Transcript Creation of the FORESEE database to support climate change

Creation of the FORESEE database to
support climate change related impact
studies
Laura DOBOR, Zoltán BARCZA, Ágnes HAVASI,
Tomáš HLÁSNY
Eötvös Loránd University, Budapest, HUNGARY
BioVeL MS11 Workshop
2013-06-06/07
Outline
• Motivation
• FORESEE database
• Construction
– Data selection for the past/future
– Bias correction
• Structure of the created database
• Availability
Motivation
• QUESTION: What is the expected effect of climate
change on plant productivity and carbon balance of
croplands in Hungary?
• MAIN AIM: perform impact study using process based
biogeochemical model + crop model
• FIRST STEP: Need for daily meteorological data for the
past and also for the future.
• We did not find good database, so we decided to create
one!
FORESEE database
Daily meteorological database
Maximum/minimum temperature, precipitation
1/6 x 1/6 DEGREE
REGULAR GRID
1. 1951-2009: Observation based data
2. 2010-2100: Bias corrected climate model data based
on 10 different regional climate models (RCMs).
Construction
- Data selection Collect reliable data in daily time steps for grid points
PAST
E-OBS database
(daily)
1951
1961
2009
PR
V3.0
V5.0
TMAX
V3.0
V7.0
TMIN
V3.0
V7.0
• Modification: Monthly correction with CRU TS 1.2 database
(monthly)  fit the monthly averages (tmax,tmin) and totals (pr)
1951
2010
ENSEMBLES project
(daily)
FUTURE: 10 regional climate model
(A1B scenario)
• Modification: Bias correction based on the modified E-OBS
2100
Construction
- List of the models RCM-GCM
1
ALADIN-ARPEGE
2
CLM-HadCM3Q0
Developing institute
Lower horizontal
National
Centre
for Meteorological
Research (CNRM)
resolution
(~ 150
km)
Whole
Earth Institute of Technology Zürich (ETHZ)
Swiss Federal
3
HadRM3Q0HadCM3Q0
Hadley Centre for Climate Prediction and Research
(HC)
4
HIRHAM5-ARPEGE
Danish Meteorological Institute (DMI)
5
HIRHAM5-ECHAM5
Danish Meteorological Institute (DMI)
6
RACMO2-ECHAM5
Royal Netherlands Meteorological Institute (KNMI)
7
RCA-ECHAM5
Sweden's Meteorological and Hydrological Institute
(SMHI)
8
RCA-HadCM3Q0
9
REGCM3-ECHAM5
10
REMO-ECHAM5
Sweden's Meteorological and Hydrological Institute
Higher
resolution (~ 25 km)
(SMHI)
Limited area
The Abdus Salam International
for Theoretical
DetailedCentre
topography
and
Physics
(ICTP)
physical parameterizations
Max-Planck-Institute for Meteorology (MPI)
Example to demonstrate the problem of inherent
systematic errors
1961-1990 precipitation - REMO-ECHAM5
The model
underestimates
the annual
preicipitation
The model
overestimates
the annual
preicipitation
-2000
mm
-700
mm
+2000
mm
+700 mm
Original climate model after interpolation minus CRU corrected E-OBS
[~observation]
Construction – Bias correction I
• Every climate model output contains systematic
errors
• If we would like to use them directly [as realistic
weather], we have to execute a bias correction
on them
• Assumption: the systematic errors are stable in
the time
Determine the errors based on the past
(observation and model comparison)
Correct the models using correction factors
(same for the past and the future)
Construction – Bias correction II
•
Based on monthly cumulative distribution function
(CDF) fitting technique (Reference period: 1951-2009)
Precipitation: amount and frequency (2 steps)
1. Create or delete wet days
2. Fit the observed and the simulated precipitation time
series distribution function
Observation
based data
Climate model
result
70% dry days
50% dry days
When the model
simulates too much
wet days
• Determine quanitle functions
(1000 portions)
• Model simulation
• Observation
• To every quantiles belong one-one
precipitation amount
 Ratios of these amounts give the
correction array (1000 elements)
• Correction steps:
–
–
–
In which quanitle is ordered the given
precipitation amount?
What is the correction factor at that
quantile?
Scale with that correction factor!
Daily
[mm/day]
amout [mm/day]
precipitation amout
Daily precipitation
data
Observed
Observed data
Precipitation amount correction
Daily
Dailyprecipitation
precipitationamout
amout[mm/day]
[mm/day]
Corrected
Raw climate
climatemodel
modeldata
data
Precipitation frequency correction:
is it important? yes! - an example -
Spatial averages for the
whole target area
Wet day: prec. amount
>=0.1 mm
― CRU corrected E-OBS (obs)
Solid lines: non-corrected, raw climate model
results
Dotted lines: bias corrected climate model
results
Model selection for impact studies:
How can we visualize 10 different model in one plot?
THERMOPLUVIOGRAM FOR CENTRAL EUROPE
• yhf
RCM-GCM
1
ALADIN-ARPEGE
2
CLM-HadCM3Q0
3
HadRM3Q0- HadCM3Q0
4
HIRHAM5-ARPEGE
5
HIRHAM5-ECHAM5
6
RACMO-ECHAM5
7
RCA-ECHAM5
8
RCA-HadCM3Q0
9
REGCM-ECHAM5
10
REMO-ECHAM5
Availability
- Webpage • 33 NetCDF (Network Common Data Form)
files, size: about 22 GB
• Open access:
http://nimbus.elte.hu/FORESEE
• More information, news/updates, publications
• Current version of database: FORESEE v1.1
• It is possible to create a workflow for the data
retreival
Summary
• We created an open access daily database for
the 1951-2100 period (FORESEE), which
contains
– Observation based datasets for the past
– 10 bias corrected climate model results for the future
• The goal was to create the meteorological
background (precipitation, maximum/minimum
temperature)
– for climate change-related impact studies
– for other modeling purposes
Thank you for
your attention!