Selecting Variables¶

By default, table(g) reads a curated set of variables for each product. You can customize which columns you get.

Viewing defaults¶

using SpaceLiDAR

g = granule("ATL08_20201121151145_08920913_006_01.h5")
vars = SpaceLiDAR.default_variables(g)

Each Variable has a name, HDF5 path, element type, and optional transform.

Adding extra columns¶

Append to the defaults:

vars = SpaceLiDAR.default_variables(g)
push!(vars, Variable(:slope, "land_segments/terrain/h_te_slope", Float32))
push!(vars, Variable(:canopy_h, "land_segments/canopy/h_mean_canopy", Float32))

t = table(g; variables=vars)

Canopy height variants¶

ATL08 and GEDI have pre-defined canopy variable sets:

# ATL08 canopy (reads h_mean_canopy_abs instead of h_te_mean)
t = table(g; variables=SpaceLiDAR.atl08_canopy_variables())

# GEDI canopy (reads elev_highestreturn instead of elev_lowestmode)
t = table(g; variables=SpaceLiDAR.gedi_l2a_canopy_variables())

Custom variable from scratch¶

A Variable needs at minimum a name, path, and type:

Variable(:my_col, "land_segments/some/dataset", Float32)

With a transform (applied at read time):

Variable(:is_land, "land_segments/surf_type", Int8, ToBool())
Variable(:time, "land_segments/delta_time", Float64,
    ToDateTime("/ancillary_data/atlas_sdp_gps_epoch", SpaceLiDAR.gps_offset))

Direct H5Table construction uses the same Variable form for transformed columns:

H5Table(file; vars=[
    Variable(:is_land, "land_segments/surf_type", Int8, ToBool()),
])

Available transforms:

Transform	Effect
`ToBool()`	nonzero → `true`
`InvertBool()`	zero → `true`
`ToDateTime(epoch_path, offset)`	delta_time → DateTime
`ToDateTimeConst(offset)`	delta_time → DateTime (fixed epoch)
`SliceRow(n)`	take row `n` from 2D dataset

Attributes¶

Attributes are scalar metadata attached to each track (partition):

attrs = SpaceLiDAR.default_attributes(g)

They become constant columns in the resulting table (e.g., :detector_id, :strong_beam).

Explore interactively¶

Don't know what's in the file? Use explore:

t = explore(g)

This opens an interactive browser showing all datasets, their dimensions, and lets you select variables visually. The result is an H5Table ready for DataFrame(t).