Antarctica on the scales

download report

Transcript Antarctica on the scales

Linking probabilistic climate
scenarios with downscaling
methods for impact studies
Dr Hayley Fowler
School of Civil Engineering and Geosciences
University of Newcastle, UK
With Contributions from:
Claudia Tebaldi (NCAR)
Stephen Blenkinsop, Andy Smith (Newcastle University)
Aim
Develop a framework for the construction of
probabilistic climate change scenarios to assess
climate change impacts at the:

regional (~100,000 to 250,000 km2)

river basin (~10,000 to ~100,000 km2)

catchment (~1000 to ~5000 km2) scales
Motivation

Different GCMs produce different climate change
projections, especially on a regional scale

Therefore no one model provides a true representation

Most probabilistic scenarios to date have been produced
for large regions or globally

Regional scale studies more relevant for impacts

How can we combine probabilistic climate scenarios
with downscaling methods to study impacts at the
catchment scale?
How can we combine probabilistic climate
scenarios with downscaling methods to study
impacts at the catchment scale?

Examining how well different RCMs simulate
different statistical properties of current
climate in their control climates

Do different RCM-GCM combinations produce
different future projections?

How can we combine the estimates of
different models to produce probabilistic
scenarios?
Case-study Locations
1 British Isles
2 Eden
3 Ebro
4 Gallego
5 Meuse
6 Dommel
7 Brenta
8 Scandinavia
9 Eastern Europe
Method: RCMs + WG
PRUDENCE
RCMs
Extract CFs
(Catchment)
Tebaldi Bayesian
UK Regions
EARWIG
Weather Generator
Calibrated Eden R-R
model
λ
Monte-Carlo
resampling of flow
sections based on λs
Data available for UK
RCM data – 50km x 50km
Control 1961-90
Future SRES A2 2070-2100
Interpolated observations – 5km x 5km
Data – Observations & Models
Observed series - Aggregated 5km interpolated precipitation dataset
Regional Climate Models – PRUDENCE (http://prudence.dmi.dk/)
RCM
Driving Data
Danish
Meteorological
Institute (DMI)
HIRHAM
Swedish
Meteorological
and Hydrological Institute
(SMHI)
Hadley Centre – UK Met
Office
Météo-France, France
RCAO
HadRM3P
HadAM3H A2
ECHAM4/OPYC
(OGCM SSTs)
HadAM3H A2
ECHAM4/OPYC
A2
HadAM3P
Arpège
Observed SST
PRUDENCE
Acronym
HC1
ecctrl
AquaTerra
Acronym
HIRHAM-H
HIRHAM-E
HCCTL
MPICTL
RCAO-H
RCAO-E
adeha
HAD-P
DA9
ARP-A
How well do RCMs represent
the seasonal cycle?
Mean Rainfall Comparison
Mean Daily Rainfall (mm)
6
5
HIRHAM_E
HIRHAM_H
4
RCAO_E
3
RCAO_H
2
HAD_P
ARPEGE_C
1
OBSERVED
0
jan
feb
mar
apr may
jun
jul
Month
aug sep
oct
nov dec
How well do RCMs represent
the seasonal cycle?
Mean Temperature Comparison
Mean Temperature (DegC)
16
14
HIRHAM_E
12
HIRHAM_H
RCAO_E
10
8
RCAO_H
6
HAD_P
4
ARPEGE_C
OBSERVED
2
0
jan
feb
mar apr may
jun
jul
Month
aug sep
oct
nov dec
How well do RCMs represent
the seasonal cycle?
Daily Rainfall Variance Comparison
35
HIRHAM_E
30
(mm2)
Variance of Daily Rainfall
40
HIRHAM_H
25
20
RCAO_E
RCAO_H
15
HAD_P
10
ARPEGE_C
OBSERVED
5
0
jan
feb mar apr may jun
jul
Month
aug sep
oct nov dec
Summer Skewness Coefficient
UK Regions
Method: RCMs + WG
PRUDENCE
RCMs
Extract CFs
(Catchment)
Tebaldi Bayesian
UK Regions
EARWIG
Weather Generator
Calibrated Eden R-R
model
λ
Monte-Carlo
resampling of flow
sections based on λs
Model weighting (a la Tebaldi)

Bayesian statistical model delivers a fully
probabilistic assessment of the uncertainty of
climate change projections at regional scales

Based on:

Reliability Ensemble Average method (Giorgi and
Mearns, 2002)

Summary measures of regional climate change,
based on a WEIGHTED AVERAGE of different
climate model responses
Model weighting (a la Tebaldi)

Weights account for:

BIAS - the performance of GCMs when
compared to present day climate ( i.e. results
from model validation)

CONVERGENCE - the degree of consensus
among the various GCMs’ responses/
Model weighting (a la Tebaldi)





pdf of change in temperature and precipitation
fitted using area-averages of the model output
Prior pdfs are assumed to be uninformative
Data from regional models/observation
incorporated through Bayes’ theorem, to derive
posterior pdfs
Model-specific “reliabilities parameters” estimated
as a function of model performance in reproducing
current climate (1961-1990) and agreement with
the ensemble consensus for future projections
These are standardised and applied as weights in
the downscaling step
NWE Seasonal Mean λ
Precipitation
ARP_C
HAD_P
HIRH_E
HIRH_H
RCAO_E
RCAO_H
DJF
0.07
0.19
0.25
0.26
0.08
0.15
MAM
0.08
0.05
0.11
0.23
0.26
0.27
JJA
0.15
0.06
0.16
0.23
0.18
0.22
SON
0.11
0.11
0.21
0.20
0.14
0.23
Temperature
ARP_C
HAD_P
HIRH_E
HIRH_H
RCAO_E
RCAO_H
DJF
0.23
0.22
0.12
0.19
0.11
0.13
MAM
0.17
0.22
0.15
0.26
0.09
0.1
JJA
0.08
0.18
0.09
0.25
0.16
0.25
SON
0.12
0.23
0.16
0.24
0.13
0.12
Method: RCMs + WG
PRUDENCE
RCMs
Extract CFs
(Catchment)
Tebaldi Bayesian
UK Regions
EARWIG
Weather Generator
Calibrated Eden R-R
model
λ
Monte-Carlo
resampling of flow
sections based on λs
EArWiG
EA Weather Generator




Developed for EA for
catchment scale Decision
Support Tool models
Generates series of daily
rainfall, T, RH, wind, sunshine
and PET on 5km UK grid
Observed and climate change
based on UKCIP02 scenarios
Collaborative with CRU, UEA
EArWiG


Map viewer interface developed
Can select catchments, time periods and
different UKCIP02 scenarios
Toolbar
Catchments tab
Model tab
Catchment finder
OSGB locator
OSGB pointer coords
Map window
Neyman-Scott Rectangular Pulses
Rainfall Model
• Storm origins arrive in a Poisson
time
process with arrival rate λ
• Raincell duration is exponentially
distributed with parameter η
time
intensity
• Each storm origin generates C
raincells separated from the storm
origin by time intervals exponentially
distributed with parameter β
time
• Rainfall intensity is equal to the sum
of the intensities of all the active cells
at that instant
total intensity
• Raincell intensity is exponentially
distributed with parameter ξ
time
Weather Generator

Depending on whether the day is wet or dry, other meteorological
variables are determined by regression relationships with
precipitation and values of the variables on the previous day

Regression relationships maintain both the cross- and autocorrelations between and within each of the variables
Change factor fields

Change factor fields are applied to the fitted rainfall model
statistics:






Mean
Variance
PD
Skewness Coefficient
Lag 1 Autocorrelation
Change factor fields are applied to the weather generator
statistics:


Mean temperature
Temperature SD
CF Summer mean temperature
CF Winter mean precipitation
CF Spring PD
Method: RCMs + WG
PRUDENCE
RCMs
Extract CFs
(Catchment)
Tebaldi Bayesian
UK Regions
EARWIG
Weather Generator
Calibrated Eden R-R
model
λ
Monte-Carlo
resampling of flow
sections based on λs
Rainfall-runoff model




ADM model, simplified version of Arno
Calibrated for Eden catchment on observed
data
R2=0.73, 0.78
Each simulated climate used to produce
simulated flow series (30 years) for each
climate model using P and PET
EARWIG run for each RCM
1
2071-2100
2
Had_P
RCAO_E
1961-1990 Control
Each series is 30 years in length
3
4
… 1000
NWE Seasonal Mean λ
Precipitation
ARP_C
HAD_P
HIRH_E
HIRH_H
RCAO_E
RCAO_H
DJF
0.07
0.19
0.25
0.26
0.08
0.15
MAM
0.08
0.05
0.11
0.23
0.26
0.27
JJA
0.15
0.06
0.16
0.23
0.18
0.22
SON
0.11
0.11
0.21
0.20
0.14
0.23
Temperature
ARP_C
HAD_P
HIRH_E
HIRH_H
RCAO_E
RCAO_H
DJF
0.23
0.22
0.12
0.19
0.11
0.13
MAM
0.17
0.22
0.15
0.26
0.09
0.1
JJA
0.08
0.18
0.09
0.25
0.16
0.25
SON
0.12
0.23
0.16
0.24
0.13
0.12
Re-sampling




Monte-Carlo re-sampling technique used to
weight models according to λ values from
Bayesian weighting
Random numbers used to choose a control and
future run for a particular RCM, then seasonal
statistics of change in mean flow, SD flow, 5th
and 95th percentiles calculated.
If seasonal λ=0.14 then random number
generator produces 140 resamples from a
particular RCM
Generates total of 1000 change statistics for
each season – pdf fitted used kernel density
2080s
2020s
Questions for the audience

Should we weight models (CG)?

Should we be weighting on statistics other than
mean?

If so, what?

Should we be looking at weighting by some
spatial bias measure rather than a simple
regional average? Makes the statistics harder…

Models may produce reasonable mean
statistics and get higher order statistics
important for impact studies wrong