In [ ]:
Copied!
# Uncomment to install climdata in Google Colab or other environments
# !pip install climdata
# Uncomment to install climdata in Google Colab or other environments
# !pip install climdata
CMIP–Observations Comparison (Tutorial)¶
Purpose: Compare CMIP model output with local observations at a single point. This notebook demonstrates extracting observational and CMIP data using climdata, computing standard performance metrics (RMSE, Pearson correlation, MAPE, MNB), and producing visual summaries (boxplots + time series + portrait plot).
Prerequisites:
climdatapackage- Additional python packages:
cftime,hvplot,holoviews,bokeh,matplotlib,numpy
How to run:
- Run cells top-to-bottom in order. Edit the
overridesblocks to changedataset,lat,lon,time_range,variables, and CMIPsource_idlist. - Outputs: an interactive HTML per variable (e.g.
distribution_tas.html) and static portrait plot shown inline.
Notes & tips:
- The metrics function supports optional time resampling (e.g.,
time_resample='MS'for monthly means).
In [1]:
Copied!
import climdata
import pandas as pd
import xarray as xr
import cftime
import logging
logging.basicConfig(
level=logging.INFO,
format="%(levelname)s | %(message)s",
force=True,
)
extractor = climdata.ClimData()
extractor_CMIP = climdata.CMIP(extractor.cfg)
print(extractor_CMIP.get_experiment_ids())
print(extractor_CMIP.get_source_ids('historical'))
print(extractor_CMIP.get_variables(experiment_id='historical',source_id=['GISS-E2-1-H','CanESM5','GFDL-ESM4']))
import climdata
import pandas as pd
import xarray as xr
import cftime
import logging
logging.basicConfig(
level=logging.INFO,
format="%(levelname)s | %(message)s",
force=True,
)
extractor = climdata.ClimData()
extractor_CMIP = climdata.CMIP(extractor.cfg)
print(extractor_CMIP.get_experiment_ids())
print(extractor_CMIP.get_source_ids('historical'))
print(extractor_CMIP.get_variables(experiment_id='historical',source_id=['GISS-E2-1-H','CanESM5','GFDL-ESM4']))
['historical', 'ssp119', 'ssp126', 'ssp245', 'ssp370', 'ssp434', 'ssp460', 'ssp585']
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
INFO | 65 models found for experiment 'historical'
['ACCESS-CM2', 'ACCESS-ESM1-5', 'AWI-CM-1-1-MR', 'AWI-ESM-1-1-LR', 'BCC-CSM2-MR', 'BCC-ESM1', 'CAMS-CSM1-0', 'CAS-ESM2-0', 'CESM2', 'CESM2-FV2', 'CESM2-WACCM', 'CESM2-WACCM-FV2', 'CIESM', 'CMCC-CM2-HR4', 'CMCC-CM2-SR5', 'CMCC-ESM2', 'CNRM-CM6-1', 'CNRM-CM6-1-HR', 'CNRM-ESM2-1', 'CanESM5', 'CanESM5-CanOE', 'E3SM-1-0', 'E3SM-1-1', 'E3SM-1-1-ECA', 'EC-Earth3', 'EC-Earth3-AerChem', 'EC-Earth3-CC', 'EC-Earth3-Veg', 'EC-Earth3-Veg-LR', 'EC-Earth3P-VHR', 'FGOALS-f3-L', 'FGOALS-g3', 'FIO-ESM-2-0', 'GFDL-CM4', 'GFDL-ESM4', 'GISS-E2-1-G', 'GISS-E2-1-G-CC', 'GISS-E2-1-H', 'GISS-E2-2-H', 'HadGEM3-GC31-LL', 'HadGEM3-GC31-MM', 'ICON-ESM-LR', 'IITM-ESM', 'INM-CM4-8', 'INM-CM5-0', 'IPSL-CM5A2-INCA', 'IPSL-CM6A-LR', 'IPSL-CM6A-LR-INCA', 'KACE-1-0-G', 'KIOST-ESM', 'MCM-UA-1-0', 'MIROC-ES2H', 'MIROC-ES2L', 'MIROC6', 'MPI-ESM-1-2-HAM', 'MPI-ESM1-2-HR', 'MPI-ESM1-2-LR', 'MRI-ESM2-0', 'NESM3', 'NorCPM1', 'NorESM2-LM', 'NorESM2-MM', 'SAM0-UNICON', 'TaiESM1', 'UKESM1-0-LL'] ['hurs', 'pr', 'sfcWind', 'tas', 'tasmax', 'tasmin']
Setup & imports 🔧¶
The first cell imports core packages used in this tutorial. If any imports fail, install requirements via:
# plotting support
pip install hvplot holoviews bokeh
If you are working on a cluster, ensure the data_dir and any data access permissions are correct before running extraction cells.
Observational data extraction (step)¶
This cell demonstrates extracting local/observational data using ClimData with overrides.
dataset: choose dataset (e.g.,hyrasormswx)lat/lon: location for point extractiontime_range.start_dateand.end_date: extraction periodvariables: list of variable names to extract
After running this cell you should have:
ds_obs: an xarray Dataset with observation variablesdf_obs: a Pandas long-form DataFrame (time,variable,value,units,source)
The notebook converts time values to cftime.DatetimeNoLeap where needed to support calendar-aware comparisons.
In [2]:
Copied!
from climdata import ClimData
overrides = [
"dataset=hyras",
"lat=52",
"lon=13",
# "region=germany",
f"time_range.start_date=1989-01-01", # Start date for data extraction
f"time_range.end_date=2014-12-31", # End date for data extraction
"variables=[tasmin,tasmax,pr,tas]", # Variables to extract: min/max temp and precipitation
"data_dir=./data", # Local directory to store raw/intermediate files
]
extractor = ClimData(overrides=overrides)
ds_obs = extractor.extract()
df_obs = extractor.to_dataframe()
from climdata import ClimData
overrides = [
"dataset=hyras",
"lat=52",
"lon=13",
# "region=germany",
f"time_range.start_date=1989-01-01", # Start date for data extraction
f"time_range.end_date=2014-12-31", # End date for data extraction
"variables=[tasmin,tasmax,pr,tas]", # Variables to extract: min/max temp and precipitation
"data_dir=./data", # Local directory to store raw/intermediate files
]
extractor = ClimData(overrides=overrides)
ds_obs = extractor.extract()
df_obs = extractor.to_dataframe()
⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1989_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1989_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1990_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1990_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1991_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1991_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1992_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1992_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1993_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1993_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1994_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1994_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1995_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1995_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1996_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1996_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1997_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1997_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1998_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1998_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_1999_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_1999_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2000_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2000_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2001_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2001_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2002_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2002_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2003_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2003_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2004_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2004_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2005_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2005_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2006_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2006_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2007_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2007_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2008_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2008_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2009_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2009_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2010_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2010_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2011_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2011_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2012_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2012_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2013_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2013_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_min/tasmin_hyras_1_2014_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMIN/tasmin_hyras_1_2014_v6-0_de.nc
<frozen importlib._bootstrap>:241: RuntimeWarning: numpy.ndarray size changed, may indicate binary incompatibility. Expected 16 from C header, got 96 from PyObject
⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1989_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1989_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1990_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1990_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1991_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1991_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1992_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1992_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1993_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1993_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1994_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1994_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1995_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1995_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1996_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1996_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1997_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1997_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1998_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1998_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_1999_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_1999_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2000_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2000_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2001_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2001_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2002_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2002_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2003_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2003_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2004_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2004_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2005_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2005_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2006_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2006_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2007_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2007_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2008_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2008_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2009_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2009_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2010_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2010_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2011_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2011_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2012_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2012_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2013_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2013_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_max/tasmax_hyras_1_2014_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TASMAX/tasmax_hyras_1_2014_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1989_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1989_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1990_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1990_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1991_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1991_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1992_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1992_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1993_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1993_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1994_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1994_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1995_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1995_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1996_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1996_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1997_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1997_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1998_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1998_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_1999_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_1999_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2000_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2000_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2001_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2001_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2002_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2002_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2003_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2003_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2004_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2004_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2005_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2005_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2006_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2006_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2007_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2007_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2008_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2008_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2009_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2009_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2010_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2010_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2011_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2011_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2012_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2012_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2013_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2013_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/precipitation/pr_hyras_1_2014_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/PR/pr_hyras_1_2014_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1989_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1989_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1990_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1990_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1991_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1991_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1992_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1992_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1993_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1993_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1994_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1994_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1995_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1995_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1996_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1996_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1997_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1997_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1998_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1998_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_1999_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_1999_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2000_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2000_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2001_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2001_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2002_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2002_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2003_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2003_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2004_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2004_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2005_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2005_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2006_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2006_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2007_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2007_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2008_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2008_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2009_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2009_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2010_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2010_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2011_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2011_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2012_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2012_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2013_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2013_v6-0_de.nc ⬇️ Checking: https://opendata.dwd.de/climate_environment/CDC/grids_germany/daily/hyras_de/air_temperature_mean/tas_hyras_1_2014_v6-0_de.nc ✔️ Exists locally: /beegfs/muduchuru/data/hyras/TAS/tas_hyras_1_2014_v6-0_de.nc
CMIP models extraction (step)¶
This loop extracts CMIP model output for a list of source_id values and concatenates models along source_id:
dataset=cmip, chooseexperiment_id(e.g.,historical)source_id: set a single model name per loop iteration (e.g.,CanESM5)table_id: typicallyAmonfor monthly output- Result variables stored in
ds_cmip(xarray, concatenated) anddf_cmip(Pandas long-form concatenated DataFrame)
Tip: narrow time_range to speed up downloads and processing during testing.
In [ ]:
Copied!
ds_cmip = []
df_cmip = []
import cftime
import xarray as xr
for source_id in ['ACCESS-CM2','CanESM5','GFDL-ESM4']:
overrides = [
"dataset=cmip", # Choose the MSWX dataset for extraction
"lat=52",
"lon=13",
f"time_range.start_date=1989-01-01", # Start date for data extraction
f"time_range.end_date=2014-12-31", # End date for data extraction
f"experiment_id=historical",
f"source_id='{source_id}'",
f"table_id=Amon",
"variables=['tasmin','tasmax','pr','tas']", # Variables to extract: min/max temp and precipitation
"data_dir=./data", # Local directory to store raw/intermediate files
]
extractor = ClimData(overrides=overrides)
ds = extractor.extract()
df_cmip.append(extractor.to_dataframe())
df_cmip = pd.concat(df_cmip,axis=0)
ds_cmip = []
df_cmip = []
import cftime
import xarray as xr
for source_id in ['ACCESS-CM2','CanESM5','GFDL-ESM4']:
overrides = [
"dataset=cmip", # Choose the MSWX dataset for extraction
"lat=52",
"lon=13",
f"time_range.start_date=1989-01-01", # Start date for data extraction
f"time_range.end_date=2014-12-31", # End date for data extraction
f"experiment_id=historical",
f"source_id='{source_id}'",
f"table_id=Amon",
"variables=['tasmin','tasmax','pr','tas']", # Variables to extract: min/max temp and precipitation
"data_dir=./data", # Local directory to store raw/intermediate files
]
extractor = ClimData(overrides=overrides)
ds = extractor.extract()
df_cmip.append(extractor.to_dataframe())
df_cmip = pd.concat(df_cmip,axis=0)
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
/home/muduchuru/miniforge3/envs/sdba/lib/python3.10/site-packages/intake_esm/core.py:475: PydanticDeprecatedSince20: The `dict` method is deprecated; use `model_dump` instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
cat = self.__class__({'esmcat': self.esmcat.dict(), 'df': esmcat_results})
Compute CMIP-vs-Observations metrics 🧮¶
This cell defines compute_cmip_metrics(df_cmip, df_obs, variables=None, time_resample=None) which:
- Aggregates spatially (mean over lat/lon), optionally resamples time (e.g., monthly
'MS'), and computes:- RMSE (Root Mean Square Error)
- CORR (Pearson correlation)
- MAPE (Mean Absolute Percentage Error)
- MNB (Mean Normalized Bias)
Output: a DataFrame metrics_df with one row per (variable,source_id) and columns like RMSE, CORR, MAPE, MNB, Model Mean, Observation Mean.
Notes:
- The function converts
cftimedates topandas.Timestampwhere possible for reliable joins and resampling. - Missing or misaligned dates between model and observations are handled by an inner merge on
timein this implementation.
In [4]:
Copied!
import pandas as pd
import numpy as np
def cftime_to_datetime(df, time_col='time'):
"""
Convert cftime.Datetime* to pd.Timestamp where possible.
"""
import cftime
times = df[time_col].values
new_times = []
for t in times:
if isinstance(t, (cftime.DatetimeNoLeap, cftime.DatetimeJulian, cftime.DatetimeProlepticGregorian)):
# convert to pandas Timestamp; keep year, month, day
# ignore hour/min/sec for safety
new_times.append(pd.Timestamp(f"{t.year:04d}-{t.month:02d}-{t.day:02d}"))
else:
new_times.append(pd.Timestamp(t))
df[time_col] = new_times
return df
def compute_cmip_metrics(df_cmip, df_obs, variables=None, time_resample=None):
"""
Compute RMSE, CORR, MAPE, MNB for CMIP models vs observations with optional time resampling.
Parameters
----------
df_cmip : pd.DataFrame
Long-form DataFrame with columns ['time','lat','lon','variable','value','units','source_id','source']
df_obs : pd.DataFrame
Long-form DataFrame with columns ['time','lat','lon','variable','value','units','source']
variables : list or None
List of variable names to compute metrics. If None, all variables in df_obs are used.
time_resample : str or None
Optional pandas resample frequency: 'D' (daily), 'MS' (month start), 'YS' (year start).
If None, no temporal aggregation is performed.
Returns
-------
pd.DataFrame
Metrics per variable and source_id with columns ['variable','source_id','RMSE','CORR','MAPE','MNB']
"""
if variables is None:
variables = df_obs['variable'].unique()
# Ensure time is datetime
df_cmip = cftime_to_datetime(df_cmip, time_col='time')
df_obs = cftime_to_datetime(df_obs, time_col='time')
# 1️⃣ Aggregate spatially: mean over lat/lon
df_cmip_mean = (
df_cmip.groupby(['time','variable','source_id'])['value']
.mean()
.reset_index()
)
df_obs_mean = (
df_obs.groupby(['time','variable'])['value']
.mean()
.reset_index()
)
# 2️⃣ Resample in time if requested
if time_resample is not None:
# CMIP
df_cmip_mean = (
df_cmip_mean.set_index('time')
.groupby(['variable','source_id'])
.resample(time_resample)['value']
.mean()
.reset_index()
)
# Observations
df_obs_mean = (
df_obs_mean.set_index('time')
.groupby('variable')
.resample(time_resample)['value']
.mean()
.reset_index()
)
records = []
# 3️⃣ Loop over variables and source_ids
for var in variables:
o = df_obs_mean[df_obs_mean['variable']==var]
for src in df_cmip_mean['source_id'].unique():
m = df_cmip_mean[(df_cmip_mean['variable']==var) & (df_cmip_mean['source_id']==src)]
# Merge on time
df_combined = pd.merge(
o[['time','value']],
m[['time','value']],
on='time',
suffixes=('_obs', f'_{src}')
)
obs_vals = df_combined[f'value_obs'].values
mod_vals = df_combined[f'value_{src}'].values
# Small jitter to avoid divide-by-zero in MAPE
epsilon = obs_vals.std() * 0.001
obs_safe = obs_vals + epsilon
rmse = np.sqrt(np.mean((mod_vals - obs_vals)**2))
corr = np.corrcoef(obs_vals, mod_vals)[0,1]
mape = np.mean(np.abs((mod_vals - obs_safe)/obs_safe)) * 100
mnb = np.mean(mod_vals - obs_vals) / np.mean(obs_safe) * 100 # mean normalized bias
records.append({
'variable': var,
'source_id': src,
'Model Mean': mod_vals.mean(),
'Observation Mean': obs_vals.mean(),
'RMSE': rmse,
'CORR': corr,
'MAPE': mape,
'MNB': mnb
})
df_metrics = pd.DataFrame(records)
return df_metrics
import pandas as pd
import numpy as np
def cftime_to_datetime(df, time_col='time'):
"""
Convert cftime.Datetime* to pd.Timestamp where possible.
"""
import cftime
times = df[time_col].values
new_times = []
for t in times:
if isinstance(t, (cftime.DatetimeNoLeap, cftime.DatetimeJulian, cftime.DatetimeProlepticGregorian)):
# convert to pandas Timestamp; keep year, month, day
# ignore hour/min/sec for safety
new_times.append(pd.Timestamp(f"{t.year:04d}-{t.month:02d}-{t.day:02d}"))
else:
new_times.append(pd.Timestamp(t))
df[time_col] = new_times
return df
def compute_cmip_metrics(df_cmip, df_obs, variables=None, time_resample=None):
"""
Compute RMSE, CORR, MAPE, MNB for CMIP models vs observations with optional time resampling.
Parameters
----------
df_cmip : pd.DataFrame
Long-form DataFrame with columns ['time','lat','lon','variable','value','units','source_id','source']
df_obs : pd.DataFrame
Long-form DataFrame with columns ['time','lat','lon','variable','value','units','source']
variables : list or None
List of variable names to compute metrics. If None, all variables in df_obs are used.
time_resample : str or None
Optional pandas resample frequency: 'D' (daily), 'MS' (month start), 'YS' (year start).
If None, no temporal aggregation is performed.
Returns
-------
pd.DataFrame
Metrics per variable and source_id with columns ['variable','source_id','RMSE','CORR','MAPE','MNB']
"""
if variables is None:
variables = df_obs['variable'].unique()
# Ensure time is datetime
df_cmip = cftime_to_datetime(df_cmip, time_col='time')
df_obs = cftime_to_datetime(df_obs, time_col='time')
# 1️⃣ Aggregate spatially: mean over lat/lon
df_cmip_mean = (
df_cmip.groupby(['time','variable','source_id'])['value']
.mean()
.reset_index()
)
df_obs_mean = (
df_obs.groupby(['time','variable'])['value']
.mean()
.reset_index()
)
# 2️⃣ Resample in time if requested
if time_resample is not None:
# CMIP
df_cmip_mean = (
df_cmip_mean.set_index('time')
.groupby(['variable','source_id'])
.resample(time_resample)['value']
.mean()
.reset_index()
)
# Observations
df_obs_mean = (
df_obs_mean.set_index('time')
.groupby('variable')
.resample(time_resample)['value']
.mean()
.reset_index()
)
records = []
# 3️⃣ Loop over variables and source_ids
for var in variables:
o = df_obs_mean[df_obs_mean['variable']==var]
for src in df_cmip_mean['source_id'].unique():
m = df_cmip_mean[(df_cmip_mean['variable']==var) & (df_cmip_mean['source_id']==src)]
# Merge on time
df_combined = pd.merge(
o[['time','value']],
m[['time','value']],
on='time',
suffixes=('_obs', f'_{src}')
)
obs_vals = df_combined[f'value_obs'].values
mod_vals = df_combined[f'value_{src}'].values
# Small jitter to avoid divide-by-zero in MAPE
epsilon = obs_vals.std() * 0.001
obs_safe = obs_vals + epsilon
rmse = np.sqrt(np.mean((mod_vals - obs_vals)**2))
corr = np.corrcoef(obs_vals, mod_vals)[0,1]
mape = np.mean(np.abs((mod_vals - obs_safe)/obs_safe)) * 100
mnb = np.mean(mod_vals - obs_vals) / np.mean(obs_safe) * 100 # mean normalized bias
records.append({
'variable': var,
'source_id': src,
'Model Mean': mod_vals.mean(),
'Observation Mean': obs_vals.mean(),
'RMSE': rmse,
'CORR': corr,
'MAPE': mape,
'MNB': mnb
})
df_metrics = pd.DataFrame(records)
return df_metrics
Visualization & outputs 📊¶
This section creates two linked visual summaries per selected variable:
- Top panel: a box plot of model value distributions by
source_idcombined with point markers showing model mean and hover details (RMSE, CORR, MAPE, MNB, Obs mean). - Bottom panel: time series of each model (by
source_id) with observations overplotted (if available).
Files saved:
distribution_<variable>.html— interactive Holoviews/HvPlot HTML for the chosen variable.- Inline portrait plot created with Matplotlib shows per-model RMSE (upper-left triangle) and correlation (lower-right triangle).
Tip: Change var to iterate through variables and call hv.save() for each variable to persist results to disk.
In [9]:
Copied!
from bokeh.models import HoverTool
import hvplot.pandas
var = 'pr'
# Rename columns to remove spaces
metrics_df = compute_cmip_metrics(df_cmip, df_obs, time_resample='MS')
metrics_df = metrics_df.rename(columns={
'Model Mean': 'Model_Mean',
'Observation Mean': 'Obs_Mean'
})
# Update your HoverTool to match
custom_hover = HoverTool(
tooltips=[
("Source ID", "@source_id"),
("Model Mean", "@Model_Mean"),
("Obs Mean", "@Obs_Mean"), # No braces needed now
("RMSE", "@RMSE"),
("Pearson r", "@CORR"),
],
mode='vline'
)
# Update your scatter call
points = metrics_df[metrics_df['variable']==var].hvplot.scatter(
x='source_id',
y='Model_Mean',
hover_cols=['RMSE','CORR', 'MAPE', 'MNB', 'Obs_Mean'], # Use new name
size=15,
tools=[custom_hover]
)
# 3. Create the box plot (disable its default hover so they don't clash)
box_plot = df_cmip[df_cmip['variable']==var].hvplot.box(
y='value',
by=['source_id'],
ylabel=var,
hover=False # We want the 'points' hover to take the lead
)
top_plot = (box_plot * points).opts(
title=f"Distribution: {var}",
width=800,
height=400
)
# 2. New Time Series Plot (Bottom Panel)
# We filter df_cmip for the variable, then plot time vs value
ts_plot = df_cmip[df_cmip['variable'] == var].hvplot.line(
x='time', # Ensure this matches your date/time column name
y='value',
by='source_id',
ylabel=var,
title=f"{var} Time Series by Model",
width=800,
height=300,
alpha=0.6 # Slight transparency helps if lines overlap
)
# Optional: Add the observation line to the time series for reference
if 'df_obs' in locals():
obs_ts = df_obs[df_obs['variable'] == var].hvplot.line(
x='time',
y='value',
color='black',
line_width=2,
label='Observations'
)
ts_plot = ts_plot * obs_ts
# 3. Combine and stack vertically
# The '+' operator creates a layout, '.cols(1)' forces the vertical stack
layout = (top_plot + ts_plot).cols(1)
layout
from bokeh.models import HoverTool
import hvplot.pandas
var = 'pr'
# Rename columns to remove spaces
metrics_df = compute_cmip_metrics(df_cmip, df_obs, time_resample='MS')
metrics_df = metrics_df.rename(columns={
'Model Mean': 'Model_Mean',
'Observation Mean': 'Obs_Mean'
})
# Update your HoverTool to match
custom_hover = HoverTool(
tooltips=[
("Source ID", "@source_id"),
("Model Mean", "@Model_Mean"),
("Obs Mean", "@Obs_Mean"), # No braces needed now
("RMSE", "@RMSE"),
("Pearson r", "@CORR"),
],
mode='vline'
)
# Update your scatter call
points = metrics_df[metrics_df['variable']==var].hvplot.scatter(
x='source_id',
y='Model_Mean',
hover_cols=['RMSE','CORR', 'MAPE', 'MNB', 'Obs_Mean'], # Use new name
size=15,
tools=[custom_hover]
)
# 3. Create the box plot (disable its default hover so they don't clash)
box_plot = df_cmip[df_cmip['variable']==var].hvplot.box(
y='value',
by=['source_id'],
ylabel=var,
hover=False # We want the 'points' hover to take the lead
)
top_plot = (box_plot * points).opts(
title=f"Distribution: {var}",
width=800,
height=400
)
# 2. New Time Series Plot (Bottom Panel)
# We filter df_cmip for the variable, then plot time vs value
ts_plot = df_cmip[df_cmip['variable'] == var].hvplot.line(
x='time', # Ensure this matches your date/time column name
y='value',
by='source_id',
ylabel=var,
title=f"{var} Time Series by Model",
width=800,
height=300,
alpha=0.6 # Slight transparency helps if lines overlap
)
# Optional: Add the observation line to the time series for reference
if 'df_obs' in locals():
obs_ts = df_obs[df_obs['variable'] == var].hvplot.line(
x='time',
y='value',
color='black',
line_width=2,
label='Observations'
)
ts_plot = ts_plot * obs_ts
# 3. Combine and stack vertically
# The '+' operator creates a layout, '.cols(1)' forces the vertical stack
layout = (top_plot + ts_plot).cols(1)
layout
Out[9]:
In [6]:
Copied!
import holoviews as hv
hv.save(layout, f'distribution_{var}.html')
import holoviews as hv
hv.save(layout, f'distribution_{var}.html')
In [8]:
Copied!
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.patches import Polygon
from matplotlib.colors import Normalize
import matplotlib.cm as cm
rmse = metrics_df.pivot(index="source_id", columns="variable", values="RMSE")
corr = metrics_df.pivot(index="source_id", columns="variable", values="CORR")
models = rmse.index.tolist()
variables = rmse.columns.tolist()
# -------------------------------------------------------
# Colormap and normalization
# -------------------------------------------------------
cmap_rmse = plt.cm.jet
rmse_norm = Normalize(vmin=0, vmax=10)
cmap_corr = plt.cm.Blues
corr_norm = Normalize(vmin=-1, vmax=1)
# -------------------------------------------------------
# Plot
# -------------------------------------------------------
fig, ax = plt.subplots(figsize=(12, 8))
for i, model in enumerate(models):
for j, var in enumerate(variables):
x0, x1 = j, j + 1
y0, y1 = i, i + 1
# Upper-left triangle (RMSE)
tri_rmse = Polygon(
[[x0, y1], [x0, y0], [x1, y1]],
closed=True,
facecolor=cmap_rmse(rmse_norm(rmse.loc[model, var])),
edgecolor="none"
)
ax.add_patch(tri_rmse)
# Lower-right triangle (CORR)
tri_corr = Polygon(
[[x1, y0], [x0, y0], [x1, y1]],
closed=True,
facecolor=cmap_corr(corr_norm(corr.loc[model, var])),
edgecolor="none"
)
ax.add_patch(tri_corr)
# -------------------------------------------------------
# Axes formatting
# -------------------------------------------------------
ax.set_xticks(np.arange(len(variables)) + 0.5)
ax.set_xticklabels(variables, fontsize=14)
ax.set_yticks(np.arange(len(models)) + 0.5)
ax.set_yticklabels(models, fontsize=14)
ax.set_xlim(0, len(variables))
ax.set_ylim(0, len(models))
ax.invert_yaxis()
ax.set_aspect("equal")
ax.set_xlabel("Variables", fontsize=16)
ax.set_ylabel("", fontsize=16)
ax.set_title("Portrait Plot (RMSE & CORR)", fontsize=18)
ax.tick_params(axis="both", which="major", labelsize=14)
# -------------------------------------------------------
# Colorbars (larger, better spacing)
# -------------------------------------------------------
cbar_rmse = fig.colorbar(
cm.ScalarMappable(norm=rmse_norm, cmap=cmap_rmse),
ax=ax,
fraction=0.05,
pad=0.02,
shrink=0.8,
aspect=20
)
cbar_rmse.set_label("Root Mean Sq Error", fontsize=14)
cbar_rmse.ax.tick_params(labelsize=12)
cbar_corr = fig.colorbar(
cm.ScalarMappable(norm=corr_norm, cmap=cmap_corr),
ax=ax,
fraction=0.15,
pad=0.02,
shrink=0.8,
aspect=20
)
cbar_corr.set_label("Pearson Correlation Coefficient", fontsize=14)
cbar_corr.ax.tick_params(labelsize=12)
plt.tight_layout()
plt.show()
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.patches import Polygon
from matplotlib.colors import Normalize
import matplotlib.cm as cm
rmse = metrics_df.pivot(index="source_id", columns="variable", values="RMSE")
corr = metrics_df.pivot(index="source_id", columns="variable", values="CORR")
models = rmse.index.tolist()
variables = rmse.columns.tolist()
# -------------------------------------------------------
# Colormap and normalization
# -------------------------------------------------------
cmap_rmse = plt.cm.jet
rmse_norm = Normalize(vmin=0, vmax=10)
cmap_corr = plt.cm.Blues
corr_norm = Normalize(vmin=-1, vmax=1)
# -------------------------------------------------------
# Plot
# -------------------------------------------------------
fig, ax = plt.subplots(figsize=(12, 8))
for i, model in enumerate(models):
for j, var in enumerate(variables):
x0, x1 = j, j + 1
y0, y1 = i, i + 1
# Upper-left triangle (RMSE)
tri_rmse = Polygon(
[[x0, y1], [x0, y0], [x1, y1]],
closed=True,
facecolor=cmap_rmse(rmse_norm(rmse.loc[model, var])),
edgecolor="none"
)
ax.add_patch(tri_rmse)
# Lower-right triangle (CORR)
tri_corr = Polygon(
[[x1, y0], [x0, y0], [x1, y1]],
closed=True,
facecolor=cmap_corr(corr_norm(corr.loc[model, var])),
edgecolor="none"
)
ax.add_patch(tri_corr)
# -------------------------------------------------------
# Axes formatting
# -------------------------------------------------------
ax.set_xticks(np.arange(len(variables)) + 0.5)
ax.set_xticklabels(variables, fontsize=14)
ax.set_yticks(np.arange(len(models)) + 0.5)
ax.set_yticklabels(models, fontsize=14)
ax.set_xlim(0, len(variables))
ax.set_ylim(0, len(models))
ax.invert_yaxis()
ax.set_aspect("equal")
ax.set_xlabel("Variables", fontsize=16)
ax.set_ylabel("", fontsize=16)
ax.set_title("Portrait Plot (RMSE & CORR)", fontsize=18)
ax.tick_params(axis="both", which="major", labelsize=14)
# -------------------------------------------------------
# Colorbars (larger, better spacing)
# -------------------------------------------------------
cbar_rmse = fig.colorbar(
cm.ScalarMappable(norm=rmse_norm, cmap=cmap_rmse),
ax=ax,
fraction=0.05,
pad=0.02,
shrink=0.8,
aspect=20
)
cbar_rmse.set_label("Root Mean Sq Error", fontsize=14)
cbar_rmse.ax.tick_params(labelsize=12)
cbar_corr = fig.colorbar(
cm.ScalarMappable(norm=corr_norm, cmap=cmap_corr),
ax=ax,
fraction=0.15,
pad=0.02,
shrink=0.8,
aspect=20
)
cbar_corr.set_label("Pearson Correlation Coefficient", fontsize=14)
cbar_corr.ax.tick_params(labelsize=12)
plt.tight_layout()
plt.show()
In [ ]:
Copied!