starman / code /layers /frend_weh.py
enzeeoai's picture
Publish verified Core native layers and full geomorphometry
a8e0542 verified
Raw History Blame Contribute Delete
5.79 kB
"""TGO FREND water-equivalent hydrogen (Malakhov et al. 2022) -> per-cell `ice_wt_pct_frend*` (planet mars).
Earth twin: ``ggd318`` ``ice_wt_pct`` (master decision 14). Input is the smoothed map ``WEHMaps.txt``
(columns X Y WEHMIN WEH WEHMAX; 1 deg pixels, lon -179.5..179.5 east-positive, lat -49.5..49.5, latitude
assumed planetocentric). The map was produced with a 12-degree FWHM Gaussian, so the 1 deg pixels are not
independent. ``SuppMap.txt`` (the raw neutron-suppression map) is NOT used for any published column. FREND is
not filled from GRS: the two stay in separate columns.
"""
from __future__ import annotations
from pathlib import Path
import numpy as np
import pyarrow as pa
from . import _climatology as clim
SOURCE = {
'product': 'Water content in the upper 1 m of the Martian regolith from TGO FREND (Zenodo 6282972)',
'version': 'Zenodo record 6282972 v1, WEHMaps.txt (smoothed)',
'url': 'https://doi.org/10.5281/zenodo.6282972',
'licence': 'CC BY 4.0 (Zenodo record); cite Malakhov et al. 2022, JGR Planets, doi:10.1029/2022JE007258.',
'planet': 'mars',
'file': 'WEHMaps.txt',
}
VERSION = 'v0.1'
MAP_NAME = 'WEHMaps.txt'
UNUSED_FILES = ('SuppMap.txt',) # recorded in the manifest only
LAT_LIMIT_DEG = 50.0
SRC_RES_KM = 700.0 # 12 deg FWHM at the equator (paper's own figure UNVERIFIED)
_C = {'dtype': 'float32', 'unit': 'wt%', 'comparison': 'planet_specific'}
COLUMNS = {
'ice_wt_pct_frend': {**_C, 'description': 'FREND water-equivalent hydrogen, upper ~1 m, wt % (smoothed map '
'interpolated to the cell; Earth twin: ggd318 ice_wt_pct)'},
'ice_wt_pct_frend_min': {**_C, 'description': 'FREND WEHMIN: -1 sigma statistical suppression pushed through the model'},
'ice_wt_pct_frend_max': {**_C, 'description': 'FREND WEHMAX: +1 sigma statistical suppression pushed through the model'},
'src_res_km': {'dtype': 'float32', 'unit': 'km', 'comparison': 'planet_specific',
'description': 'effective resolution of the smoothed map (12 deg FWHM at the equator)'},
'interp_flag': {'dtype': 'string', 'unit': '', 'comparison': 'categorical',
'description': 'bilinear interpolation status'},
}
LIMITATIONS = [
'Coverage is +-50 deg latitude only; cells poleward are NaN with outside_coverage and are not filled from GRS.',
'The map is smoothed (12 deg FWHM, ~700 km), far coarser than the 1 deg pixels and a 5 km cell.',
'WEHMIN / WEHMAX are 1-sigma statistical ranges, not model or systematic uncertainty.',
'The WEH zero point is tied to one GRS-derived value at Solis Planum, so FREND and GRS are not independent in absolute level.',
'Latitude type is assumed planetocentric. Depth ~1 m. SuppMap.txt (raw suppression) is not published.',
]
def read_map(path) -> dict[str, np.ndarray]:
"""Parse the whitespace table (header row X Y WEHMIN WEH WEHMAX) into column arrays."""
with open(path) as fh:
header = fh.readline().split()
data = np.loadtxt(path, skiprows=1, ndmin=2)
if data.shape[1] != len(header):
raise ValueError(f'{path}: {data.shape[1]} columns but header {header}')
return {name: data[:, i] for i, name in enumerate(header)}
def to_grid(table: dict[str, np.ndarray], column: str):
lon, lat = table['X'], table['Y']
glon, glat = np.unique(lon), np.unique(lat)
field = np.full((len(glat), len(glon)), np.nan)
field[np.searchsorted(glat, lat), np.searchsorted(glon, lon)] = table[column]
return glat, glon, field
def plan(band_group, ctx):
return [{'key': MAP_NAME, 'url': str(Path(ctx.raw_dir) / MAP_NAME), 'method': 'local'}]
def prepare(ctx) -> None:
ctx.source_checksum = ctx.record_source_file(Path(ctx.raw_dir) / MAP_NAME)
for name in UNUSED_FILES: # recorded in the manifest only
path = Path(ctx.raw_dir) / name
if path.exists():
ctx.record_source_file(path)
def reduce(band_group, items, ctx):
table = read_map(items.all()[0].path)
cells = ctx.cells(band_group)
return build_rows(cells['region_id'], cells['center_lat'].to_numpy(zero_copy_only=False),
cells['center_lon'].to_numpy(zero_copy_only=False), table)
def build_rows(region_id, lat, lon, table) -> pa.Table:
lat = np.asarray(lat, dtype=np.float64)
out, flags_interp = {}, []
glat = glon = None
outside = np.abs(lat) > LAT_LIMIT_DEG
for column, name in (('WEH', 'ice_wt_pct_frend'), ('WEHMIN', 'ice_wt_pct_frend_min'),
('WEHMAX', 'ice_wt_pct_frend_max')):
glat, glon, field = to_grid(table, column)
# the map's edge pixel centres are +-49.5 but it covers to +-50: query at the edge row inside that strip
lat_q = np.clip(lat, glat[0], glat[-1])
values, flag = clim.bilinear_at(lat_q, lon, glat, glon, field)
clamped = (lat != lat_q) & ~outside
flag = np.where(clamped & (flag == 'bilinear'), 'nearest_fallback', flag)
out[name], flags_interp = np.where(outside, np.nan, values), flags_interp + [flag]
interp = clim.worst_flag(*flags_interp)
interp = np.where(outside, 'outside_coverage', interp)
flags = clim.Flags(len(lat))
flags.add(outside, 'outside_coverage')
flags.add(~outside & ~np.isfinite(out['ice_wt_pct_frend']), 'no_source')
data = {'region_id': pa.array(list(region_id), type=pa.string())}
for name, values in out.items():
data[name] = pa.array(values.astype(np.float32))
data['src_res_km'] = pa.array(np.full(len(lat), SRC_RES_KM, dtype=np.float32))
data['interp_flag'] = pa.array([str(x) for x in interp], type=pa.string())
data['quality_flags'] = pa.array(flags.strings(), type=pa.string())
return pa.table(data)