Commits adding functions to mads_calibration/SA_post_hoc_analysis.py for post sensitivity analysis comparison of equilibrium quality metrics and time series data to either user input thresholds or prescribed defaults. Additionally, there is an update to the nitrogen checking function here to produce further outputs. These are highlighted below.
plot_equilibrium_metrics_scatter plots all results as a scatter plot
for considering spread of results compared with either user input or
default equilibrium metrics.
plot_equilibrium_metrics_boxplot plots boxplots similarly to above but
as a single figure for more concise analysis
plot_equilibrium_relationships plots equilibrium time series against
target (observed) value for all targeted variables
equilibrium_check create Pandas DataFrames showing pass/fail statistics
for each target variable, bool for each variable following all tests,
and bool for each variable with each test for closer inspection.
Additionally plots pass/fail bar graphs for each variable.
nitrogen_check function for post sensitivity analysis run
nitrogen limitation description in SA_post_hoc_analysis.py
script.
Requires INGPP, GPP, AVLN output variables with nitrogen
feedback enabled during SA_setup_and_run.py.
Nitrogen_check function analyses ratio between INGPP:GPP
and compares to range associated with boreal or tundra
biome, providing pandas DataFrames giving pass/fail %
and INGPP:GPP ratio, AVLN for final 10 years of equilibrium.
Creates plots for INGPP:GPP and AVLN time series for visual
inspection and bar plot showing pass/fail percentage. Plot
titles suggest possible adjustment for parameters or new
available nitrogen target.
Functions require either a path to a sample directory or output eq_quality and targets
csv files. Nitrogen check function specifically requires a path but also that INGPP, GPP, and AVLN are specified when the sensitivity analysis is first run, with nitrogen feedback enabled. Tested on AC-REFAC branch with combinations of variables containing non-pft, pft, and compartments. May require further debugging after
broader uptake.
Commits adding functions to
mads_calibration/SA_post_hoc_analysis.pyfor post sensitivity analysis comparison of equilibrium quality metrics and time series data to either user input thresholds or prescribed defaults. Additionally, there is an update to the nitrogen checking function here to produce further outputs. These are highlighted below.plot_equilibrium_metrics_scatter plots all results as a scatter plot for considering spread of results compared with either user input or default equilibrium metrics.
plot_equilibrium_metrics_boxplot plots boxplots similarly to above but as a single figure for more concise analysis
plot_equilibrium_relationships plots equilibrium time series against target (observed) value for all targeted variables
equilibrium_check create Pandas DataFrames showing pass/fail statistics for each target variable, bool for each variable following all tests, and bool for each variable with each test for closer inspection. Additionally plots pass/fail bar graphs for each variable.
nitrogen_check function for post sensitivity analysis run nitrogen limitation description in SA_post_hoc_analysis.py script.
Requires INGPP, GPP, AVLN output variables with nitrogen feedback enabled during SA_setup_and_run.py.
Nitrogen_check function analyses ratio between INGPP:GPP and compares to range associated with boreal or tundra biome, providing pandas DataFrames giving pass/fail % and INGPP:GPP ratio, AVLN for final 10 years of equilibrium.
Creates plots for INGPP:GPP and AVLN time series for visual inspection and bar plot showing pass/fail percentage. Plot titles suggest possible adjustment for parameters or new available nitrogen target.
Functions require either a path to a sample directory or output eq_quality and targets csv files. Nitrogen check function specifically requires a path but also that INGPP, GPP, and AVLN are specified when the sensitivity analysis is first run, with nitrogen feedback enabled. Tested on AC-REFAC branch with combinations of variables containing non-pft, pft, and compartments. May require further debugging after broader uptake.