xleaf

xleaf

Leaf and canopy radiative transfer modeling tools built on PROSPECT-D and SAIL.

---

Introduction

🌳 xleaf is a python package for running leaf and canopy simulation models using PROSAIL. It provides python bindings to the PROSPECT-D & 4SAIL Fortran code.

🌿 It includes sensible defaults that make it easy to get up and running quickly, and clear code documentation in the form of docstrings and type hints.

📚 All credit for the fundamental modeling code and for the underlying science belongs to the original researchers. xleaf is mostly a wrapper. Please cite their most recent research:

@article{feret2017prospect,
  title={PROSPECT-D: Towards modeling leaf optical properties through a complete lifecycle},
  author={Feret, J-B and Gitelson, AA and Noble, SD and Jacquemoud, S},
  journal={Remote Sensing of Environment},
  volume={193},
  pages={204--215},
  year={2017},
  publisher={Elsevier}
}

🧙 Shout out to my man JB.

If you used xleaf in your research and want to credit it, please cite:

@software{anderson_xleaf_2026,
  author       = {Anderson, Christopher B.},
  title        = {xleaf: Leaf and canopy radiative transfer modeling tools for PROSPECT-D and SAIL},
  month        = aug,
  year         = 2026,
  version      = {1.2.2},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.21786969},
  url          = {https://github.com/earth-chris/xleaf}
}

---

Install

Install from PyPI:

pip install xleaf

Development

This project uses pixi to manage dependencies and builds. To set up a development environment:

pixi install

This installs all dependencies including the Fortran compiler (gfortran), meson, and numpy. To run tests:

pixi run test

---

Leaf and canopy simulations

import xleaf
import matplotlib.pyplot as plt

# run with off-the-shelf defaults
leaf = xleaf.simulate_leaf()

# or specify detailed parameters
canopy = xleaf.simulate_canopy(
    chl = 40, # ug/cm2
    car = 8, # ug/cm2
    antho = 0.5, # ug/cm2
    ewt = 0.01, # cm
    lma = 0.009, # g/cm2
    N = 1.5, # unitless
    lai = 3.0, # m2/m2
    lidf = 30, # degrees
    soil_dryness = 0.75, # %
    solar_zenith = 35, # degrees
    solar_azimuth = 120, # degrees
    view_zenith = 0, # degrees
    view_azimuth = 60, # degrees
    hot_spot = 0.01, # unitless
)

# and plot them together
plt.plot(xleaf.wavelengths, leaf, label='leaf')
plt.plot(xleaf.wavelengths, canopy, label='canopy')
plt.legend()

📄 The definitions and expected range of values for each parameter are described in the xleaf docstrings.

---

Chaining leaf and canopy models

🔗 simulate_canopy runs PROSPECT and 4SAIL together. If you’d rather run the canopy model on a leaf spectrum you already have — a measured spectrum, or one from a different leaf model — use simulate_sail. It takes leaf reflectance and transmittance directly, so you can pair it with simulate_leaf(transmittance=True):

import xleaf
import matplotlib.pyplot as plt

# simulate a leaf, returning both reflectance and transmittance
reflectance, transmittance = xleaf.simulate_leaf(
    chl = 40, # ug/cm2
    car = 8, # ug/cm2
    antho = 0.5, # ug/cm2
    ewt = 0.01, # cm
    lma = 0.009, # g/cm2
    N = 1.5, # unitless
    transmittance = True,
)

# then run the canopy model on that leaf spectrum
canopy = xleaf.simulate_sail(
    reflectance,
    transmittance,
    lai = 3.0, # m2/m2
    lidf = 30, # degrees
    soil_dryness = 0.75, # %
    solar_zenith = 35, # degrees
    solar_azimuth = 120, # degrees
    view_zenith = 0, # degrees
    view_azimuth = 60, # degrees
    hot_spot = 0.01, # unitless
)

plt.plot(xleaf.wavelengths, reflectance, label='leaf')
plt.plot(xleaf.wavelengths, canopy, label='canopy')
plt.legend()

This enables canopy simulations from field-measured reflectance and transmittence.

---

Random forests

📊 xleaf provides classes for generating random parameters within the global range of expected values. These classes have a .sample() method for generating an appropriate random value based on a literature review.

import xleaf
import matplotlib.pyplot as plt

# generate 5 random leaf spectra from global defaults
for idx in range(5):
    chl = xleaf.ChlorophyllSampler.sample()
    car = xleaf.CarotenoidSampler.sample()
    antho = xleaf.AnthocyaninSampler.sample()
    ewt = xleaf.EWTSampler.sample()
    lma = xleaf.LMASampler.sample()
    N = xleaf.NSampler.sample()
    leaf = xleaf.simulate_leaf(chl, car, antho, ewt, lma, N)
    plt.plot(xleaf.wavelengths, leaf, label=f"leaf {idx+1}")

plt.legend()

🧪 Or experiment by setting the range of values yourself:

import xleaf
import matplotlib.pyplot as plt

MyLAISampler = xleaf.UniformSampler(min=2, max=6)
MyVZASampler = xleaf.NormalSampler(mean=0, stdv=3, min=-10, max=10)

# generate 5 random canopy spectra just varying LAI/VZA
for idx in range(5):
    lai = MyLAISampler.sample()
    vza = MyVZASampler.sample()
    canopy = xleaf.simulate_canopy(lai=lai, view_zenith=vza)
    plt.plot(xleaf.wavelengths, canopy, label=f"lai: {lai:0.2f}, vza: {vza:0.2f}")

plt.legend()

⚡ These parameters don’t always vary independently. Try to exercise caution when constructing parameter estimates to ensure biological realism.

---

Developed by

Christopher Anderson[^1]

[^1]: Planet Labs PBC, San Francisco