"""Modifications to the downscaling by Fisher et al. (2016).
Changes from the Olsen-Randerson downscaling are primarily to prevent
discontinuities at month change.
I do not know whether the rolling windows used in Fisher et al. (2016)
were centered on the given time or ending on the given time. At
present, this code uses rolling windows ending on the given day,
because that is easy to get pandas to do.
"""
from . import NEP_TO_GPP_FACTOR, Q10, T0
INPUT_FREQUENCY = "1M"
"""The frequency at which the input data are given.
Used to ensure the downscaled fluxes match the predictions of the
original fluxes.
"""
[docs]def downscale_timeseries(flux_nee, temperature, par):
"""Downscale the columns of flux_nee.
The parts of the downscaled flux corresponding to the first and
last time periods in the original flux will be incomplete. It is
recommended to provide an extra time period on each end to avoid
this.
Parameters
----------
flux_nee : pd.DataFrame[N_large, M]
:abbr:`NEE (Net Ecosystem Exchange)`, at the large timesteps.
Must have datetime index. Positive indicates carbon is
entering the atmosphere. Units must have time in denominator.
temperature : pd.DataFrame[N, M]
Temperature at the small timesteps.
Must have datetime index with a set frequency
Unit is expected to be degrees Celsius.
par : pd.DataFrame[N, M]
:abbr:`PAR (Photosynthetically Active Radiation)` at the small
timesteps. Must be greather than or equal to zero. Must have
datetime index with a set frequency.
Returns
-------
flux_nee : pd.DataFrame[N, M]
The downscaled :abbr:`NEE (Net Ecosystem Exchange)`.
References
----------
Fisher, J. B., Sikka, M., Huntzinger, D. N., Schwalm, C., and Liu,
J., 2016: Technical note: 3-hourly temporal downscaling of monthly
global terrestrial biosphere model net ecosystem exchange,
*Biogeosciences*, vol. 13, no. 14, 4271--4277,
:doi:`10.5194/bg-13-4271-2016`.
"""
estimated_gpp = -NEP_TO_GPP_FACTOR * flux_nee
flux_gpp = downscale_gpp_timeseries(
estimated_gpp, par
)
flux_resp = downscale_resp_timeseries(
estimated_gpp + flux_nee, temperature
)
downscaled_nee = flux_resp - flux_gpp
# # This is where the NEE rebalancing would happen
# month_lengths = downscaled_nee.resample(INPUT_FREQUENCY).size()
# nee_left_out = (
# flux_nee.resample(INPUT_FREQUENCY).sum().multiply(
# month_lengths, axis=0
# ) -
# downscaled_nee.resample(INPUT_FREQUENCY).sum()
# )
# downscaled_nee = (
# downscaled_nee.resample(INPUT_FREQUENCY) +
# nee_left_out.divide(
# month_lengths, axis=0
# )
# ).transform(lambda x: x)
return downscaled_nee
[docs]def downscale_gpp_timeseries(flux_gpp, par):
"""Downscale the columns of flux_nee.
Parameters
----------
flux_gpp : pd.DataFrame[N_large, M]
:abbr:`GPP (Gross Primary Productivity)` at the larger
timesteps. Must have a datetime index. Units must have time
in the denominator.
par : pd.DataFrame[N, M]
:abbr:`PAR (Photosynthetically Active Radiation)` at the small
timesteps. Must be greather than or equal to zero. Must have
datetime index with a set frequency.
Returns
-------
flux_gpp : pd.DataFrame[N, M]
The :abbr:`GPP (Gross Primary Productivity)` downscaled to the
smaller time steps.
References
----------
Fisher, J. B., Sikka, M., Huntzinger, D. N., Schwalm, C., and Liu,
J., 2016: Technical note: 3-hourly temporal downscaling of monthly
global terrestrial biosphere model net ecosystem exchange,
*Biogeosciences*, vol. 13, no. 14, 4271--4277,
:doi:`10.5194/bg-13-4271-2016`.
"""
# This is mean over thirty days prior to the given day.
# I can't figure out how to get a centered window.
par_mean = par.rolling("30D").mean()
# Get the GPP timeseries to the same timestep as par
flux_gpp_baseline = flux_gpp.resample(
par.index.freq
).interpolate(method="time")
# This would be where I would deal with the first and last several
# timesteps.
return flux_gpp_baseline / par_mean * par
[docs]def downscale_resp_timeseries(flux_resp, temperature):
"""Downscale the columns of flux_resp.
Parameters
----------
flux_resp : pd.DataFrame[N_large, M]
Respiration fluxes at the larger timesteps.
Must have a datetime index.
Units must have time in denominator.
temperature : pd.DataFrame[N, M]
Temperature at the small timesteps.
Must have datetime index with a set frequency
Unit is expected to be degrees Celsius.
Returns
-------
flux_resp : pd.DataFrame[N, M]
The respiration fluxes downscaled to the smaller time steps.
References
----------
Fisher, J. B., Sikka, M., Huntzinger, D. N., Schwalm, C., and Liu,
J., 2016: Technical note: 3-hourly temporal downscaling of monthly
global terrestrial biosphere model net ecosystem exchange,
*Biogeosciences*, vol. 13, no. 14, 4271--4277,
:doi:`10.5194/bg-13-4271-2016`.
"""
resp_scaling = Q10 ** ((temperature - T0) / 10)
# This is mean over thirty days prior to the given day.
# I can't figure out how to get a centered window.
resp_mean = resp_scaling.rolling("30D").mean()
# Get the Respiration timeseries on the same timestep as
# temperature
flux_resp_baseline = flux_resp.resample(
temperature.index.freq
).interpolate(method="time")
# This is where I would deal with the first and last several
# timesteps.
return flux_resp_baseline / resp_mean * resp_scaling