import typing as t
import numpy as np
import pandas as pd
from ._core import InstancedCanopyElement
from ._tree import MeshTree
from ..spectra import InterpolatedSpectrum
from ... import data
from ...units import unit_registry as ureg
[docs]
def wellington_citrus_orchard(
padding=0, properties: t.Literal["rami", "hyperspectral"] = "rami"
) -> dict:
"""
Generate a base experiment dictionary based on the Wellington Citrus Orchard
scene from the RAMI benchmark series.
Parameters
----------
padding : int, optional, default: 0
Amount of padding applied to the generated canopy.
properties : {"rami", "hyperspectral"}, optional, default: "rami"
If set to ``"rami"``, the radiative properties used for the RAMI
benchmark series is used. If set to ``"hyperspectral"``, leaf and soil
reflectance are replaced with spectral datasets.
Returns
-------
dict
A dictionary suitable for keyword argument specification for the
:class:`.CanopyExperiment` and :class:`CanopyAtmosphereExperiment`
constructors. It contains the surface and canopy specifications.
Warnings
--------
* This is an experimental feature, please use with caution and report issues.
* The underlying data is sourced from the scene file archived distributed by
the DART team and post-processed to accommodate Eradiate-specific frame
conventions and formats. We do not recommend using it to run RAMI
benchmark cases.
"""
spectrum_table = pd.read_csv(
data.data_store.fetch("trees/citrus_sinensis/spectral.txt"),
sep=r"\s+",
comment="#",
index_col=0,
)
# Collect wavelengths
w = (
0.5
* (spectrum_table.loc["WLMIN"].values + spectrum_table.loc["WLMAX"].values)
* ureg.nm
) # Central wavelength for each band
# Collect trunk reflectance
trunk_reflectance = InterpolatedSpectrum(
quantity="reflectance",
wavelengths=w,
values=spectrum_table.loc["CTR[1-10]_WOOD_R"].values,
)
if properties == "rami":
# Collect ground reflectance
ground_reflectance = InterpolatedSpectrum(
quantity="reflectance",
wavelengths=w,
values=spectrum_table.loc["BGROUND_REFL"].values,
)
# Collect foliage reflectance/transmittance and define individual trees
citrus_sinensis = []
for i in range(1, 11):
transmittance = InterpolatedSpectrum(
quantity="transmittance",
wavelengths=w,
values=spectrum_table.loc[f"CTR{i}_FOLI_TRAN"].values,
)
reflectance = InterpolatedSpectrum(
quantity="reflectance",
wavelengths=w,
values=spectrum_table.loc[f"CTR{i}_FOLI_REFL"].values,
)
citrus_sinensis.append(
MeshTree(
id=f"cisi{i}",
mesh_tree_elements=[
{
"id": f"trunk{i}",
"mesh_filename": data.data_store.fetch(
f"trees/citrus_sinensis/CISI{i}/trunk.ply"
),
"reflectance": trunk_reflectance,
},
{
"id": f"foliage{i}",
"mesh_filename": data.data_store.fetch(
f"trees/citrus_sinensis/CISI{i}/foliage.ply"
),
"transmittance": transmittance,
"reflectance": reflectance,
},
],
)
)
elif properties == "hyperspectral":
# Collect ground reflectance
ground_ds = data.open_dataset("spectra/reflectance/lambertian_soil.nc").sel(
brightness="dark", drop=True
)
ground_reflectance = InterpolatedSpectrum(
quantity="reflectance",
wavelengths=ureg.Quantity(
ground_ds.w.values,
ground_ds.w.units,
),
values=ground_ds["reflectance"].values,
)
# Collect foliage reflectance/transmittance and define individual trees
citrus_sinensis = []
foliage_ds = data.open_dataset(
"spectra/reflectance/bilambertian_leaf_cellulose.nc"
)
w = ureg.Quantity(foliage_ds.w.values, foliage_ds.w.units)
transmittance = InterpolatedSpectrum(
quantity="transmittance",
wavelengths=w,
values=foliage_ds.transmittance.values,
)
reflectance = InterpolatedSpectrum(
quantity="reflectance",
wavelengths=w,
values=foliage_ds.reflectance.values,
)
for i in range(1, 11):
citrus_sinensis.append(
MeshTree(
id=f"cisi{i}",
mesh_tree_elements=[
{
"id": f"trunk{i}",
"mesh_filename": data.data_store.fetch(
f"trees/citrus_sinensis/CISI{i}/trunk.ply"
),
"reflectance": trunk_reflectance,
},
{
"id": f"foliage{i}",
"mesh_filename": data.data_store.fetch(
f"trees/citrus_sinensis/CISI{i}/foliage.ply"
),
"transmittance": transmittance,
"reflectance": reflectance,
},
],
)
)
else:
raise ValueError(f"unknown spectral property specification '{properties}'")
# Load instance positions
citrus_sinensis_locations = []
for i in range(1, 11):
loc = np.genfromtxt(
data.data_store.fetch(f"trees/citrus_sinensis/CISI{i}/location.txt")
)
zeros = np.zeros((len(loc), 1))
citrus_sinensis_locations.append(np.concatenate((loc, zeros), axis=1))
instanced_elements = []
for i, tree, locations in zip(
range(1, 11), citrus_sinensis, citrus_sinensis_locations
):
instanced_elements.append(
InstancedCanopyElement(
canopy_element=tree,
instance_positions=locations,
)
)
return {
"surface": {"type": "lambertian", "reflectance": ground_reflectance},
"canopy": {
"type": "discrete_canopy",
"instanced_canopy_elements": instanced_elements,
"size": [100, 100, 10],
"padding": padding,
},
}
