Climate data recipes
This section provides examples for climate data downloading and simple pre-processing.
The Copernicus climate data store section contains a list of recommended data sets to run the Virtual Ecosystem and describes how to download climate data from the Copernicus Climate Data Store (CDS) and basic pre-processing options using the CDS toolbox.
Note
At present, the pre-processing does not include scaling or topographic adjustment.
Metadata:
Muñoz-Sabater,J. et al: ERA5-Land: A state-of-the-art global reanalysis dataset for land applications, Earth Syst. Sci. Data,13, 4349-4383, 2021. https://doi.org/10.5194/essd-13-4349-2021
Product type: Monthly averaged reanalysis
Variable: 2m dewpoint temperature, 2m temperature, Surface pressure, Total precipitation
Year: 2013, 2014
Month: January, February, March, April, May, June, July, August, September, October, November, December
Time: 00:00
Sub-region extraction: North 6°, West 116°, South 4°, East 118°
Format: NetCDF3
We have used a simple recipe from this data source to create the climate data used in the example data. The code in that recipe is shown below:
climate_example_data.py
"""Simple climate data pre-processing example for `ve_run` example data.
This section illustrates how to perform simple manipulations to adjust ERA5-Land data to
use in the Virtual Ecosystem. This includes reading climate data from netcdf,
converting the data into an input format that is suitable for the abiotic module (e.g.
Kelvin to Celsius conversion), adding further required variables, and writing the output
in a new netcdf file. This does not include spatially interpolating coarser resolution
climate data and including the effects of local topography.
Input file: ERA5_land.nc
Metadata:
* Muñoz-Sabater,J. et al: ERA5-Land: A state-of-the-art global reanalysis dataset for
land applications, Earth Syst. Sci. Data,13, 4349-4383, 2021.
[https://doi.org/10.5194/essd-13-4349-2021](https://doi.org/10.5194/essd-13-4349-2021)
* Product type: Monthly averaged reanalysis
* Variable: 2m dewpoint temperature, 2m temperature, Surface pressure, Total
precipitation
* Year: 2013, 2014
* Month: January, February, March, April, May, June, July, August, September, October,
November, December
* Time: 00:00
* Sub-region extraction: North 6°, West 116°, South 4°, East 118°
* Format: NetCDF3
Once the new netcdf file is created, the final step is to add the grid information to
the grid config `TOML` to load this data correctly when setting up a Virtual Ecosystem
Simulation. Here, we can also add the 45 m offset to position the coordinated at the
centre of the grid cell.
[core.grid]
cell_nx = 9
cell_ny = 9
cell_area = 8100
xoff = -45.0
yoff = -45.0
"""
import numpy as np
import xarray as xr
from xarray import DataArray
from virtual_ecosystem.example_data.generation_scripts.common import (
time_index,
x_cell_ids,
y_cell_ids,
)
# 1. Load ERA5_Land data in low resolution
dataset = xr.open_dataset("../source/ERA5_land.nc")
# 2. Convert temperatures units
# The standard output unit of ERA5-Land temperatures is Kelvin which we need to convert
# to degree Celsius for the Virtual Ecosystem. This includes 2m air temperature and
# 2m dewpoint temperature which are used to calculate relative humidity in next step.
dataset["t2m_C"] = dataset["t2m"] - 273.15 # 2m air temperature
dataset["d2m_C"] = dataset["d2m"] - 273.15 # 2m dewpoint temperature
# 3. Calculate relative humidity
# Relative humidity (RH) is not a standard output from ERA5-Land but can be calculated
# from 2m dewpoint temperature (DPT) and 2m air temperature (T)
dataset["rh2m"] = 100.0 * (
np.exp(17.625 * dataset["d2m_C"] / (243.04 + dataset["d2m_C"]))
/ np.exp(17.625 * dataset["t2m_C"] / (243.04 + dataset["t2m_C"]))
)
# 4. Convert precipitation units
# The standard output unit for total precipitation in ERA5-Land is meters which we need
# to convert to millimeters. Further, the data represents mean daily accumulated
# precipitation for the 9x9km grid box, so the value has to be scaled to monthly (here
# 30 days). TODO handel daily inputs
dataset["tp_mm"] = dataset["tp"] * 1000 * 30
# 5. Convert surface pressure units
# The standard output unit for surface pressure in ERA5-Land is Pascal (Pa) which we
# need to convert to Kilopascal (kPa).
dataset["sp_kPa"] = dataset["sp"] / 1000
# 6. Clean dataset and rename variables
# In this step, we delete the initial temperature variables (K), precipitation (m), and
# surface pressure(Pa) and rename the remaining variables according to the Virtual
# Ecosystem naming convention.
dataset_cleaned = dataset.drop_vars(["d2m", "d2m_C", "t2m", "tp", "sp"])
dataset_renamed = dataset_cleaned.rename(
{
"sp_kPa": "atmospheric_pressure_ref",
"tp_mm": "precipitation",
"t2m_C": "air_temperature_ref",
"rh2m": "relative_humidity_ref",
}
)
# 7. Add further required variables
# In addition to the variables from the ERA5-Land datasset, a time series of atmospheric
# CO2 is needed. We add this here as a constant field. Mean annual temperature
# is calculated from the full time series of air temperatures; in the future, this
# should be done for each year.
dataset_renamed["atmospheric_co2_ref"] = DataArray(
np.full_like(dataset_renamed["air_temperature_ref"], 400),
dims=["time", "latitude", "longitude"],
)
dataset_renamed["wind_speed_ref"] = DataArray(
np.full_like(dataset_renamed["air_temperature_ref"], 0.1),
dims=["time", "latitude", "longitude"],
)
dataset_renamed["mean_annual_temperature"] = dataset_renamed[
"air_temperature_ref"
].mean(dim="time")
# 8. Change coordinates to x-y in meters
# The following code segment changes the coordinate names from `longitude/latitude` to
# `x/y` and the units from `minutes` to `meters`. The ERA5-Land coordinates are treated
# as the centre points of the grid cells which means that when setting up the grid, an
# offset of 4.5 km has to be added.
dataset_xy = (
dataset_renamed.rename_dims({"longitude": "x", "latitude": "y"})
.assign_coords({"x": np.arange(0, 180000, 9000), "y": np.arange(0, 180000, 9000)})
.drop({"longitude", "latitude"})
)
# 9. Scale to 90 m resolution
# The Virtual Ecosystem example data is run on a 90 x 90 m grid. This means that some
# form of spatial downscaling has to be applied to the dataset, for example by spatially
# interpolating coarser resolution climate data and including the effects of local
# topography. This is not yet implemented!
# For the purpose of a example data in the development stage, the coordinates can be
# overwritten to match the Virtual Ecosystem grid and we can select a smaller area.
# Note that the resulting dataset does no longer match a digital elevation model for the
# area!
dataset_xy_100 = (
dataset_renamed.rename_dims({"longitude": "x", "latitude": "y"})
.assign_coords({"x": np.arange(0, 1800, 90), "y": np.arange(0, 1800, 90)})
.drop({"longitude", "latitude"})
)
dataset_xy_example = dataset_xy_100.isel(x=x_cell_ids, y=y_cell_ids)
# 10. Add time_index
# At the moemnt, the example model iterates over time indices rather than real datetime.
# Therefore, we add a `time_index` coordinate to the dataset:
dataset_xy_timeindex = (
dataset_xy_example.rename_dims({"time": "time_index"})
.assign_coords({"time_index": time_index})
.drop("time")
)
# 11. Save netcdf
# Once we confirmed that our dataset is complete and our calculations are correct, we
# save it as a new netcdf file. This can then be fed into the code data loading system
# here {mod}`~virtual_ecosystem.core.data`.
dataset_xy_timeindex.to_netcdf("../data/example_climate_data.nc")
In more detail, that script carries out the main steps perfomed to create the following input variables for the Virtual Ecosystem:
air temperature, [C]
relative humidity, [-]
atmospheric pressure, [kPa]
precipitation, [mm month^-1]
atmospheric \(\ce{CO_{2}}\) concentration, [ppm]
mean annual temperature, [C]
Adjustment of units
The standard output unit of ERA5-Land temperatures is Kelvin which needs to be converted to degree Celsius for the Virtual Ecosystem. This includes 2m air temperature and 2m dewpoint temperature which are used to calculate relative humidity later. The standard output unit for total precipitation in ERA5-Land is meters which we need to convert to millimeters. Further, the data represents mean daily accumulated precipitation for the 9x9km grid box, so the value has to be scaled to monthly (here 30 days). The standard output unit for surface pressure in ERA5-Land is Pascal (Pa) which we need to convert to Kilopascal (kPa).
Addition of missing variables
In addition to the variables from the ERA5-Land data, a time series of atmospheric \(\ce{CO_{2}}\) is needed. We add this here as a constant field across all grid cells and vertical layers. Mean annual temperature is calculated from the full time series of air temperatures; in the future, this should be done for each year.
Relative humidity (RH) is also not a standard output from ERA5-Land but can be calculated from 2m dewpoint temperature (DPT) and 2m air temperature (T) as follows:
Matching Virtual Ecosystem grid and naming conventions
Once all input units are adjusted, the variables are re-named according to the Virtual
Ecosystem naming convention. The coordinate names have to be changed from
longitude/latitude to x/y and the units from minutes to meters. The ERA5-Land
coordinates are treated as the centre points of the grid cells which means that when
setting up the grid, an offset of 4.5 km has to be added.
Note
The example data is run using a 90 x 90 m grid. This means that some form of spatial downscaling has to be applied to the dataset, for example by spatially interpolating coarser resolution climate data and including the effects of local topography. This is not yet implemented!
For the purpose of a example data simulation in the development stage, the script curently selects a 9 by 9 sample of the grid and overwrites the coordinates to align to the example grid resolution. Note that the resulting dataset does no longer match a digital elevation model for the area!
At the moment, the dummy model iterates over time indices rather than real datetime.
Therefore, we add a time_index dimension and coordinate to the dataset.