plant-food-research-open / fieldnbalance Goto Github PK

View Code? Open in Web Editor NEW

2.0 3.0 4.0 189.93 MB

Nitrogen balance model components and test procedures

Home Page: https://plant-food-research-open.github.io/FieldNBalance/

License: GNU General Public License v3.0

C# 0.51% Jupyter Notebook 99.32% Python 0.17% Shell 0.01%

biophysical-modeling crop-model crop-modeling mechanistic-modeling mechanistic-models nitrogen-balance dotnet python

fieldnbalance's Introduction

FieldNBalance

Website: FieldNBalance

Nitrogen balance model components and test procedures

fieldnbalance's People

Contributors

Stargazers

Watchers

Forkers

hamishbrownpfr rezaresystems nibble-byte jbris

fieldnbalance's Issues

Unit testing requirements

@HamishBrownPFR

I've added some scaffold code for unit testing here: https://github.com/Plant-Food-Research-Open/FieldNBalance/blob/main/SVSModel.Tests/SVSModelTest.cs

This works fine. Most recent test results from GitHub Actions: https://github.com/Plant-Food-Research-Open/FieldNBalance/runs/21364924990

Anyway, I'll let this one sit until you can secure and add the validation data - or we can plan out the required unit tests in more detail

Remove met data from model interface

It should not be necessary to pass the met data in through the interface as all the data is in the source code. Could easily refactor so the model gets its met data locally from the metfile name provided in the interface.

Moisture test data leading to overlapping points in matplotlib graphs

@HamishBrownPFR

Regarding the issue with the matplotlib graphs...

The issue is specific to Moisture test sets and graphs

In https://github.com/Plant-Food-Research-Open/FieldNBalance/blob/main/TestGraphs/MakeGraphs/Moisture.py#L40, you're loading your test sets from the TestComponents/TestSets/Moisture/OutPuts subdirectory

If I read the test set data from OutPuts, and use that to generate visualisations, then the graphs look ok: https://plant-food-research-open.github.io/FieldNBalance/

That's where your Moisture test sets are currently stored: https://github.com/Plant-Food-Research-Open/FieldNBalance/tree/main/TestComponents/TestSets/Moisture/OutPuts

However, in the current pipeline, the Moisture test sets will be written to the TestComponents/TestSets/Moisture/Outputs subdirectory. After this happens, the visualisations look strange...with overlapping points

All the other test sets are written to and plotted from an Outputs subdirectory. So this is basically a typo that's specific to Moisture

Anyway, I think we have a data quality issue with the generated Moisture test set data (duplicate data maybe?). I'm not familiar enough with this data to help, so you'll probably need to inspect it yourself

Drainage coefficients

Add table with drainage coefficients for each met site and mechinisum to read them into the model

403 error on GitHub Pages pull request - gh-pages plugin

Bring Soil N test calculations into the model code so they don't have to be done on the client side

Currently the model requires soil test results in kg N/ha requiring the client web app to calculate this from test results. It would be better in the long run to bring these calculations into the code so the client only needs to provide the information inputs to these calculations through the UI.

Add additional tests

WS1 Data
HWEON project test set

Crop coefficient

Incorporate data from WS1 analysis
Review forage coefficients based on Steves analysis
Incorporate carrot coeffs
Incorporate data from WS2
Make potato separate category and add cultivars

copy right notice

<one line to give the program's name and a brief idea of what it does.>
Author: Hamish Brown
Copyright (c) 2024 The New Zealand Institute for Plant and Food Research Limited

Residue model coefficients

Add final coefficients for residue model from fitting

Do N balance on a daily loop

The model is currently structured to calculate each of the components for the duration of the simulation and then stick them together to do the N balance. This worked well when the model was quite simple. However, we have had to add more complexity into the loss model and feedbacks to reduce uptake if soil N is limited which has resulted in a code structure that is quite hard to read.
It would be better to turn this into a true daily time-step model so it is easier to follow for others needing to ready of debug the code in the future.