the
boundarylayer. The radiation fluxes at the surface are reasonably
well simulated during winter for the SHEBA point.In summer,
shortwave radiation is overestimated while longwave radiation is
underestimated. This bias islikely to be caused by the
underestimation of summertime cloud cover. The CRCM results are
relativelygood for the whole domain when compared to the ECMWF
re-analyses. However, the cloud cover is notcaptured by the model
and all ARCMIP models miss the annual cycle of cloud cover (see
Figure 4.1.5.1).Results of this analysis will be submitted shortly
for publication (Wyser et al. [4]).In order to resolve this
problem, we have implemented a new boundary-layer scheme [5] based
on thesimilarity theory, which has shown its robustness in other
models such as the ECMWF. Figure 4.1.5.2 showsthe improvement of
the wintertime and summertime boundary-layer profile of temperature
and humidity afterthe implementation of the new scheme. These
results now compare very well to the observations. With thenew
boundary-layer scheme, CRCM still agrees well with the observed
radiation fluxes on the surface.Compared to the original one, it
increases the simulated cloud cover, especially in summer, while in
winter itoverestimates the cloud cover. However, the new version
now well captures the lowest clouds both insummer and in winter. To
further improve the model simulations, we focus on the
cloud-coverparameterization. Three methods of cloud cover
parameterizations will be tested in CRCM. One is using Xu&
Randall (1996) scheme [6], in which the cloud cover is based on the
relative humidity and cloud watercontent. A second one is based on
Slingo (1987) [7], in which the cloud cover is based on relative
humiditywith a threshold RH depending on height. The last one is an
offline test to adjust the threshold of relative 4 Wyser, et al.
Evaluation of an Ensemble of eight Arctic Regional Climate Models:
cloud and radiation, manuscript.5 Troen and Mahart, Bound. Lay.
Met., 37, 129-1486 Xu,K.-M.,and D.A.Randall, 1996a A semiempirical
cloudiness parameterization for use in climate models, J. Atmos.
Sci., 53, 3084–3102.7 Slingo, J.M., 1987: The development and
verification of a cloud prediction scheme for the ECMWF model,
Q.J.R.Meteotol.Soc., 113, 899-927.
-
Progress Report – CRCM Network – 2005-2006 11 / 34
humidity and then to be used to calculate cloud cover to fit the
Arctic conditions. We are currently in theprocess of testing these
cloud-cover schemes. So far, only wintertime simulations have been
performed.Preliminary results with the Xu and Randall (1996) [8]
show no significant improvements. Other simulationshave to be
performed with other parameter settings and with a prognostic cloud
scheme to test this cloudcover parameterization.
Fig. 4.1.5.1 The monthly average value of shortwave (SW) down,
longwave (LW) down, SFC albedo (surface albedo), cloud coverand
precipitable water from all of the RCMs for ARCMIP and the
observations. The red line is for CRCM.
Fig. 4.1.5.2 Wintertime and summertime averaged temperature and
specific humidity vertical profile for the SHEBA observation(obs),
original version of the CRCM (ori) and modified version (new
boundary layer scheme) of the CRCM (ver).
8 Xu, K.-M., and D. A. Randall, 1996b :Evaluation of
statistically based cloudiness Parameterizations used in climate
models, J. Atmos.Sci., 53,3103–3119.
-
Progress Report – CRCM Network – 2005-2006 12 / 34
Sub-project 4.1.6 "Project to Intercompare Regional Climate
Simulations (PIRCS)"; Caya, Laprise andCôté.MSc student, Raphaël
Brochu, UQAM
2005 – 2006 : • Use the PIRCS experimental protocol as a
standard benchmark to validate new versions ofthe CRCM.
Within the framework of PIRCS_1c, Mr Raphaël Brochu, MSc at
UQAM, has evaluated the surface energyand water budgets of the
operational version CRCM_3.6.1 and developmental version CRCM_4
includingCLASS with re-analyses and available observations. The
conclusion of his study is that the CRCM_4constitutes an
improvement over the operational version, particularly for the
precipitation, evaporation anddiurnal screen temperature range.
Raphaël has completed his MSc in September 2005; a paper has
beensubmitted and is currently being reviewed.The newly installed
North American Climate Change Assessment Program (NARCCAP) is
taking overPIRCS for validation of RCMs over North America. Since
December 2005, NARCCAP is the maininternational initiative for
intercomparison of RCM simulations over North America. The Iowa
StateUniversity (ISU) team leading PIRCS has now taken the
responsibilities for all RCMs simulations driven byreanalyses in
NARCCAP following the protocol developed in PIRCS. The NARCCAP
domain is muchlarger than the PIRCS domain and covers almost the
complete North American continent making it muchmore appropriate
for validation of the Canadian RCM. The NARCCAP domain and the AMNO
domain arenow the basic domains for validating the CRCM. The
validation simulations on these larger domains are alsomuch longer
in time and cover the 25-year period of 1979-2003.
THEME 2. Optimisation of dynamical downscaling
Sub-project 4.2.1 "The Big-Brother Experiment"; Laprise and
Caya.MSc students, Emilia-Paula Diaconescu and Martin Leduc,
UQAM.
2004 – 2006 : • Perform simulations with various degree of LBC
imperfection and analyse the impact onthe RCM-simulated
climate.
The central objective of this study is to investigate the impact
of the lateral boundary condition (LBC) errorson the climate of a
nested Regional Climate Model (RCM). We assess if the Canadian
Regional ClimateModel (CRCM) amplifies or attenuates these errors
and if these large-scale errors affect the small scalesgenerated by
the CRCM.The methodology is based on a perfect-model framework
nick-named the “Big-Brother Experiment”designed by Denis et al.
(2002a) [9]. This method permits to evaluate the errors due to the
nesting processexcluding other model errors. First, a
high-resolution (45 km) RCM simulation is made over a large
domain.This simulation, called the Perfect Big Brother (PBB), is
driven by reanalyses from the National Centres forEnvironmental
Prediction (NCEP); it serves as reference virtual-reality climate
to which other RCM runs willbe compared. Errors of adjustable
magnitude are introduced by performing RCM simulations
withincreasingly larger domains at lower horizontal resolution (90
km); such simulations are called the ImperfectBig-Brother (IBB)
simulations and they are used, after removing small scales in order
to achieve low-resolution typical of today’s General Circulation
Models (GCM), as LBCs for smaller domain high-resolution RCM runs.
These small-domain high-resolution simulations are called Little
Brother (LB)simulations. The climate statistics of the LB are
compared to those of the PBB in order to estimate the
errorsresulting solely from nesting with imperfect LBCs, while the
difference between the climate statistics of theIBB and those of
PBB simulations mimic errors of the nesting model. 9 Denis, B., R.
Laprise, D. Caya and J. Côté, 2002: Downscaling ability of
one-way-nested regional climate models: The Big-Brother
experiment.Clim. Dyn. 18, 627-646.
-
Progress Report – CRCM Network – 2005-2006 13 / 34
MSc student Emilia Diaconescu has performed all the simulations
for five February months: from 1990 to1994. The statistical
analyses were performed over the common zone of LB simulations
excluding thesponge zone of 10 points. A spatial decomposition was
applied to separate fields into their large-scale andsmall-scale
components. A temporal decomposition of fields was also performed
to separate stationary andtransient components. The errors in each
component are indicated by the statistical coefficients displayed
intoTaylor diagram diagrams (Denis et al. 2003 [10], Taylor 2001
[11]). The results for the precipitation field aresummarized in the
above figure, which presents the Taylor diagrams for the stationary
(a and c) and transient(b and d) components of the large scales (a
and b) and for the small scales (c and d).
a) Stationary part - larges scales b) Transient part - larges
scales c) Stationary part - smalls scales d) Transient part -
smalls scales
Fig. 4.2.1. Summary Taylor diagrams showing the errors induced
in the IBB and LB precipitation rate fields, for the stationary
andtransient parts of the large- and small-scale components of the
field.
Errors are present in both stationary and transient parts for
the IBB simulations, but the transient componentsof the field
exhibit the largest errors due to rather weak temporal correlation.
The points corresponding to theLB fields are close to those
corresponding to the driving IBBs for all four components of the
fields,indicating the presence of similar errors in the
precipitation rate fields of LBs to those contained in
thecorresponding fields of IBBs. The LB reproduces a great part of
the large-scales errors of corresponding IBB.For the small scales
there is a slight improvement in regions with important orographic
forcing, andgenerally, a reproduction of most part of the
driving-model small-scale errors, even if these small scale donot
take part in the nesting process. This result suggests that, for
this particular LB domain and period, thelarge scales precondition
the small scales. Similar results were observed for the mean sea
level pressure andtemperature at 850 hPa fields (Diaconescu et al.
2005). A paper has been submitted and is now beingreviewed. In
conclusion for this study of the impact of LBC errors on the
climate of a nested RCM, thequality of lateral boundary data plays
a critical role in regional climate modelling, highlighting the
need forgood LBCs (Diaconescu 2006). Therefore it is necessary to
provide the correct large-scale circulation at thelateral boundary
of RCM in order to obtain the correct small scales too.Furthermore,
RCMs are increasingly used to add small-scale features that are not
present in their LBCs. It iswell known that the limited area over
which a model integrates must be large enough to allow the
fulldevelopment of small-scale features (Jones et al., 1995[12]).
On the other hand, integrations on very largedomains have shown
important departures from the driving data, unless nudging of the
large scales is applied(e.g., Castro and Pielke, 2005[13]). MSc
student Martin Leduc is assessing the effects of domain size on
thedevelopment of small scales using the "Big-Brother" approach.
Similarly to the previous study, a referenceclimate is established
by performing a high-resolution simulation over a large domain (the
Big Brother). The 10 Denis, B., R. Laprise and D. Caya, 2003:
Sensitivity of a Regional Climate Model to the spatial resolution
and temporal updating frequency ofthe lateral boundary conditions.
Clim. Dyn., 20, 107-126.11 Taylor K.E., 2001: Summarizing multiple
aspects of model performance in a single diagram. J. Geophys. Res.
106: 7183-7192.12 Jones, R. G., J. M. Murphy, and M. Noguer, 1995:
Simulation of climate change over Europe using a nested
regional-climate model. I:Assessment of control climate, including
sensitivity to location of lateral boundaries. Quart. J. Roy.
Meteorol. Soc., 121, 1413-1449.13 Castro, C. L. and R. A. Pielke,
2004: Dynamical Downscaling: Assessment of Value Retained and Added
Using the Regional AtmosphericModeling System (RAMS). J. Geophys.
Res., 110, 1-21.
-
Progress Report – CRCM Network – 2005-2006 14 / 34
next step is to degrade this dataset with a low-pass filter
based on discrete cosine transform (DCT; Denis etal., 2002b [14])
to emulate coarse-resolution LBCs that are usually taken from GCMs
or reanalyses. A secondsimulation (the Little Brother) is driven by
the coarse-resolution LBCs and generates its own
small-scalefeatures inside the new smaller domain. Driven and added
scales of the Little Brother can then be comparedwith the
Big-Brother (unfiltered) ones by using the DCT-filter again. Three
February months (1990, 1991and 1992) were integrated over a
continental grid (Big Brother: 196x196 gridpoints) with grid mesh
of 45 kmcovering almost the entire North-America. After filtering,
this dataset is used to drive five simulations withvarying domain
size (48x48, 72x72, 96x96, 120x120 and 144x144) centred on the same
geographic location;all other parameters are kept constant. Monthly
statistics of the five Little Brothers are compared with thevirtual
reference (Big Brother) over the common domain (28x28)
corresponding to the smallest Little Brotherbut without its sponge
zone (Fig. 4.2.1.2). Results show that temporal correlation of
large-scale eventsincreases when the domain size is reduced from
144x144 to 48x48. For the same domain change, thecorrelation
improves in small-scale features, suggesting that their consistence
is in some way helped by theincreased correlation of the
large-scale flow. A second effect of the domain size on small
scales is based onthe fact that these scales need a spin-up time to
fully develop from the low-resolution lateral boundaryconditions.
Variance-ratio maps show that the relative intensity of
small-scales tend to grow from theirentrance inside the domain
until they reach the full amplitude of the Big-Brother variance. A
spin-up zone isobserved for small-scales of all examined fields
(e.g. geopotential, temperature, relative humidity and
relativevorticity). The spin-up area grows in size at higher levels
where winds are stronger, suggesting that small-scale features are
advected out of the domain area before they have time to fully
develop.
Fig. 4.2.1.2 The percentage of the Big-Brother's variance in
small-scales for the relative vorticity field at 700-hPa is given
for thefive Little-Brother simulations which have domain sizes of
a) 144x144, b) 120x120, c) 96x96, d) 72x72 and e) 48x48,all
displayed through a window of 28x28 gridpoints.
14 Denis, B., J. Côté and R. Laprise, 2002: Spectral
decomposition of two-dimensional atmospheric fields on limited-area
domains using discretecosine transforms (DFT). Mon. Wea. Rev. 130
(7),1812-1829.
-
Progress Report – CRCM Network – 2005-2006 15 / 34
Sub-project 4.2.2 " Influence of surface forcing and large-scale
nudging on RCM internal variability ";Caya, Laprise and de Elía;PhD
student, Philippe Lucas-Picher, UQÀMNon-Network MSc student,
Jean-Philippe Paquin, UQAM.
2005 – 2006 : • Study the forcing from the CLASS multi-layer
land-surface scheme and investigate theinfluence of its long memory
(order of years) on the relatively short (order of decades)regional
climate simulations; Perform a scale decomposition of the
variability andinvestigate the influence of the land-surface
boundary on the development of fine-scaledetails in RCM
simulations.
Unlike global climate models (GCM), regional climate models
(RCM) simulations require to be driven at thelateral boundaries of
their limited area domain by continuous atmospheric information.
This nesting imposesan additional forcing on the RCM, which reduces
the internal variability. The intensity of this forcing isfunction
of the flow regime, domain size and season. Despite this forcing,
the RCM exhibits a certain level ofinternal variability (IV).
Ensemble of simulations with the Canadian RCM started with
different initialconditions for various domain sizes were realized
by PhD student Philippe Lucas-Picher. Statistical analysisof the
simulations shows that lateral boundary forcing is less effective
as the domain expanded, increasingthe internal variability. An
attempt was made to investigate the cause of this dependence. With
this objective,an ageing tracer was implemented in the Canadian RCM
to measure the residency time of the atmosphericparcels into the
limited-area domain. This tool can serve to determine the
flow-regime properties and toevaluate the degree of constraint
exerted by the lateral boundary information on the RCM simulation
(Fig.4.2.2). A quasi-linear relation between the residency time and
the RCM’s internal variability was found.Surface temperature and
mean sea level pressure internal variability increase linearly with
the residency time.It was also found that the internal variability
increases more in winter than in summer with the residencytime.
This idea is opposite to general perception.
Fig. 4.2.2 Ten-year mean atmospheric circulation for summer
season (JJA) at 500 hPa. The colour scale represents the
timeresidency in days while the arrows indicate the wind speed in
m/s.
In the meantime, Jean-Philippe Paquin, M.Sc. student at UQAM, is
performing simulations on a hemisphericdomain with the CRCM at a
lower resolution (180 km instead of the usual 45 km) in order to
test theimprovement of using an additional step in the nesting
within the driving data with an intermediate-resolutionversion of
the model. The idea is to test the approach developed at the Hadley
Center in UK over NorthAmerica. Simulations carried at the Hadley
Center over Europe have shown that the use of the
intermediate-resolution helps in representing the large-scale
circulation in the high-resolution model. Some problems in
-
Progress Report – CRCM Network – 2005-2006 16 / 34
the development of the intermediate version of the CRCM have
introduced some delays in Jean-Philippe'sproject. He has now
resolved most of the problems and has resumed his simulations.
Sub-project 4.2.3 " Internal variability of RCM in ensemble
simulations "; De Elía, Caya and Laprise;M.Sc. students Adelina
Alexandru, Leo Separovic, UQAM. Both students continued the
researchstarted by PDF Ramón de Elía (hired by Ouranos Consortium
in August 2004).
2005 – 2006 : •Comparison of internal variability of both the
GCM and the RCM over a season.Characterize the internal variability
for different weather regimes.
Regional climate models do not behave like a “magnifying glass”
when used to increase fine details fromlow-resolution fields. On
the contrary, they do show some freedom (known as internal
variability), whichforces the user to carefully assess the validity
of the obtained high-resolution features.Adelina Alexandru studied
the internal variability (IV) in RCMs by means of an ensemble
approach over theNorth American region. Several twenty-member
ensemble simulations over different domains weregenerated, allowing
for a detailed study of the spatial and temporal variation of IV
with domain size. Whilepreviously it was believed that IV grows
with domain size, it was shown that the increase in IV is
notmonotonic with domain size. In addition, it was shown that the
geographical distribution of the IV canchange substantially with
domain size, and that areas of large IV in a small domain can lie
within areas ofsmall IV in a larger domain.A detailed investigation
of the temporal evolution of the IV in different domain sizes
showed that, althoughsmall domains tend to develop in average less
IV, they do have isolated episodes of strong IV. These
isolatedepisodes leave a trace in the IV seasonal average,
suggesting that even for small integration domains theensemble
approach may be necessary. Adelina has been submitted her Master
thesis at the end of June 2006;she is now working on an article
focussing on internal variability in regional climate downscaling
at theseasonal scale.
Fig. 4.2.3.1 Internal variability experiment, Ensemble of 20
simulations with a domain of 120 x 120: Seasonal ensemble spread
forthe precipitation (mm/day) (left-side panel) and for the 850-hPa
geopotential (m) (right-side panel).MSc student Leo Separovic took
advantage of the large database generated by A. Alexandru and
investigatedthe amount of information (in the spectral space)
contained in the ensemble average and the departures fromthe
average. This has been done on an instantaneous basis (at each
archived timestep) as well as on theseasonal average.Results show
that some of the small scales added by the dynamical downscaling
(scales absent in the drivingfields) are present in the ensemble
average, while others exist only in the deviations from the
ensemble (Fig.4.2.3.2). The dominant term varies with weather
pattern: for periods of intense IV almost no small-scaleinformation
remains in the ensemble average, while periods of low IV show
significant development of smallscales in the ensemble everage.
These results show that in certain cases the low-resolution
boundary
-
Progress Report – CRCM Network – 2005-2006 17 / 34
conditions that drive the RCM predetermine the small scales
inside the integration domain, while in certaincases (high IV), the
regional model’s generation of small scales is less dependent on
boundary conditions.
Fig. 4.2.3.2. Spectrally decomposed spatial variance of
instantaneous 925-hPa geopotential generated by the ensemble of
CRCMruns. The variance is sampled every 6 hours, and averaged
during June, July, and August 1993. Dashed line represents the
variancebetween members within the ensemble average, the dotted
line represents the variance within the deviations from the
ensembleaverage, and full line is the sum of the two previous ones.
The red line represents the spectral variance of regridded
NCEPreanalyses used to define the lateral boundary conditions. The
intersection of the dashed and the doted curve at wavenumber
20(~270 km) indicates the spatial scale at which the dynamical
downscaling process generates information dominantly in
stochasticform.
THEME 3. Development and improvements in the CRCM
Sub-project 4.3.1 "Regional ocean coupling"; Saucier, Caya,
Laprise and Boer;RA, Simon Senneville, UQAR, Dr Minwei Qian, UQAM;
MSc Student, Marko Markovic,UQAM.Non-CRCM Network PhD student, Marc
Defossey.
2005 – 2006 : • Evaluate performance of coupled system with
available observations;• Investigate the effect of coupling over
the Hudson Bay on the simulated climate and in
particular on the water cycle, and evaluate with available
observations.One of the primary atmospheric inputs controlling the
quality of simulated regional ocean model surfacetemperatures and
ice coverage is the incoming solar and terrestrial radiation at the
ocean surface. In order toincrease confidence in the quality of the
simulated surface radiation budget in Regional Climate Models
usedby the CRCM Network, MSc student Marko Markovic has performed
an in-depth evaluation of the surfaceradiation simulated by3 RCMs
integrated over North America, using analysed lateral boundary
conditions forthe recent observed past.NOAA US-Surface Radiation
Network observations have been used to evaluate the models surface
radiationbudget across a variety of climatic regimes over North
America and throughout the seasonal cycle. Threemodels have been
evaluated for the period 1999-2004 are:1. The latest version of the
CRCM model, version 4.1, including the most recent physical
parameterisationpackage under development collaboratively between
the CRCM Network and Ouranos scientists.2. The climate version of
GEM-LAM. This is the RCM targeted to be the next default Canadian
RegionalClimate Model and is presently being extensively evaluated
as a Regional Climate Model.
-
Progress Report – CRCM Network – 2005-2006 18 / 34
3. The Rossby Centre Regional Atmosphere Model version 3 (RCA3).
In forthcoming CRCMD Network [15],it is intended to introduce the
RCA3 physics package, along with the CRCM version 4.1 physics into
theGEM-LAM dynamical core. From this perspective it is worthwhile
to include this model in our evaluation ofthe surface radiation
budget.The US-Surface radiation Network offers high temporal
frequency (hourly) observations of total-skyshortwave and longwave
downwelling radiation, along with fractional cloud coverage at 6
sites across NorthAmerica. For the simulated period 1999-2004, RCM
grid box values of these variables have been extractedand compared
to the surface observations. Total-sky and clear-sky downwelling
radiation have been analysedalong with the simulated cloud cover.
From this analysis the annual cycle of surface cloud-radiative
forcingcan be constructed and compared to observed values.
Systematic biases in aspects of the cloud and radiationinteraction
in the 3 RCMs have been identified and updated models are in the
process of being evaluated.
Figure 4.3.1.1Mean annual cycle of downwelling solar and
terrestrial radiation at all US Surface Radiation Network sites and
assimulated by the 3 respective RCMs for the period
1999-2004.Figure 4.3.1.1 shows the observed and simulated mean
annual cycle of downwelling solar and terrestrialradiation averaged
across the 6 surface observation stations, along with the annual
cycle of RCM errors inthese two quantities. Parallel analysis has
enabled a comparison of the simulated surface
cloud-radiativeforcing and assisted in identifying which aspects of
the cloud-radiation or clear-sky radiation are responsiblefor the
total surface radiation errors.The high temporal frequency of the
observations also allows an evaluation of the simulated mean
diurnalcycle of surface radiation and cloud cover. Climate models
frequently have problems simulating the meandiurnal cycle of
convection and associated cloud fields during the summer season
often resulting in large
15 Canadian Regional Climate Modelling and Diagnosis Network;
CFCAS has approved the research Network initiative lead by Colin
Jones,including 16 co-Is from Universities, Ouranos and
Meteorological service of Canada (MSC / CCCma and RPN) for the
period 2006-2010.
-
Progress Report – CRCM Network – 2005-2006 19 / 34
errors in the seasonal mean surface radiation budget and a
gradual drying out of soil moisture. Figure 4.3.1.2shows the mean
diurnal cycle of surface radiation for the summer season
(June-July-August) averaged acrossthe 6 observation sites and the
years 1999-2004. Results are shown separately for all-sky and
clear-skyconditions, along with the derived diurnal cycle of
cloud-radiative forcing. This type of analysis assists
inunderstanding the source of errors in the seasonal mean surface
radiation budget simulated by RCMs.
Fig. 4.3.1.2 Mean diurnal cycle of JJA surface solar radiation
for all sky and clear-sky conditions along with the mean diurnal
cycleof cloud-radiative forcing from observations and the 3 RCMs.
Values are averaged over the 6 Observation sites and for the
years1999-2004In the meantime, Mr Marc Defossez, Ph.D. student, has
investigated the deep waters formation andcirculation in the Hudson
Bay system during his first year at the University of Québec at
Rimouski. He hasfocused on the deep waters renewal in Foxe Basin
(Defossez et al., 2005) using Saucier's et al. (2004 [16])
seaice-ocean model and year-long time series observations retrieved
from a mooring at the bottom of FoxeChannel in 2004. He now
examines the dense waters formation mechanisms and their dependance
toatmospheric changes by sensitivity experiments. This research is
done with the collaboration of CLIVARNetwork scientists Saucier,
Myers and Caya.Dr. Minwei Qian continued his work on the coupled
model used to perform a new simulation from 1st Aug.1991 to 31st
Jan. 1998. This simulation was used to demonstrate the effect of
the NAO index on inter-annualvariation in ice freeze-up and and
break-up period. NAO is the most important mode of
atmosphericvariability over the North Atlantic Ocean, and plays a
major role in sea-ice formation, sea surfacetemperature (SST) and
surface current over the Hudson Bay. However, NAO is not the only
mode. TheSouthern Oscillation (SO) is another important mode that
has impact on Hudson Bay. During the period from1992 to 1996, there
are neither strong El Niño nor strong La Niña events, while 1992
and 1993 are strongpositive-NAO years and 1995 and 1996 are strong
negative-NAO years. By make a comparison between thesimulations in
positive- and negative-NAO episodes, the variations of sea-ice
cover, SST and surface currentare obtained.The results shows:1. The
coupled model captures the interannual variation of sea-ice cover.
In the sea-ice freeze-up period from
21st November to 10th December, the sea-ice cover in
positive-NAO episode is 31% more than that innegative-NAO
episode.
2. The interannual variation of SST in October is well
simulated. The SST in positive-NAO episode is 1.3°Clower than that
in negative-NAO episode.
3. The average cyclonic surface current in
October-November-December season is 12 to 14 cm/s (Fig.4.3.1.2 in
left panel). However, the surface current in positive-NAO episode
is 6 to 10 cm/s faster than that innegative-NAO episode in
southwestern and eastern coasts of the Hudson Bay (Fig. 4.3.1.2,
right panel).
16 Saucier, F.J., Senneville, S., Prinsenberg, S., Roy, F.,
Smith, G., Gachon, P., Caya, D., and Laprise, R. 2004. Modelling
the sea ice-oceanseasonal cycle in Hudson Bay, Foxe Basin and
Hudson Strait, Canada. Clim. Dyn., 23: 303-326.
-
Progress Report – CRCM Network – 2005-2006 20 / 34
Fig. 4.3.1.2 Average sea surface current (cm/s) in Hudson Bay
(left panel) and Sea surface current difference between
positive-andnegative-NAO episodes (right panel).
Sub-project 4.3.2 " River-routing and surface water processes ";
Caya, Slivitzky, Larocque, Laprise andSaucier;MSc, Ivana Popadic,
RA, Dr Laxmi Sushama, UQAM (hired by Ouranos Consortium in
January2006).
2004 – 2006 : • Continue efforts on more fundamental issues to
identify potential candidates for improvedmodels of land surface,
river, lake and permafrost for adaptation to the CRCM.
The sensitivity of the Canadian Regional Climate Model (CRCM)
projected changes to the climatologicalmeans and extremes of
selected basin-scale surface fields to model errors, for six basins
covering the majorclimate regions in North America, were studied
using current and future (A2 and IS92a scenarios)
climatesimulations performed with two versions of CRCM (Sushama et
al., 2006a). Climate change is commonlyevaluated as differences
between simulated climates under future and current forcing, based
on theassumption that systematic errors in the current climate
simulation do not affect the climate-change signal.Assessment of
errors in two CRCM versions suggests the presence of non-negligible
biases in the surfacefields, due primarily to the internal dynamics
and physics of the regional model and to the errors in thedriving
data at the boundaries. In general, results demonstrate that, in
spite of the errors in the two modelversions, the simulated
climate-change signals associated with the long-term monthly
climatology of varioussurface water balance components (such as
precipitation, evaporation, snow water equivalent, runoff and
soilmoisture) are consistent in sign, but differ in magnitude (Fig.
4.3.2.1). The same is found for projectedchanges to the low-flow
characteristics such as frequency, timing and return levels.
High-flowcharacteristics, particularly the seasonal distribution
and return levels, appear to be more sensitive to modelversion. It
should be noted that the precipitation and runoff biases in the
model were partly due to its simpleland-surface scheme. Study
suggests the need to have a more physically based land-surface
scheme toimprove the near surface processes which could improve the
high flows along with other fields.
-
Progress Report – CRCM Network – 2005-2006 21 / 34
Fig. 4.3.2.1 Relative magnitude of the 10-year 7-day low flows
for the six basins, with 95% confidence intervals for
currentCRCM-CGCM2-s (black), CRCM-CGCM2-u (green) climates and
future CRCM-A2-s (blue), CRCM-IS92a-s (red) and CRCM-A2-u (purple)
climates.In parallel with the above, an evaluation of changes in
the soil thermal regime for Northeastern Canada wasperformed using
a one-dimensional heat conduction model (Popadic, 2006; Sushama et
al., 2006b). Projectedchanges were estimated as the difference
between two simulations of the soil model corresponding to theIPCC
IS92a future scenario (2041–2070), which has effective CO2
concentration increasing at 1% per year(2041–2070), and current
(1961–1990) climates. The surface temperature and snow cover from
time series oftransient climate simulations with the CRCM were used
to drive the soil model.
ALT (1961-1990) ALT (2041-2070)
Fig.4.3.2.2 Simulated distributions of average ALT for SM_CRCM
permafrost zones. The average ALT is at an interval of 2 m.
-
Progress Report – CRCM Network – 2005-2006 22 / 34
Results suggest significant warming trends in the annual mean,
maximal and minimal near-surface soiltemperatures, with the mean
annual near-surface soil temperature increasing by 4°C for the
continuouspermafrost zone and by 2–3°C for the rest of the
permafrost zones in Northeastern Canada (Fig. 4.3.2.2).Results also
suggest significant deepening of the active layer for the period
2041–2070, with its thickness(ALT) increasing by 40 to 80% for the
continuous permafrost region.
Sub-project 4.3.3 " Physical parameterisation including
land-surface processes "; Jones, Laprise, Cayaand McFarlane;RA, Dr
Yanjun Jiao, UQAM; MSc student, Cynthia Papon, UQAM.
2005 – 2006: • Perform and analyse regional climate model
experiments to assess and improve therepresentation of convection
and convectively forced clouds in the CRCM and GEMmodels.
•Repeat the CRCM integrations with higher model resolution.
Analyse the representation ofconvection/clouds and compare to the
low-resolution models. Identify areas requiringimprovement for the
representation of convection at higher resolution.
The present version of the CRCM, version 4.1, has been
extensively improved in order to accuratelyrepresent stratocumulus
clouds, shallow cumulus and deep convection, and the phase
transition betweenthese regimes. This work has been performed by RA
Dr Yanjun Jiao within the GEWEX Pacific Cross-section
Intercomparison project (GPCI), an international project to
evaluate and improve the representationof cloud and precipitation
processes in weather prediction and climate models. In the GPCI,
different modelsand observations are analyzed and compared along a
cross-section over the Pacific Ocean from thestratocumulus regions
off the coast of California, across the shallow cumulus areas, to
the deep convectionregions of the ITCZ.According to the requirement
of the GPCI, the CRCM_4 was run on a 180-km grid mesh over the
PacificOcean with 115x75 grid points in the polar-stereographic
projection (Fig.4.3.3.1). Instantaneous modelresults are output
every 3 hours for the periods of June-July-August 1998 and 2003.To
get better results, based on the preliminary results produced by
the original CRCM_4, a number ofmodifications have been introduced
in the Bechtold-Kain-Fritsch convective scheme, vertical diffusion
in theboundary layer and cloud parameterization schemes of the
CRCM_4.
Fig. 4.3.3.1. Computational domain (outlined by the thickblue
line) of the CRCM_4 over the Pacific Ocean. Thedotted line
indicates the location of the cross-sectionproposed by the GPCI and
the solid white line represents thecommon area required by the
GPCI.
-
Progress Report – CRCM Network – 2005-2006 23 / 34
The modifications are briefly summarized as:1. Shallow
convection is turned off once the deepconvection had been detected
at a given grid point.2. A temperature perturbation based on the
relativehumidity in the mixing layer has been added in the
triggerfunction of shallow convection.3. The free convective
vertical velocity scale has beenused in the cloud-base mass flux
closure of the shallowconvection.4. Cloud radius of the deep
convection, which controls themaximum possible entrainment rate,
has been specified tovary as a function of vertical velocity at
liftingcondensation level.5. The minimum cloud-depth threshold has
beenparameterized according to the cloud-base temperaturerather
than remaining constant.6. A dilute updraft ascent has been used to
calculateconvective available potential energy (CAPE),
whichprovides a more accurate calculation in convection rainfalland
mass flux.7. The turbulent diffusion scheme of the ECMWF hasbeen
used to calculate momentum, heat and moisturetransfers in boundary
layer.8. A cloud scheme that considers the cloud liquid water
incalculating cloud amount, has been introduced.9. The evaporation
of the larger scale precipitation hasalso been considered.
Fig. 4.3.3.2. June-July-August averaged cross-sections of
relative humidity (a) simulated by theoriginal CRCM4, (b) simulated
by the modifiedCRCM4, and (c) the ECMWF analysis used as
areference.
These modifications have a significant beneficial influence on
the CRCM_4 simulation such as the verticalstructure of the boundary
layer (Fig. 4.3.3.2), cloud amount and distribution, the location
and amount of theconvective and large-scale precipitation. The
results have been submitted to the GPCI working group.One aspect of
this development has involved the introduction of a more physically
based parameterisation ofcloud amounts associated with shallow and
deep convection. This has been the work of MSc student
CynthiaPapon.In the original version of CRCM_4.1 cloud amounts
associated with shallow and deep convection wereassumed to be a
constant value when and where either convective type occurred.
There was no formalparameterisation of the cloud amount which
linked cloud fraction to the intensity of convection,
convectivewater content or the environmental conditions within
which the convection occurred. In an attempt toincrease the
physical realism of the cloud amounts associated with shallow and
deep convection, two newparameterisations of these cloud types have
been implemented into the CRCM model. These new approachesformally
link the amount of convective cloud to the intensity of convection,
the amount of convectivelydetrained cloud and ice water and are
influenced by the thermodynamic conditions within which the
(a)
(b)
(c)
-
Progress Report – CRCM Network – 2005-2006 24 / 34
convection is embedded. The parameterisation schemes were
initially evaluated in a constrained single-column setting and are
now presently being evaluated in full 3D CRCM integrations
following the GEWEXGPCI experiment protocol. Figure 4.3.3.3 shows
some initial results from the updated CRCM (in black)along with
observations (in green). Results have been extracted for a
cross-section through the Pacificstratocumulus, shallow cumulus and
deep convection regimes extending from the coast of California to
thewest Pacific warm-pool region. Figure 4.3.3.3 shows the
simulated and observed seasonal mean integratedliquid water path
(LWP), total cloud cover and precipitation along this cross-section
for the period June-July-August 1998. The updated CRCM has a
relatively accurate simulation of the various cloud types, only a
clearunderestimate of LWP in the stratocumulus region is evident.
This problem is presently under investigation.
Fig. 4.3.3.3. Parameterisation of cloud amounts associated with
shallow and deep convection: CRCM (in black) versusobservations (in
green).This updated CRCM version is being evaluated over North
America for present climate conditions to assessthe improvement in
the representation of clouds and surface radiation as a result of
the updated shallow anddeep convective cloud parameterisation.
-
Progress Report – CRCM Network – 2005-2006 25 / 34
Sub-project 4.3.4 " Computer code parallelisation "; Zadra,
Jones, Côté, Laprise, Zwiers and Caya;RA, Dr Ron McTaggart-Cowan,
UQAM.
2004 – 2006 : • Develop a distributed-memory parallelised
version of CRCM, CRCM_5, based on thedynamical kernel of the
Canadian operational forecast model GEM in a limited-areaversion
called GEM-LAM.
One of the primary goals of this sub-project is to develop the
GEM model for implementation as the primaryCanadian regional
climate model. The limitations involved with using the Environment
Canada IBMsupercomputer at the Canadian Meteorological Centre in
Dorval are significant, given the heavy load on themachine that
leads to queued wait times ranging between 6 and 24 hours for
typical regional climate modelgrid sizes. We have therefore made
considerable efforts to ensure that the GEM model will run
effectively ona variety of platforms, including those currently in
use at Ouranos.As a result of this project, the GEM model and its
supporting libraries have now been ported to architecturesincluding
64-bit Linux, SUPER-UX, and Catamount. As part of a CFI [17]
acquisition process, a series ofbenchmarks has been run on these
platforms in order to obtain accurate measures of performance and
scaling.One of the primary findings of this work is that the
performance of the GEM model on a distributed memorycluster with
fast interconnects exceeds the expectations of the model
developers. This is important forupcoming supercomputer
acquisitions since commodity-based cluster solutions tend to be
more cost effectivethan their “fat node” or vector
counterparts.Another contribution of this project has been the
vectorization of the GEM model. Although the code wasoriginally
written to take advantage of the vector machines at the Canadian
Meteorological Centre in the late1990’s, it has since been
optimized for use on cache-based architectures. The vector
operation ratio of thecode was increased from 78% to >98% and
the wall-clock run time of the benchmark simulation reduced bya
factor of 3. These significant performance enhancements, to be
implemented and fully supported in the nextrelease of the GEM
model, will allow the regional climate version of the GEM to
effectively leverage thevector-based computing platforms (NEC SX-6
running a SUPER-UX operating system) available atOuranos. This will
make the conversion from the existing non-parallel regional model
to the GEM climateversion possible without the need for
restructured computing facilities.
Fig. 4.3.4 GEM RCM climate scaling: Almost linear to 200 PEs18
offering fast integrations speeds on parallel systems.
17 Canadian Foundation Innovation18 PE Processor Equivalent
-
Progress Report – CRCM Network – 2005-2006 26 / 34
As a part of the GEM climate model verification project (again,
in preparation for its implementation as theprimary Canadian
regional climate model), a regional climatology for southern North
America and CentralAmerica is being produced. This region is of
particular interest because it encompasses the western
NorthAtlantic, Gulf of Mexico and Eastern Pacific hurricane genesis
regions, as well as containing the intertropicalconvergence zone
during the summer months. Accurate simulation of each of these
features is known to beproblematic in global models, and it is
hoped that this next-generation forecasting model, properly
adaptedfor extended climate simulation, will be more successful in
resolving the processes and forcings that areunique to this region.
While the development of this climatology continues, a series of
collaborativediagnostic studies are being performed in order to
assess tropical cyclone development in this region, and itsimpacts
on the global circulation.
2. ImpactThe Network includes Universities, Ouranos, IML and MSC
(CCCma and RPN) scientists working todevelop and use advanced
modelling tools in a complementry fashion. Together these
researchers aim atdeveloping, validating and improving a complete
regional-scale climate modelling and diagnostics system,coupling
the atmosphere, coastal oceans, lakes, ice and land surfaces. This
initiative represents the Canadiancontribution to the international
efforts in regional-scale climate modelling and
analysis.Governmental institutions are the logical end-users of the
science results produced by the Network. Theparticipation of
scientists of Ouranos, CCCma, RPN and IML to the Network ensures
continuity of thecommunication channels, and these co-applicants
are part of the technology transfer mechanism. The CRCMNetwork
continues, as in the past, to maintain contact with the Impact and
Adaptation (IA) community,through participating in and organising
users' workshops. Facing the management of consequences ofextreme
climate events, the policy makers show a growing interest for the
climate change at regional scaleand tools such as the CRCM, in
order to be prepared to climate-change consequences. The IA
communityrequires the highest possible resolution projections in
order to establish detailed climate-change scenarios forthe
Canadian community. Scientific publications and the participation
in Canadian and internationalconferences and workshops are the
principal external communication strategy.As an important client
for the R&D of the CRCM Network, Ouranos also relieves the
Network from climate-change simulations via its Climate Simulations
Team (CST) under the leadership of Dr D. Caya, Co-I ofCRCM Network.
The CST has the national mandate, in collaboration with MSC, of the
production ofCanada-wide regional climate-change projections. The
collaboration with Ouranos is consistent with theacademic research,
training and technology transfer mandate of the Network. The
products of the Network’sresearch is intended to give operational
sectors, governmental services and Canadian companies access to
theprivileged knowledge, expertise and access to highly qualified
personnel, thus offering opportunities foreconomic benefits to
Canada.Another important focus of the Network is the training of
HQP in the field of regional climate modelling.The expertise of
this CRCM Network is unique in Canada, and it constitutes an ideal
nest for developing theneeded HQP. As a consequence, HPQ trained
through this Network are currently active in the Canadian
andinterational regional climate regional modelling field.Despite
the limited Canadian resources, the global and regional modelling
researchers involved in the CRCMNetwork at UQAM and MSC are widely
recognised as important players in the field. This is reflected
bysubstantial contributions through publication of peer-reviewed
papers and participation in internationalprojects through Model
Intercomparison Projects (MIPs) and World Climate Research
Programme (WCRP)(see section 4 ‘Contributions’). In particular, the
involment of several members as Contributing or LeadAuthors in the
forthcoming IPCC (Inter-governmental Panel on Climate Change)
Fourth Assessment Report
-
Progress Report – CRCM Network – 2005-2006 27 / 34
(AR4) helps ensure to maintain the CRCM Network's research at
the forefront of science, and enhance thecurrent high profile of
Canadian research in this area.
3. Level of supportThe proportion of the total CRCM Network
budget provided by CFCAS was around 68% during thereporting period.
The Finance Department of UQÀM has issued a financial statement of
the CRCM activitiesfinanced by CFCAS.The CLIVAR Network (funded by
NSERC and CFCAS), the US Department of Energy (US DoE), and
theQuébec Government, through the Global Environmental and
Climate-Change Centre (GEC3, funded byFQRNT) and the Consortium
Ouranos, « Consortium sur le climat régional et l’adaptation
aux changementsclimatiques » have provided the funds required to
complete the budget of the CRCM Network for the periodcovered by
this report.Since September 2002, the Consortium Ouranos provides
space and facilities to house a part of the CRCMNetwork at UQÀM at
its research laboratories, at 550 Sherbrooke West, 19th Floor, West
Tower, Montréal(Québec). The value of this in-kind contribution to
the CRCM Network amounts to about 200 K$ per year.Substantial
in-kind contributions are also provided by Environment Canada, in
terms of significantcomputational facilities and scientists
contributing to the network objectives.The blend between the
research focus of the Universtities, the model development focus of
MSC and theapplications focus of Ouranos provided the
infrastructure necessary to the CRCM Network. Specifically
thiscomputing infractructure located at Ouranos includes a CLUMEQ
CFI-funded system composed of a 32-CPU Origin 3000 SGI
computer, a DMF automatic archival system, RAID mass storage and a
UPS unit. Thislocal equipment complements the Ouranos and MSC
supercomputer for CRCM computing.
4. Dissemination
(1) Peer Rewieved Publications (published, in press or accepted
for publication)Beaulne, A, H. Barker and J.P. Blanchet, 2005:
Estimating Cloud Optical Depth from Surface Radiometric
Observations:
Sensitivity to Instrument Noise and Aerosol Contamination, J.
Atmosph. Sc, 62 (11): 4095–4104. Bielli, S., and R. Laprise
2006: A methodology for regional scale-decomposed atmospheric water
budget: Application to a
simulation of the Canadian RCM nested by NCEP reanalyses over
North America. Mon. Wea. Rev., 134: 854-873.de Elía, R. and R.
Laprise, 2005: Probabilities, probabilities, and probabilities.
Essay in Bull. Amer. Meteo. Soc. 86, 1224-1225.Desgagné, M., R.
McTaggart-Cowan, W. Ohfuchi, G. Brunet, M. K. Yau, J. Gyakum, Y.
Furukawa and M. Valin, 2006: Large
atmospheric computation on the Earth Simulator: the LACES
project. Scientific Computing (in press).Dimitrijevic, M., and R.
Laprise, 2005: Validation of the nesting technique in a RCM and
sensitivity tests to the resolution of the
lateral boundary conditions during summer. Clim. Dyn. 25,
555-580.Fox-Rabinovitz, M.S., J. Côté, B. Dugas, M. Deque and J.
McGregor, 2006: Variable-Resolution GCMs: Stretched-Grid Model
Intercomparison Project (SGMIP). Journal of Geophysical Research
– Atmospheres (in press).Girard., É., and B. Bekcic, 2005.
Sensitivity of an Artic Regional Climate Model to the horizontal
resolution during winter:
implications for aerosol simulation. Internat. J. Climat. 25
(11):1455-1473.Girard, E., Blanchet, J.-P., and Y., Dubois. 2004.
Effects of arctic sulphuric acid aerosols on wintertime low-level
atmospheric ice
cyrstals, humidity, and temperature at Alert, Nunavut. Atmos.
Res., 73 (1-2): 131-149.Hu, R.M., J.-P.Blanchet, and E.Girard,
2005a: Evaluation of the Direct and Indirect Radiative and Climate
Effects of Aerosols over
the Western Arctic, J. Geophys. Res., 110 (11) (DOI
10.1029/2004JD005043). Hu, R.M., J.-P.Blanchet, and E.Girard,
2005b: Aerosol Effect on Surface Cloud Radiative Forcing in the
Arctic. Atmos. Chem.
Phys., 5, 9039-9063, 2005.
-
Progress Report – CRCM Network – 2005-2006 28 / 34
Jiao, Y., and D. Caya, 2006 : An investigation of summer
precipitation simulated by the Canadian Regional Climate Model.
Mon.Wea. Rev. 134 (3), 919–932.
Le Fouest, V., B. Zakardjian, F.J. Saucier and M. Starr,
2005 : Seasonal versus synoptic variability in planktonic
production in ahigh-latitude marginal sea: the Gulf of St. Lawrence
(Canada). J. Geophys. Res. 110 (9) (DOI 10.1029/2004JC002423).
McTaggart-Cowan, R., L. F. Bosart, J. R. Gyakum and E. Atallah,
2006: Hurricane Juan (2003). Part II: forecasting and
numericalsimulation. Mon. Wea. Rev. (in press).
Plummer, D., D. Caya, H. Côté, A. Frigon, S. Biner, M. Giguère,
D. Paquin, R. Harvey and R. De Elía 2006: Climate and climatechange
over North America as simulated by the Canadian Regional Climate
Model. J. Climate (in press).
Rinke, A., K. Dethloff, J. Cassano, J. H. Christensen, J. A.
Curry, P. Du, E. Girard, J.-E. Haugen, D. Jacob, C. G. Jones,
M.Køltzow, R. Laprise, A. H. Lynch, S. Pfeifer, M. C. Serreze, M.
J. Shaw, M. Tjernström, K. Wyser and M. Zagar, 2006.Evaluation of
an ensemble of Arctic regional climate models: spatiotemporal
fields during the SHEBA year. Clim. Dyn., 26:459-472.
Smith G, F. J. Saucier, and D. Straub, 2005. Circulation in the
St. Lawrence Estuary in wintertime. J. Phys. Oceanogr.
(accepted)Sushama, L., R. Laprise, D. Caya, A. Frigon and M.
Slivitzky, 2006a: Canadian RCM projected climate change signal and
its
sensitivity to model errors. Int. J. Climatol. DOI:
10.1002/joc.1362 (in press).Sushama, L., R. Laprise, I. Popadic and
M. Allard, 2006b : Modeled current and future soil thermal regime
for North East Canada.
J. Geophys. Res. – Atmosphere (accepted for
publication).Tjernström, M., M. Zagar, G. Svensson, J.J. Cassano,
S. Pfeifer, A. Rinke, K. Wyser, K. Dethloff, C. Jones, T. Semmler
and M.
Shaw, 2005 : Modelling the arctic boundary layer : An
evaluation of six ARCMIP regional-scale models using data from
theSHEBA project. Boundary Layer Meteorology 117 (2): 337
Wyser, K., and C.G. Jones, 2005 : Aerosol and Clouds –
Modeled and observed clouds during surface heat budget of Arctic
Ocean(SHEBA). J. Geophys. Res. 110 (9) (DOI
10.1029/2004JD004751).
Zhao, T.L., S.L. Gong, X.Y. Zhang, J.-P. Blanchet, I.G. McKendry
and Z. J. Zhou, 2005: A Simulated Climatology of Asian DustAerosol
and its Trans-Pacific Transport: 1. Mean Climate and Validation. J.
Clim., 19 (1): 88–103.
(2) Peer Rewieved Publications (submitted)Brochu, R. and R.
Laprise: Surface water and energy budgets as simulated by two
generations of the Canadian Regional Climate
Model over the Mississippi and Columbia River Basins. Accepted
with corrections in Atmosphere and Ocean, 2006de Elía, R. and D.
Caya: Extracting additional information from Taylor diagrams.
Submitted to J. Climate, 2006.de Elía, R., D. Caya, A. Frigon, H.
Côté, M. Giguère, D. Paquin, S. Biner, R. Harvey, D. Plummer:
Evaluation of uncertainties in
the CRCM-simulated North American climate: nesting-related
issues. Submitted to Climate Dynamics, 2006.Dethloff, K., C. Jones,
Rinke, J. Cassano, J.H. Christensen, J.A. Curry, P. Du, E. Girard,
J.-E. Haugen, D. Jacob, M. Køltzow, A.H.
Lynch, S. Pfeifer, M.C Serreze, M.J. Shaw, M. Tjernström, K.
Wyser, M. Zagar, 2005: Clouds and radiation in ARCMIP.
(enpréparation).
Diaconescu, E. P., R. Laprise and L. Sushama, 2006: The impact
of lateral boundary data errors on the simulated climate of anested
regional climate model. Climate Dynamics (submitted November
2005)
Hu, R., J. Blanchet, and E. Girard, 2005: Aerosol effect on
surface cloud radiative forcing in the Arctic. Atmospheric
Chemistryand Physic (in review process).
Long, Z., W. Perrie, J. Gyakum, R. Laprise, and D. Caya:
Northern Lake Impacts on Local Seasonal Climate. Submitted to. J.
ofHydrometeorology.
McTaggart-Cowan, R., L. F. Bosart, J. R. Gyakum and E. Atallah:
Evolution and global impacts of a diabatically-generated warmpool:
Hurricane Katrina (2005). Monthly Weather Review (in review).
Music, B., and D. Caya: Evaluation of the Water Cycle over the
Mississippi River Basin as simulated by the Canadian
RegionalClimate Model (CRCM). Journal of Hydrometeorology (in
review).
(3) Chapter in reports or booksAlexandru, A., R. de Elía and R.
Laprise, 2006 : Geographical Distribution of Internal
Variability in Regional Climate
Downscaling. Research activities in Atmospheric and Oceanic
Modelling, WMO/TD, edited by J. Côté, April 2006, xx.Barrette, N.,
and R. Laprise, 2005: A one-dimensional model for simulating the
vertical transport of dissolved CO2 and CH4 in
hydroelectric reservoirs. Chap. 24, 575-595, in: Greenhouse Gas
Emissions: Fluxes and Processes, Hydroelectric Reservoirsand
Natural Environments. A. Tremblay, L. Varfalvy, C. Roehm and M.
Garneau (Eds.). Environmental Science Series,Springer, Berlin –
Heidelberg – New York, 732 pp.
-
Progress Report – CRCM Network – 2005-2006 29 / 34
Diaconescu, E. P., and René Laprise The impact of lateral
boundary data errors on the simulated climate of a nested
RegionalClimate Model. Research activities in Atmospheric and
Oceanic Modelling, WMO/TD, edited by J. Côté, April 2006, xx.
Fox-Rabinovitz, M.S., J. Cote, B. Dugas, M. Deque and J.
McGregor, 2006: Regional Modeling with Variable-Resolution
GCMs:International SGMIP, WMO/WGNE, Research Activities, 2006
Edition, p. 3-07.
Frigon, A., M. Slivitzky1 and D. Caya, 2006 : Hydrology of
Northern Quebec as seen by the Canadian Regional Climate
Model.Research activities in Atmospheric and Oceanic Modelling,
WMO/TD, edited by J. Côté, April 2006, xx.
Laprise, R., 2007 (a Lead Author to Chap. 11): « Regional
Climate Projections ». IPCC Fourth Assessment Report (AR4)
«ClimateChange 2007: The Physical Science Basis ».
Leduc M. and R. Laprise, 2006: CRCM sensitivity to domain size.
Research activities in Atmospheric and Oceanic Modelling,WMO/TD,
edited by J. Côté, April 2006, xx.
Lucas-Picher, P., D. Caya and S. Biner, 2006: Relation between
RCM’s internal variability and residency time of the
atmosphericparcels into the limited area domain. Research
activities in Atmospheric and Oceanic Modelling, WMO/TD, edited by
J. Côté,April 2006, xx.
Music, B., and D. Caya, 2006: Evaluation and validation of the
hydrological cycle simulated by the Canadian Regional ClimateModel
(CRCM) using an integrative approach. Research activities in
Atmospheric and Oceanic Modelling, WMO/TD, editedby J. Côté, April
2006, xx.
Radojevic, M., P. Zwack and R. Laprise, 2006 a:
Northern-Hemisphere extra-tropical cyclone activity in 1961-1990:
Comparisonof the CGCM3 with the NCEP/NCAR reanalyses. Research
activities in Atmospheric and Oceanic Modelling, WMO/TD,edited by
J. Côté, April 2006 – xxxx.
Radojevic, M., P. Zwack and R. Laprise, 2006 b: Impact of
enhanced greenhouse gases on Northern Hemisphere
extra-tropicalcyclone activity in 2041-2070 as simulated by the
CGCM3. Research activities in Atmospheric and Oceanic
Modelling,WMO/TD, edited by J. Côté, April 2006 – xxxx.
Rosu, C., and R. laprise, 2006 : The Relationship between
Cyclone Characteristics and Annual Hydrological Resources
overQuébec. Research activities in Atmospheric and Oceanic
Modelling, WMO/TD, edited by J. Côté, April 2006 – xxxx.
Separovic, L., R. de Elia and R. Laprise, 2006: Stochastic and
deterministic components in limited-area model downscaling.Research
activities in Atmospheric and Oceanic Modelling, WMO/TD, edited by
J. Côté, April 2006 – xxxx.
(4) Papers in preparationAlexandru, A., R. de Elía and R.
Laprise, 2006: internal variability in regional climate downscaling
at the seasonal scale.Lucas-Picher P., D. Caya and S. Biner, 2005:
Influence of domain size, domain location and spectral nudging on
RCM's internal
variability.Qian, M., F. Saucier, D.Caya and R. Laprise, 2005:
The study of sea ice in Hudson Bay and its effect on regional
climate using
coupled regional atmospheric model with an ice-ocean model. To
be submitted to Monthly Weather Review.Laprise, R., 2006 : Regional
climate modelling. J. Comp. Phys. (Invited paper), Special issue on
« Predicting weather, climate and
extreme events » (in preparation).Wyser, Girard, Hu, Jones et
al. Evaluation of an Ensemble of eight Arctic Regional Climate
Models: cloud and radiation (in
preparation)
(5) Conferences ProceedingsDiaconescu, E. P., and R. Laprise :
La réponse d’un modèle régional du climat aux erreurs du pilote.
Conference Proceeding of the
Ateliers de modélisation atmosphérique (AMA). 18-20 janvier
2006. Toulouse, France.Leduc, M., and R. Laprise : Effets reliés à
la taille du domaine d'intégration d'une simulation climatique
régionale. Conference
Proceeding of the Ateliers de modélisation atmosphérique (AMA).
18-20 janvier 2006. Toulouse, France.Lucas-Picher P., D. Caya and
S. Biner : Corrélation entre la variabilité interne d’un MRC et le
temps de résidence de l'écoulement
atmosphérique dans le domaine. Conference Proceeding of the
Ateliers de modélisation atmosphérique (AMA). 18-20 janvier2006.
Toulouse, France.
(6) SeminarsCaya, D., 2006 (inivited speaker): Regional Climate
Simulations for Impact Studies. Ministry of Forestry, Victoria, BC,
3 mai
2006.
Caya, D., 2006 (inivited speaker):: La simulation climatique :
un outil de planification. Alcan, Rencontre des cadres, Alma, QC,
26avril 2006.
-
Progress Report – CRCM Network – 2005-2006 30 / 34
Caya, D., 2006 (inivited speaker):: Latest Canadian Regional
Climate Model (CRCM) results for all of Canada. Hydro Power
andClimate Change Workshop Agenda, Winnipeg, Manitoba, 2 et 3 mars
2006.
Caya, D., 2005 (inivited speaker):: La simulation climatique
régionale à Ouranos, Centre d’études nordiques, Université
Laval,Québec, QC, 2 décembre 2005.
McTaggart-Cowan, R., L. Bosart, J. Gyakum and E. Atallah, 2006:
Global impacts of a diabatically generated warm pool:Hurricane
Katrina (2005). The 27th Conference on Hurricanes and Tropical
Meteorology. April 2006, Monterey, California.
Atallah, E. H., J. R. Gyakum and R. McTaggart-Cowan, 2006:
Forecast errors associated with Hurricane Rita (2005). The
27thConference on Hurricanes and Tropical Meteorology. April 2006,
Monterey, California.
Bosart, L. F., R. McTaggart-Cowan, C. A. Davis and M. T.
Montgomery, 2006: The tropical transition of Hurricane Alex
(2004):An observational perspective. The 27th Conference on
Hurricanes and Tropical Meteorology. April 2006,
Monterey,California.
Davis, C. A., L. F. Bosart and R. McTaggart-Cowan, 2006:
Tropical transition: possible mechanisms and observational needs.
The27th Conference on Hurricanes and Tropical Meteorology. April
2006, Monterey, California.
Fox-Rabinovitz, M.S., J. Côté, B. Dugas, M. Deque and J.
McGregor, 2006: Regional modeling with variable-resolution
GCMs:Stretched-Grid Model Intercomparison Project (SGMIP), 2006
Workshop on the Solution of Partial Differential Equations onthe
Sphere, June 26-29 2006, Monterey CA.
Fox-Rabinovitz, M.S., J. Cote, B. Dugas, M. Deque and J.
McGregor, 2005: International SGMIP: Results of the phase-1 and
thepreliminary results the phase-2, progress report, WMO/WCRP/WGNE
Meeting, November 2005, St. Petersburg, Russia.
McTaggart-Cowan, R., L. F. Bosart, J. R. Gyakum and E. H.
Atallah, 2006: Global impacts of a diabatically generated warm
pool:Hurricane Katrina (2005). University at Albany Lecture Series.
March 2006, Albany, New York.
McTaggart-Cowan, R., L. F. Bosart, J. R. Gyakum and E. H.
Atallah, 2006: Global impacts of a diabatically generated warm
pool:Hurricane Katrina (2005). Rutgers University Lecture Series.
March 2006, New Brunswick, New Jersey.
McTaggart-Cowan, R., L. Bosart, J. Gyakum and E. Atallah, 2005:
The impact of Hurricane Katrina (2005) on the midlatitudeflow. The
2nd International Workshop on Extratropical Transition. December
2005, Perth, Western Australia.
McTaggart-Cowan, R., L. Bosart and C. D