GRIDMET variance / z-score cube

This recipe demonstrates how to load GRIDMET temperature data, compute temporal anomalies, and aggregate the results spatially for QA plots.

import cubedynamics as cd
from cubedynamics import pipe, verbs as v
from cubedynamics.stats.anomalies import temporal_anomaly
from cubedynamics.stats.spatial import spatial_coarsen_mean
from cubedynamics.viz.qa_plots import plot_median_over_space

tmax = cd.load_gridmet_cube(
    lat=40.05,
    lon=-105.275,
    variable="tmmx",
    start="2000-01-01",
    end="2020-12-31",
    freq="MS",
    chunks={"time": 120},
)
tmax_anom = temporal_anomaly(tmax, dim="time")
tmax_z = pipe(tmax_anom) | v.zscore(dim="time")

tmax_z_coarse = spatial_coarsen_mean(tmax_z, factor_y=2, factor_x=2)

ax = plot_median_over_space(
    tmax_z_coarse,
    ylabel="Median tmax z-score",
    title="GRIDMET tmax anomalies (median over space)",
)

See also:

• PRISM precipitation anomaly / z-score cube for high-resolution precipitation summaries.
• Sentinel-2 NDVI anomaly (z-score) cube for vegetation dynamics that can be compared against the GRIDMET cube with xr.corr or the rolling helpers in cubedynamics.stats. (v.correlation_cube will return once the streaming implementation is ready.)