A uniform method of adjusting global attributes of altered datasets

[JanisGailis]

Expected behavior

Some operations, such as subset_spatial, subset_temporal, coregistration, have a potential of changing dataset properties that should be properly reflected in the global attributes of affected datasets. There should be a uniform method of addressing this.

Actual behavior

Logic that handles updates to global attributes is part of particular operations.

Implementation hints

Create utility functions in ops/normalize.py or ops/utility.py that hamdle this. Such as adjust_spatial_attrs, adjust_temporal_attrs that work with attributes denoted in Attribute Convention for Data Discovery. Specifically:

Spatial:
geospatial_lat_min:         -90.0
geospatial_lat_max:         90.0
geospatial_lon_min:         -180.0
geospatial_lon_max:         180.0
geospatial_vertical_min:    0.0
geospatial_vertical_max:    0.0
spatial_resolution:         1km at Equator
geospatial_lat_units:       degrees_north
geospatial_lon_units:       degrees_east
geospatial_lat_resolution:  0.5
geospatial_lon_resolution:  0.5

Temporal:
time_coverage_start:        20010101T000000Z
time_coverage_end:          20010101T235959Z
time_coverage_duration:     P1D
time_coverage_resolution:   P1D

As global attributes are optional, it is better to have no attribute than to have a wrong attribute.

Task list

• Create adjust_spatial_attrs
• Create adjust_temporal_attrs
• Use in subset_spatial
• Use in subset_temporal
• Use in coregistration
• Check if has to be used in other operations
• Update operation development guide
• Update CHANGE.md

Specifications

Cate 0.9.0dev3

[JanisGailis]

See also #223