Field-Based Carbon Flux Model

A model for field-specific carbon flux estimation, based on the NASA Soil Moisture Active Passive (SMAP) Level 4 Carbon (L4C) model.

Installation

While this project is in active development, it's currently recommended to install the Python package in "editable" mode, using pip:

# From the directory containing setup.py
pip install -e .

During development, it's convenient to rely upon the pyl4c package for some functionality. However, pyl4c is not a hard requirement because it has some dependencies that may be difficult for some users to install (HDF5 and GDAL). To install with support for pyl4c:

# Install GDAL first, ensuring that the version matches the system library
pip install GDAL==$(gdal-config --version)
pip install -e .[pyl4c]

Running Tests

The test suite depends on pytest.

pip install -e .[dev]

Tests can be run with:

pytest tests/

Some tests depend on having pyl4c installed and will be skipped if the module is not available.

Getting Started

import numpy as np
from agstack.models import TCF
from agstack.drivers import drivers_for_tcf
from agstack.io import drivers_from_csv, params_dict_from_json

# For a single site, a CSV file containing surface meteorology
data, dates = drivers_from_csv('752731_field2_2D.csv')

# Load driver (meteorology)
drivers = drivers_for_tcf(data)

# Soil organic carbon state, e.g., from SoilGrids 250m, separated into
#    three different pools
state = np.array([[150, 150, 450]]).T

# Read in model parameters
params = params_dict_from_json('../agstack/data/SPL4CMDL_V7_BPLUT.json')

model = TCF(params, [7], state)
model.spin_up(dates, drivers)

nee, gpp, rh = model.forward_run(drivers)

Prior Art and Citation

Model code here is based heavily on the publicly available pyl4c package (Endsley et al. 2022). See REFERENCES for a complete list of references.

The fraction of photosynthetically active radiation (fPAR) is currently based on an empirical formula of Sentinel NDVI:

fPAR = (-18.4 + 120.3*NDVI)