File size: 2,673 Bytes
b8c97f2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 | """Small API for the processed earth_data_pack.h5 file."""
from __future__ import annotations
from pathlib import Path
import h5py
import numpy as np
class EarthDataPack:
def __init__(self, path: str | Path):
self.path = Path(path)
self.h5 = h5py.File(self.path, "r")
self.aligned = self.h5["aligned"]
def close(self) -> None:
self.h5.close()
def __enter__(self):
return self
def __exit__(self, exc_type, exc, tb):
self.close()
@property
def shape(self) -> tuple[int, int]:
return tuple(self.aligned["elevation_m"].shape)
def field(self, name: str, rows=slice(None), cols=slice(None)) -> np.ndarray:
return self.aligned[name][rows, cols]
def conditioning(self, rows=slice(None), cols=slice(None)) -> np.ndarray:
elevation = self.field("elevation_m", rows, cols).astype(np.float32)
temperature = self.field("temperature_mean_c", rows, cols).astype(np.float32)
seasonality = self.field("temperature_seasonality_c", rows, cols).astype(np.float32) * 100.0
precipitation = self.field("precipitation_mm_year", rows, cols).astype(np.float32)
precipitation_cv = self.field("precipitation_cv_percent", rows, cols).astype(np.float32)
elevation_signed_sqrt = np.sign(elevation) * np.sqrt(np.abs(elevation))
return np.stack([
elevation_signed_sqrt,
temperature,
seasonality,
precipitation,
precipitation_cv,
], axis=0)
def world_xy_to_pixel(self, x_km: float, y_km: float) -> tuple[int, int]:
height, width = self.shape
col = int(np.floor((x_km % float(self.h5.attrs["canvas_width_km"])) / float(self.h5.attrs["pixel_km_x"]))) % width
row = int(np.clip(np.floor(y_km / float(self.h5.attrs["pixel_km_y"])), 0, height - 1))
return row, col
def patch(self, center_x_km: float, center_y_km: float, width_cells: int, height_cells: int) -> dict[str, np.ndarray]:
height, width = self.shape
center_row, center_col = self.world_xy_to_pixel(center_x_km, center_y_km)
rows = np.clip(np.arange(center_row - height_cells // 2, center_row - height_cells // 2 + height_cells), 0, height - 1)
cols = np.arange(center_col - width_cells // 2, center_col - width_cells // 2 + width_cells) % width
result = {}
row_start = int(rows.min())
row_end = int(rows.max()) + 1
local_rows = rows - row_start
for name in self.aligned:
strip = self.aligned[name][row_start:row_end, :]
result[name] = strip[local_rows][:, cols]
return result
|