· Rib21 Constrained TCR · Dobler SICE rmse BNS · Dobler SICE rmse NAtl · Dobler SST rmse NAtl · Dobler SST rmse MED · Mindlin21 tropampl · Atlas Dtas WCE DJF · Atlas Dtas world DJF · Atlas Dtas world JJA · Atlas Dpr NEU DJF · Atlas Dpr WCE DJF · Atlas Dpr world DJF · Atlas Dpr NEU JJA · Atlas Dpr WCE JJA · Atlas Dpr world JJA · Can20 marle · Pri20 storm track DJF · Pri20 storm track JJA · Tok20 Constrained TCR · Fer21 Lamb TPMS · Fas20 · Cob21 · Beo21
· Atlas Dtas MED DJF · Atlas Dpr MED DJF
Located in CMIP6_studies/AR6.yaml
None
- key: AR6 TCR
doi: None
type: performance
spatial_scope: Global
temporal_scope: Annual
data_source: reference
metric:
name: TCR
long_name: Transient Climate Response
units: K
variables: tas
comment:
TCR as provided by th IPCC WGI AR6 on Table 7.SM.5 (https://www.ipcc.ch/re
port/ar6/wg1/downloads/report/IPCC_AR6_WGI_Chapter_07_Supplementary_Materi
al.pdf).
plausible_values:
- min: 1.2
max: 2.4
source: reference
comment:
This is a 90% (very likely) range for the TCR according to AR6 Technical
Summary: Based on process understanding, warming over the instrumental
record, and emergent constraints, the best estimate of TCR is 1.8 degC,
the likely range is 1.4 to 2.2 degC and the very likely range is 1.2 to
2.4 DegC. There is a high level of agreement among the different lines of
evidence (Figure TS.16c) (high confidence). {7.5.5}
Located in CMIP6_studies/Bru20.yaml
Lukas Brunner et al. (2020) Reduced global warming from CMIP6 projections when weighting models by performance and independence, https://doi.org/10.5194%2Fesd-11-995-2020
- key: Bru20 perf
doi: 10.5194/esd-11-995-2020
type: performance
spatial_scope: Global
data_source: reference
metric:
name: perf
long_name: None
units: None
variables: ['tas', 'psl']
comment:
Evaluation of the model using global criteria performance criteria (past
trends + spatial pattern) Model with performance scores below 0.006 are
considered as irrealistic. This study is based on multi-member for each
model, not on only one member. Wrt to the original article, we consider
here only the performance criteria and not the combined criteria that
takes also into account the independence.
period:
reference: 1980-2014
plausible_values:
- min: 0.006
max: 0.2
source: author
comment:
The 0.006 threshold has been provided by the author (expert judgment)
after email exchanges with S. Somot
Located in CMIP6_studies/Dalelane.yaml
['pers. comm.', 'C. Dalelane', 'DWD']
- key: Dalelane MNQS
doi: ['pers. comm.', 'C. Dalelane', 'DWD']
type: performance
spatial_scope: Global
temporal_scope: Annual
data_source: author
metric:
name: MNQS
long_name: Multivariate Network Quality Score for Global Teleconnections
units: 1
variables: tos, z500
comment:
Data is converted to seasonal anomalies, detrended with season-reliant
trend-EOFs. Seasonal variances normalized. Adjacency-matrix between all
pairs of grid cell with Spearman's rank correlation. Maximal domains in
tos (z500) with average pairwise rank correlation>0.95(0.93)- quantile of
all pairwise correlations are found. All pairwise links between domains
(area weighted average time series) calculated with Distance correlation
in tos, z500 and cross-links between tos and z500. Tested to level 0.05
with control of False Discovery Rate=0.05. New adjacency matrices
constructed based on domain links. Adjacency matrices compared to
references with Structural Similarity Index (SSIM). Exponential
transform wrt. value 1 for all 3 variables (individual Network Quality
Score-NQS). Geometric mean of NQSs over variables (Multivariate Network
Quality Score, MNQS). Average of MNQSs over references.
best: 1
worst: 0
plausible_values:
- min: 0
max: 1
source: author
comment:
the higher the better, MNQS between references in the table for comparison
Located in CMIP6_studies/Qasmi.yaml
Aur'elien Ribes et al. (2021) Making climate projections conditional on historical observations, https://doi.org/10.1126%2Fsciadv.abc0671
- key: Qasmi Constr Global Dtas ssp245 2050
doi: 10.1126/sciadv.abc0671
type: performance
spatial_scope: Global
temporal_scope: Annual
data_source: author
metric:
name: Constrained-Dtas
long_name: Observationally-contrained future climate change
units: binary
variables: tas
comment:
Constrained global annual temperature future climate change range,
2041-2060 vs 1850-1900, SSP245 (adapted from Ribes et al. 2021 by S.
Qasmi). In particular by adding recently available CMIP6 GCM, now 40 GCMs.
period:
reference: 1850-1900
target: 2041-2060
plausible_values:
- min: 1
max: 1
source: author
comment:
models lying outside the observationally-constrained 90% interval obtained
by the method are considered as implausible. The 90% interval is [1.5 ;
2.1]degC for the period 2041-2060 vs 1850-1900, SSP245. Multi-member
ensemble mean is used in this study for every model. Note that this
criteria is very strict and can potentially eliminate a large number of
GCMs.
Located in CMIP6_studies/Bra21.yaml
Swen Brands et al. (2021) A circulation-based performance atlas of the CMIP5 and 6 models, https://doi.org/10.5194%2Fgmd-2020-418
- key: Bra21 Lamb EUR
doi: 10.5194/gmd-2020-418
type: performance
spatial_scope: EUR
temporal_scope: Annual
data_source: author_adapted
metric:
name: lwtmae
long_name: MAE of 27 Lamb Weather Type relative frequencies
units: percent
variables: psl
comment:
Mean absolute error (MAE) of the simulated vs. quasi-observed (reanalysis)
relative frequencies for the 27 Lamb Weather Types representing recurrent
regional atmospheric circulation patterns. The MAE was calculated
separately for each grid box of a regular 2.5 deg lat-lon mesh extending
from 22.5W to 42.5E and 30N to 70N. The spatial median MAE is provided
here. Reference dataset to compute the metric is ERA-Interim. As
reference, the value for the JRA-55 reanalysis (EUR) is 0.0956
best: 0
worst: 100
period:
reference: 1979-2005
plausible_values:
- min: 0
max: 1
source: eurocordex_gcm_selection_team
comment:
Test value
- min: 0
max: 5
source: author
comment:
The range of plausible values is directly obtained from the reference, the
maximum MAE obtained there is here rounded to the next integer.
Located in CMIP6_studies/Div20.yaml
Paolo Davini et al. (2020) From CMIP3 to CMIP6: Northern Hemisphere Atmospheric Blocking Simulation in Present and Future Climate, https://doi.org/10.1175%2Fjcli-d-19-0862.1
- key: Dav20 blocking freq DJF
doi: 10.1175/JCLI-D-19-0862.1
type: performance
spatial_scope: EUR
temporal_scope: DJF
data_source: author
metric:
name: blocking
long_name: blocking frequency
units: categorical
variables: zg
comment:
Scoring of models on performance for blocking frequency. Blocking
frequency bias has been calculated by the method of Davini and D'Andrea
(2020). The individual CMIP6 models have then been clustered into
categories based on their RMSE, bias and correlation compared to multiple
reanalysis JRA-55, NCEP-NCAR, ERA-40, ERA-Interim. Data for individual
CMIP6 models and clustering of errors into categories are provided by the
author for Europe. Based on the method of qualitative scoring in
McSweeney et al. (2015) and adapted for CMIP6. The scoring has been
changed from the traffic light coding to numbers for EURO-CORDEX. Values
0 - Low errors over both local and remote regions. Captures key
characteristics of the criteria spatially or temporarily, 1 - Some
substantial errors present but not widespread or not present in the
local region of interest. Location of larger remote errors are not
known to have a downstream impact in the local region of interest.
Captures key characteristics of the criteria spatially or temporarily, 2 -
Substantial errors in remote regions where downstream effects could be
expected to impact on the reliability of regional information and/or
present in the local region of interest, 3 - Large widespread errors to
the extent that the model is unable to represent the present-day
climatology in a useful way and future projections by the model cannot
be interpreted in a meaningful way.
best: 0
worst: 3
period:
reference: 1961-2000
plausible_values:
- min: 0
max: 2
source: author
comment:
Large widespread errors (value 3) lead to consider the model unplausible.
Located in CMIP6_studies/Div20.yaml
Paolo Davini et al. (2020) From CMIP3 to CMIP6: Northern Hemisphere Atmospheric Blocking Simulation in Present and Future Climate, https://doi.org/10.1175%2Fjcli-d-19-0862.1
- key: Dav20 blocking freq JJA
doi: 10.1175/JCLI-D-19-0862.1
type: performance
spatial_scope: EUR
temporal_scope: JJA
data_source: author
metric:
name: blocking
long_name: blocking frequency
units: categorical
variables: zg
comment:
Scoring of models on performance for blocking frequency. Blocking
frequency bias has been calculated by the method of Davini and D'Andrea
(2020). The individual CMIP6 models have then been clustered into
categories based on their RMSE, bias and correlation compared to multiple
reanalysis JRA-55, NCEP-NCAR, ERA-40, ERA-Interim. Data for individual
CMIP6 models and clustering of errors into categories are provided by the
author for Europe. Based on the method of qualitative scoring in
McSweeney et al. (2015) and adapted for CMIP6. The scoring has been
changed from the traffic light coding to numbers for EURO-CORDEX. Values
0 - Low errors over both local and remote regions. Captures key
characteristics of the criteria spatially or temporarily, 1 - Some
substantial errors present but not widespread or not present in the
local region of interest. Location of larger remote errors are not
known to have a downstream impact in the local region of interest.
Captures key characteristics of the criteria spatially or temporarily, 2 -
Substantial errors in remote regions where downstream effects could be
expected to impact on the reliability of regional information and/or
present in the local region of interest, 3 - Large widespread errors to
the extent that the model is unable to represent the present-day
climatology in a useful way and future projections by the model cannot
be interpreted in a meaningful way.
best: 0
worst: 3
period:
reference: 1961-2000
plausible_values:
- min: 0
max: 2
source: author
comment:
Large widespread errors (value 3) lead to consider the model unplausible.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SST rmse EUR
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: EUR
temporal_scope: Annual
data_source: author
metric:
name: sstrmse
long_name: Sea Surface Temperature RMSE w.r.t. HadISST
units: K
variables: tos
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly SST are
computed. Then, over all 12 maps of biases the RMSEs are calculated. All
the models are interpolated onto the grid of the refererence HadISST 1.1
monthly average sea surface temperature (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=0 Area definitions: BNS: Baltic and North
Sea NBS: Norwegian and Barents Sea NAtl: Nordic Atlantic (replaced by NBS)
SNA: (Southern) Nord Atlantic MED: Mediterranean (disabled, use Sevault
MED SST instead. ) BLK: Black Sea EUR: Europe box Maps are provided in: h
ttps://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xG
AmnNSIiI/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 3
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/McSw15.yaml
C. F. McSweeney et al. (2014) Selecting CMIP5 GCMs for downscaling over multiple regions, https://doi.org/10.1007%2Fs00382-014-2418-8
- key: McSw15 circ DJF
doi: 10.1007/s00382-014-2418-8
type: performance
spatial_scope: EUR
temporal_scope: DJF
data_source: author
metric:
name: circulation
long_name: Large scale atmospheric circulation pattern
units: categorical
variables: ua850 va850
comment:
Qualitative large scale circulation pattern score based on overall
pattern, bias and RMSE. Based on 20 year climatology comparison with ERA5
1995-2014. Based on the method of qualitative scoring in McSweeney et al.
(2015) and adapted for CMIP6. The scoring has been changed from the
traffic light coding to numbers for EURO-CORDEX. Values 0 -
Low errors over both local and remote regions. Captures key
characteristics of the criteria spatially or temporarily, 1 - Some
substantial errors present but not widespread or not present in
the local region of interest. Location of larger remote errors are not
known to have a downstream impact in the local region of interest.
Captures key characteristics of the criteria spatially or temporarily,
2 - Substantial errors in remote regions where downstream effects could
be expected to impact on the reliability of regional information
and/or present in the local region of interest, 3 - Large widespread
errors to the extent that the model is unable to represent the
present-day climatology in a useful way and future projections
by the model cannot be interpreted in a meaningful way.
best: 0
worst: 3
period:
reference: 1995-2014
plausible_values:
- min: 0
max: 2
source: author
Located in CMIP6_studies/McSw15.yaml
C. F. McSweeney et al. (2014) Selecting CMIP5 GCMs for downscaling over multiple regions, https://doi.org/10.1007%2Fs00382-014-2418-8
- key: McSw15 circ JJA
doi: 10.1007/s00382-014-2418-8
type: performance
spatial_scope: EUR
temporal_scope: JJA
data_source: author
metric:
name: circulation
long_name: Large scale atmospheric circulation pattern
units: categorical
variables: ua850 va850
comment:
Qualitative large scale circulation pattern score based on overall
pattern, bias and RMSE. Based on 20 year climatology comparison with ERA5
1995-2014. Based on the method of qualitative scoring in McSweeney et al.
(2015) and adapted for CMIP6. The scoring has been changed from the
traffic light coding to numbers for EURO-CORDEX. Values 0 -
Low errors over both local and remote regions. Captures key
characteristics of the criteria spatially or temporarily, 1 - Some
substantial errors present but not widespread or not present in
the local region of interest. Location of larger remote errors are not
known to have a downstream impact in the local region of interest.
Captures key characteristics of the criteria spatially or temporarily,
2 - Substantial errors in remote regions where downstream effects could
be expected to impact on the reliability of regional information
and/or present in the local region of interest, 3 - Large widespread
errors to the extent that the model is unable to represent the
present-day climatology in a useful way and future projections
by the model cannot be interpreted in a meaningful way.
best: 0
worst: 3
period:
reference: 1995-2014
plausible_values:
- min: 0
max: 2
source: author
Located in CMIP6_studies/Nabat.yaml
['pers_comm', 'Pierre Nabat']
- key: Nabat EUR AOD
doi: ['pers_comm', 'Pierre Nabat']
type: performance
spatial_scope: EUR
temporal_scope: Annual
data_source: author
metric:
name: aod_rmse
long_name: plausibillity of RMSE of the European Aerosol Optical Depth
units: aod
variables: aod550
comment:
Aerosol Optical Depth (AOD) spatial RMSE, annual mean, satellite reference
dataset (MACv2, 2000-2014)
best: 0
worst: +inf
plausible_values:
- min: 0
max: 0.2
source: author
comment:
Plausibility threshold is set at 0.2. All models are below 0.2 except for
2 GCMs from the same institute that are above 0.7
Located in CMIP6_studies/Nabat.yaml
['pers_comm', 'Pierre Nabat']
- key: Nabat EUR AOD hist trend
doi: ['pers_comm', 'Pierre Nabat']
type: performance
spatial_scope: EUR
temporal_scope: Annual
data_source: author
metric:
name: aod_histtrend
long_name: plausibillity of past trend of the European Aerosol Optical Depth
units: aod
variables: aod550
comment:
Change in Aerosol Optical Depth (AOD) between 2 sub-periods of the
historical run, annual mean Difference in AOD is computed between
2000-2014 and 1976-1990. This period corresponds to the well-known
brightening period in Europe during which it is virtual certain that AOD
has decreased over Europe This metrics is inspired by Nabat et al. 2014,
doi:10.1002/2014GL060798 The MACv2 dataset can be used to obtain a low-
confidence estimate of the real value. MACv2 shows a AOD decrease of
-0.0315 between 1976-1990 and 2000-2014. Any GCM having a trend more than
2 times MACv2 (<-0.08) can be considered as showing a strong AOD past
trend that will contribute to re-inforce the historical warming in the RCM
best: 0
worst: +inf
plausible_values:
- min: -999
max: 0
source: author
comment:
Plausibility threshold is set at 0, meaning that any postive value
(increase in AOD over the period) is considered as implausible.
Located in CMIP6_studies/Oud20.yaml
Thomas Oudar et al. (2020) Drivers of the Northern Extratropical Eddy-Driven Jet Change in CMIP5 and CMIP6 Models, https://doi.org/10.1029%2F2019gl086695
- key: Oud20 jetpos
doi: 10.1029/2019GL086695
type: performance
spatial_scope: EUR
temporal_scope: ONDJFM
data_source: author
metric:
name: jetpos
long_name: Jet Stream North-South relative position
units: degrees_north
variables: ua850
comment:
Jet position bias against ERA5 in the Central Atlantic region. Note that
the bias is estimated by subtracting the ONDJFM mean eddy-driven jet
position over the period 1979-2018 in ERA5 to each model mean eddy-driven
jet position over the same period.
best: 0
worst: [90, -90]
period:
reference: 1979-2018
plausible_values:
- min: -3
max: 3
source: author
Located in CMIP6_studies/Pri20.yaml
Matthew D. K. Priestley et al. (2020) An Overview of the Extratropical Storm Tracks in CMIP6 Historical Simulations, https://doi.org/10.1175%2Fjcli-d-19-0928.1
- key: Pri20 storm track
doi: 10.1175/JCLI-D-19-0928.1
type: performance
spatial_scope: EUR
temporal_scope: DJF+JJA
data_source: author
metric:
name: storm_track
long_name: zonal mean North Atlantic storm track
units: categorical
variables: ua850 va850 MSLP 850 relative vorticity
comment:
Scoring of models on performance for the North Atlantic storm track. Based
on RMSE of the zonal mean track profile between 25-80N compared to ERA5
and qualitative assessment of the trimodal structure of the storm track.
Storm track calculated by method in Priestly et al. (2020), data and
scores provided by author. Based on the method of qualitative scoring in
McSweeney et al. (2015) and adapted for CMIP6. The scoring has been
changed from the traffic light coding to numbers for EURO-CORDEX. Values
0 - Low errors over both local and remote regions. Captures key
characteristics of the criteria spatially or temporarily, 1 - Some
substantial errors present but not widespread or not present in the
local region of interest. Location of larger remote errors are not
known to have a downstream impact in the local region of interest.
Captures key characteristics of the criteria spatially or temporarily, 2 -
Substantial errors in remote regions where downstream effects could be
expected to impact on the reliability of regional information and/or
present in the local region of interest, 3 - Large widespread errors to
the extent that the model is unable to represent the present-day
climatology in a useful way and future projections by the model cannot
be interpreted in a meaningful way.
best: 0
worst: 3
period:
reference: 1979-2014
plausible_values:
- min: 0
max: 2
source: author
comment:
Large widespread errors (value 3) lead to consider the model unplausible.
Located in CMIP6_studies/Winderlich.yaml
['pers_comm', 'K. Winderlich', 'DWD']
- key: Winderlich SCQS
doi: ['pers_comm', 'K. Winderlich', 'DWD']
type: performance
spatial_scope: EUR
temporal_scope: Annual
data_source: author
metric:
name: SCQS
long_name: Synoptic Circulation Quality Score
units: 1
variables: z500
comment:
Domain is CORDEX-EUR. Data z500 is converted to daily anomalies and
normalized by daily standard deviation. Data are then attributed to
previously obtained Synoptic Circulation (SP) classes. Number of Synoptic
Circulation classes is 43 (derived on daily ERA-Interim data, 1979-2018).
For each model, 7 variables are computed from the attributed daily data.
1) HIST - frequency of each SP-class (year through) 2) HIST_JFD -
frequency of each SP-class (winter) 3) HIST_MAM - frequency of each SP-
class (spring) 4) HIST_JJA - frequency of each SP-class (summer) 5)
HIST_SON - frequency of each SP-class (autumn) 6) SEQUENCE - matrix of
frequencies for the subsequent occurrence of the pair of two
synoptic patterns SPi?SPj. 7) PERSISTENCE - matrix of frequency for
persistence of each SP-class for 1,2,3,.. N days in a
row. The SCQS is computed as the mean of 7 individual Quality Scores
computed on each of these variables.
best: 1
worst: 0
plausible_values:
- min: 0
max: 1
source: author
comment:
The higher the better, SCQS between reference reanalysis (ERA-Interim) and
an alternative reanalysis (NCAR-NCEP1) is in the table for comparison.
Located in CMIP6_studies/Sevault.yaml
['pers_comm', 'F. Sevault', 'CNRM']
- key: Sevault MED SST
doi: ['pers_comm', 'F. Sevault', 'CNRM']
type: performance
spatial_scope: MED
temporal_scope: Annual
data_source: author
metric:
name: sst_rmse
long_name: Sea surface temperature RMSE
units: K
variables: sst
comment:
For the performance criteria, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014. It means that we first compute the
temporal average to obtain a mean seasonal cycle - 12maps- of the bias
maps and then we compute the spatio-temporal RMSE. All the models are
interpolated on the grid of the refererence dataset, and then a mask of
the Mediterranean Sea is applied (no Black Sea). The reference dataset is
a specific CMEMS product developed for the Mediterranean Sea,
GOS-L4_GHRSST-SSTfnd-OISST_HR_REP-MED-v02.0-fv02.0 data (Pisano et al.
2016, doi:10.1016/j.rse.2016.01.019, Casey et al. 2010,
doi:10.1007/978-90-481-8681-5_16 . Generated/provided by Copernicus Marine
Service and CNR - ISMAR ROME).
best: 0
worst: inf
plausible_values:
- min: 0
max: 2
source: author
comment:
The plausibility threshold is difficult to set.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SICE rmse NBS
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: NBS
temporal_scope: Annual
data_source: author
metric:
name: siconcrmse
long_name: Sea Ice RMSE w.r.t. HadICE
units: percent
variables: siconc
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly sea-ice
concentrations are computed. Then, over all 12 maps of biases the RMSEs
are calculated. All the models are interpolated onto the grid of the
refererence HadISST sea ice concentration (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=1435900047 Area definitions: BNS: Baltic
and North Sea (replaced by BAL) BAL: Baltic Sea NBS: Norwegian and Barents
Sea NAtl: Nordic Atlantic (replaced by NBS) Maps are provided in: https://
docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xGAmnNSIi
I/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 33.5
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SST rmse NBS
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: NBS
temporal_scope: Annual
data_source: author
metric:
name: sstrmse
long_name: Sea Surface Temperature RMSE w.r.t. HadISST
units: K
variables: tos
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly SST are
computed. Then, over all 12 maps of biases the RMSEs are calculated. All
the models are interpolated onto the grid of the refererence HadISST 1.1
monthly average sea surface temperature (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=0 Area definitions: BNS: Baltic and North
Sea NBS: Norwegian and Barents Sea NAtl: Nordic Atlantic (replaced by NBS)
SNA: (Southern) Nord Atlantic MED: Mediterranean (disabled, use Sevault
MED SST instead. ) BLK: Black Sea EUR: Europe box Maps are provided in: h
ttps://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xG
AmnNSIiI/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 4.5
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SST rmse BNS
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: BNS
temporal_scope: Annual
data_source: author
metric:
name: sstrmse
long_name: Sea Surface Temperature RMSE w.r.t. HadISST
units: K
variables: tos
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly SST are
computed. Then, over all 12 maps of biases the RMSEs are calculated. All
the models are interpolated onto the grid of the refererence HadISST 1.1
monthly average sea surface temperature (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=0 Area definitions: BNS: Baltic and North
Sea NBS: Norwegian and Barents Sea NAtl: Nordic Atlantic (replaced by NBS)
SNA: (Southern) Nord Atlantic MED: Mediterranean (disabled, use Sevault
MED SST instead. ) BLK: Black Sea EUR: Europe box Maps are provided in: h
ttps://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xG
AmnNSIiI/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 2.5
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SICE rmse BAL
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: BAL
temporal_scope: Annual
data_source: author
metric:
name: siconcrmse
long_name: Sea Ice RMSE w.r.t. HadICE
units: percent
variables: siconc
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly sea-ice
concentrations are computed. Then, over all 12 maps of biases the RMSEs
are calculated. All the models are interpolated onto the grid of the
refererence HadISST sea ice concentration (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=1435900047 Area definitions: BNS: Baltic
and North Sea (replaced by BAL) BAL: Baltic Sea NBS: Norwegian and Barents
Sea NAtl: Nordic Atlantic (replaced by NBS) Maps are provided in: https://
docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xGAmnNSIi
I/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 17.5
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SST rmse BLK
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: BLK
temporal_scope: Annual
data_source: author
metric:
name: sstrmse
long_name: Sea Surface Temperature RMSE w.r.t. HadISST
units: K
variables: tos
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly SST are
computed. Then, over all 12 maps of biases the RMSEs are calculated. All
the models are interpolated onto the grid of the refererence HadISST 1.1
monthly average sea surface temperature (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=0 Area definitions: BNS: Baltic and North
Sea NBS: Norwegian and Barents Sea NAtl: Nordic Atlantic (replaced by NBS)
SNA: (Southern) Nord Atlantic MED: Mediterranean (disabled, use Sevault
MED SST instead. ) BLK: Black Sea EUR: Europe box Maps are provided in: h
ttps://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xG
AmnNSIiI/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 2.5
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/Dobler.yaml
['pers. comm.', 'A. Dobler']
- key: Dobler SST rmse SNA
doi: ['pers. comm.', 'A. Dobler']
type: performance
spatial_scope: SNA
temporal_scope: Annual
data_source: author
metric:
name: sstrmse
long_name: Sea Surface Temperature RMSE w.r.t. HadISST
units: K
variables: tos
comment:
Analogue to the MED SST RMSE, we compute the spatial RMSE on the 12-month
bias maps over the period 1985-2014: first the average monthly SST are
computed. Then, over all 12 maps of biases the RMSEs are calculated. All
the models are interpolated onto the grid of the refererence HadISST 1.1
monthly average sea surface temperature (Rayner et al. 2003,
DOI:10.1029/2002JD002670) Missing values due to non-existing sea areas (in
the GCM) are coded as -99 (RMSE is strictly positive). ----- Numbers are
provided in: https://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTB
z9ctgRrCHqw5xGAmnNSIiI/edit#gid=0 Area definitions: BNS: Baltic and North
Sea NBS: Norwegian and Barents Sea NAtl: Nordic Atlantic (replaced by NBS)
SNA: (Southern) Nord Atlantic MED: Mediterranean (disabled, use Sevault
MED SST instead. ) BLK: Black Sea EUR: Europe box Maps are provided in: h
ttps://docs.google.com/spreadsheets/d/1xvqc2CtKmi1UOCftX5hTBz9ctgRrCHqw5xG
AmnNSIiI/edit#gid=334563502 R-script reports (PDF files) used for the
calcualtions are available at
https://drive.google.com/drive/folders/1MRNO_h6EGcyGs4d82vqtTLyTLtNxHaQ0
best: 0
worst: +inf
period:
reference: 1985-2014
plausible_values:
- min: 0
max: 2.5
source: eurocordex_gcm_selection_team
comment:
Upper limit: (Mean + 2*sd) of the RMSEs of 29 models, rounded up to the
next half integer.
Located in CMIP6_studies/AR6.yaml
None
- key: AR6 TCR as spread
doi: None
type: future_spread
spatial_scope: Global
temporal_scope: Annual
data_source: reference
metric:
name: TCR
long_name: Transient Climate Response
units: K
variables: tas
comment:
TCR as provided by th IPCC WGI AR6 on Table 7.SM.5 (https://www.ipcc.ch/re
port/ar6/wg1/downloads/report/IPCC_AR6_WGI_Chapter_07_Supplementary_Materi
al.pdf).
classes:
- limits: [-10, 1.2, 1.6, 2.0, 2.4, 10]
labels: ['very low', 'low', 'moderate', 'high', 'very high']
source: eurocordex_gcm_selection_team
comment:
A proposal with 0.4 degree bins in the very likely range (1.2, 2.4)
provided by the IPCC AR6
Located in CMIP6_studies/Sch20.yaml
Manuel Schlund et al. (2020) Emergent constraints on equilibrium climate sensitivity in CMIP5: do they hold for CMIP6?, https://doi.org/10.5194%2Fesd-11-1233-2020
- key: Sch20 ECS
doi: 10.5194/esd-11-1233-2020
type: future_spread
spatial_scope: Global
temporal_scope: Annual
data_source: reference
metric:
name: ECS
long_name: Equilibrium climate sensitivity
units: K
variables: tas
comment:
ECS is calculated with ESMValTool (Gregory method). CMIP5 range is [2.08,
4.67]. Ensemble member added to make it comparable to scenarioMIP runs,
although ECS is derived from 4xCO2 runs, unrelated to scenarioMIP.
Located in CMIP6_studies/Nabat.yaml
['pers_comm', 'Pierre Nabat']
- key: Nabat EUR AOD future change
doi: ['pers_comm', 'Pierre Nabat']
type: future_spread
spatial_scope: EUR
temporal_scope: Annual
data_source: author
metric:
name: aod_futurechange
long_name: Future evolution of the European Aerosol Optical Depth
units: aod
variables: aod550
comment:
Change in Aerosol Optical Depth (AOD) between periods over Europe, annual
mean Difference in AOD is computed between 2086-2100 (SSP585) and
2000-2014 (HIST)
period:
reference: 2000-2014
target: 2086-2100
classes:
- limits: [-99, -0.04, 0, 99]
labels: ['strong decrease', 'decrease', 'increase']
source: author
comment:
Located in CMIP6_studies/Oud20.yaml
Thomas Oudar et al. (2020) Drivers of the Northern Extratropical Eddy-Driven Jet Change in CMIP5 and CMIP6 Models, https://doi.org/10.1029%2F2019gl086695
- key: Oud20 jetposdelta
doi: 10.1029/2019GL086695
type: future_spread
spatial_scope: EUR
temporal_scope: ONDJFM
metric:
name: jetposdelta
long_name: Jet Stream North-South position delta change
units: degrees_north
variables: ua850
comment:
Jet position delta change estimated by subtracting the ONDJFM mean eddy-
driven jet position over the period 2080-2099 (ssp585) w.r.t.
preindustrial 1860-1900
period:
reference: 1860-1900
target: 2080-2099
classes:
- limits: [-90, -0.5, 0.5, 90]
labels: ['strong south change', 'weak change', 'strong north change']
source: author
Located in CMIP6_studies/Qasmi.yaml
Saïd Qasmi et al. (2022) Reducing uncertainty in local temperature projections, https://doi.org/10.1126%2Fsciadv.abo6872
- key: Qasmi Constr EUR Dtas ssp245 2050 JJA
doi: 10.1126/sciadv.abo6872
type: future_spread
spatial_scope: MED+NEU+CEU
temporal_scope: JJA
data_source: author
metric:
name: deltatas_class
long_name: Warming classes according to Observationally-constrained Summer European future surface air temperature change in 2041-2060 in Summer
units: categorical
variables: tas
comment:
Regional tas change in Europe MED, NEU, CEU, MED+NEU+CEU, DJF, JJA,
2041-2060 vs 1850-1900, SSP245. Values are given only for land points.
Adapted from Qasmi and Ribes 2022, Sci. Adv. by S. Qasmi following S.
Somot's request Numerical values available soon. Only warming classes for
now. We report here only warming classes for JJA and for the joined
MED+NEU+CEU domain
period:
reference: 1850-1900
target: 2041-2060
classes:
- limits: [0.5, 1.5, 2.5, 3.5, 4.5, 5.5]
labels: ['implausible cold', 'weak warming', 'medium warming', 'strong warming', 'implausible warm']
source: author
comment:
Warming classes are determined wrt an observationally-constrained range
for the future regional warming based on Ribes et al. 2021, Qasmi et al.
(in rev). The observational constraint is a global constraint on the GMST
but it allows to constraint the regional climate warming. The 90% interval
of future warming plausible range is [2.3 ; 3.3]degC The 50% interval of
future warming plausible range is [2.5 ; 3.0]degC The best estimate is a
warming of 2.8 degC Category definition (new since 5 jan 2022): Categories
1 and 5 are considered as implausible by S. Qasmi. Category 2, 3, 4 are
plausible. Class 1 is below the Q5 of the constrained range. Class 2 is
between Q5 and Q25 Class 3 is between Q25 and Q75 Class 4 is between Q75
and Q95 Class 5 is above Q95
Located in CMIP6_studies/AtlasIPCC.yaml
['Pers. Comm.', 'Jesus Fernandez']
- key: Atlas Dtas NEU DJF
doi: ['Pers. Comm.', 'Jesus Fernandez']
type: future_spread
spatial_scope: NEU
temporal_scope: DJF
data_source: author_adapted
metric:
name: delta_tas
long_name: Near surface temperature delta change 2071-2100 w.r.t. 1981-2010
units: K
variables: tas
comment:
Data derived from https://github.com/IPCC-WG1/Atlas/tree/devel/datasets-
aggregated-regionally using the tas_landsea dataset.
period:
reference: 1981-2010
target: 2071-2100
Located in CMIP6_studies/AtlasIPCC.yaml
['Pers. Comm.', 'Jesus Fernandez']
- key: Atlas Dtas NEU JJA
doi: ['Pers. Comm.', 'Jesus Fernandez']
type: future_spread
spatial_scope: NEU
temporal_scope: JJA
data_source: author_adapted
metric:
name: delta_tas
long_name: Near surface temperature delta change 2071-2100 w.r.t. 1981-2010
units: K
variables: tas
comment:
Data derived from https://github.com/IPCC-WG1/Atlas/tree/devel/datasets-
aggregated-regionally using the tas_landsea dataset.
period:
reference: 1981-2010
target: 2071-2100
Located in CMIP6_studies/AtlasIPCC.yaml
['Pers. Comm.', 'Jesus Fernandez']
- key: Atlas Dtas WCE JJA
doi: ['Pers. Comm.', 'Jesus Fernandez']
type: future_spread
spatial_scope: WCE
temporal_scope: JJA
data_source: author_adapted
metric:
name: delta_tas
long_name: Near surface temperature delta change 2071-2100 w.r.t. 1981-2010
units: K
variables: tas
comment:
Data derived from https://github.com/IPCC-WG1/Atlas/tree/devel/datasets-
aggregated-regionally using the tas_landsea dataset.
period:
reference: 1981-2010
target: 2071-2100
Located in CMIP6_studies/AtlasIPCC.yaml
['Pers. Comm.', 'Jesus Fernandez']
- key: Atlas Dtas MED JJA
doi: ['Pers. Comm.', 'Jesus Fernandez']
type: future_spread
spatial_scope: MED
temporal_scope: JJA
data_source: author_adapted
metric:
name: delta_tas
long_name: Near surface temperature delta change 2071-2100 w.r.t. 1981-2010
units: K
variables: tas
comment:
Data derived from https://github.com/IPCC-WG1/Atlas/tree/devel/datasets-
aggregated-regionally using the tas_landsea dataset.
period:
reference: 1981-2010
target: 2071-2100
Located in CMIP6_studies/AtlasIPCC.yaml
['Pers. Comm.', 'Jesus Fernandez']
- key: Atlas Dpr MED JJA
doi: ['Pers. Comm.', 'Jesus Fernandez']
type: future_spread
spatial_scope: MED
temporal_scope: JJA
data_source: author_adapted
metric:
name: delta_pr
long_name: Precipitation relative delta change 2071-2100 w.r.t. 1981-2010
units: percent
variables: pr
comment:
Data derived from https://github.com/IPCC-WG1/Atlas/tree/devel/datasets-
aggregated-regionally using the pr_land dataset.
period:
reference: 1981-2010
target: 2071-2100
Located in CMIP6_studies/Sevault.yaml
['pers_comm', 'F. Sevault']
- key: Sevault MED SST warming
doi: ['pers_comm', 'F. Sevault']
type: future_spread
spatial_scope: MED
temporal_scope: Annual
data_source: author
metric:
name: deltasst
long_name: Future sea surface temperature change
units: K
variables: sst
comment:
For the Future spread, we simply compute the basin-averaged and temporal-
averaged climate change response (annual mean) over the Mediterranean Sea
(no Black Sea) for the period 2070-2099 wrt the present-climate 1985-2014
period for the SSP585 scenario.
classes:
- limits: [2, 3, 4, 10]
labels: ['weak', 'medium', 'strong']
source: eurocordex_gcm_selection_team
comment:
The limits of the warming level categories (weak, medium, strong) are
arbitrary.
Located in CMIP6_studies/Bru20.yaml
Lukas Brunner et al. (2020) Reduced global warming from CMIP6 projections when weighting models by performance and independence, https://doi.org/10.5194%2Fesd-11-995-2020
- key: Bru20 mfamily
doi: 10.5194/esd-11-995-2020
type: other
spatial_scope: Global
data_source: reference
metric:
name: mfamily
long_name: None
units: categorical
variables: ['tas', 'psl']
comment:
From Figure 5. Model family tree for all 33 CMIP6 models, similar to
Knutti et al. (2013). Models branching further to the left are more
dependent, and models branching further to the right are more independent.
The analysis is based on global, horizontally resolved tasCLIM and pslCLIM
in the period from 1980 to 2014. Labels with the same color indicate
models with obvious dependencies, such as shared components or the same
origin, whereas models with no clear dependencies are labeled in black.
Using this figure 5, we have put in the same family, the GCM lines that
merge before the dashed line, that correponds to the independence shape
parameter. We have given family names only to families with at least 2
members. Note that we are not using the Figure 5 color code to determine
the family. This leads to some surprise such as Strangely, ACCESS-CM2 is
in the same family as UKESM and HadGEM3, but not ACCESS-ESM1-5. This was
confirmed as ok by the Australian group. Also the two FGOALS model are not
belonging to the same family with this criteria.
period:
reference: 1980-2014
Located in CMIP6_studies/Aerosol.yaml
['pers. comm.', 'Jesus Fernandez']
- key: Aer. species
doi: ['pers. comm.', 'Jesus Fernandez']
type: other
spatial_scope: special
temporal_scope: Annual
data_source: author
metric:
name: aer_species
long_name: Aerosol species for which AOD available at ESGF
units: categorical
variables: od550bb, od550bc, od550dust, od550no3, od550oa, od550so4, od550ss, od550so4so, aerasymbnd, aeroptbnd, aerssabnd
comment:
Data extracted from ESGF using https://github.com/jesusff/cmip6-for-
cordex/blob/main/util/aerosol_species.py which feeds from
https://github.com/jesusff/cmip6-for-cordex/blob/main/CMIP6_for_CORDEX.py
Also, some pers. comm. for certain models that do not provide AOD by
aerosol species. S. Yang for the EC-Earth consortium models.
best: bb, bc, dust, no3, oa, so4, ss, so4so, aerasymbnd, aeroptbnd, aerssabnd
Located in CMIP6_studies/Bra21.yaml
Swen Brands et al. (2021) A circulation-based performance atlas of the CMIP5 and 6 models, https://doi.org/10.5194%2Fgmd-2020-418
- key: Bra21 complexity
doi: 10.5194/gmd-2020-418
type: other
spatial_scope: special
temporal_scope: Annual
data_source: reference
metric:
name: complexity
long_name: Complexity of model components
units: categorical
variables: []
comment:
Model complexity from Table 1 is coded with ternary values 0 - not
considered 1 - prescribed 2 - interactive component in the following
order Atm-Lnd-Ocn-SI-Veg-Tbgc-Aer-Chem-Obgc-Gla
plausible_values:
- min: 2222000000
max: 2222222222
source: eurocordex_gcm_selection_team
comment:
At least coupled Atm-Lnd-Ocn-SI with some form of aerosol consideration
Located in CMIP6_studies/Calendar.yaml
['pers. comm.', 'Andreas Dobler']
- key: Calendar
doi: ['pers. comm.', 'Andreas Dobler']
type: other
spatial_scope: special
temporal_scope: Annual
data_source: author
metric:
name: calendar
long_name: Model calendar
units: categorical
variables: None
comment:
Data extracted from ESGF
Located in CMIP6_studies/Resolution.yaml
None
- key: atm. res. km
doi: None
type: other
spatial_scope: special
temporal_scope: Annual
data_source: author
metric:
name: resolution
long_name: Nominal resolution of the atmospheric component
units: km
variables: None
comment:
Data extracted from the CMIP github https://github.com/WCRP-
CMIP/CMIP6_CVs/blob/master/CMIP6_source_id.json using
util/resolution_to_yaml.py Manually edited to include appropriate runs.
best: 0
plausible_values:
- min: 0
max: 300
source: eurocordex_gcm_selection_team
comment:
Test value