I feel that the description of the link tables may benefit from a bit more precision about what each column represents. I also would like to correct the equivalence made between n:n relations and clusters, which are in fact not equivalent.
A list of overlapping outline pairs between RGI 7.0 and RGI 6.0 describing 1:1, 1:n, n:1 or n:n relationships as well as the overlapping area between them. For example, a perfect match between an RGI 7.0 and RGI 6.0 outline results in a 1:1 relation with 100% area match in both. If a single RGI 6.0 outline was divided into two glaciers for RGI 7.0, a 2:1 relationship (a cluster) would result with two lines in the table with twice 50% area match in RGI 6.0 and twice 100% match in RGI 7.0. In more complex cases the matches are not always perfect and the relationships less straightforward, for example when an outline was remapped.
Could become something like:
A list of all overlaps (greater than 200 m2) between RGI 7.0 (rgi7_id) and RGI 6.0 (rgi6_id) glacier outlines. The overlap area (overlap_area_km2) is divided by the area of the original outlines to also provide area fractions for RGI 7 and RGI 6 (rgi7_area_fraction and rgi6_area_fraction), respectively. Columns n_rgi7 and n_rgi6 indicate whether each overlap is isolated (a 1:1 relation) or part of a more complex 1:n, n:1, or n:n relation. For example, a perfect match between an RGI 7 and RGI 6 outline results in a 1:1 relation with overlap area fractions of 1. If an RGI 6 outline is divided into two outlines (of equal area) in RGI 7, the table will contain two rows with n_rgi6: 1, n_rgi7: 2, rgi6_area_fraction: 0.5, rgi7_area_fraction: 1. Often the relation between RGI7 and RGI6 is more complex, for example when an outline was remapped in RGI 7 and partially overlaps many in RGI 6.
Note that n_rgi7 and n_rgi6 only count direct relatives – that is, the total number of RGI 7 outlines overlapping rgi6_id and the total number of RGI 6 outlines overlapping rgi7_id. Column cluster_id groups together all overlapping RGI 6 and RGI 7 outlines such that each cluster does not overlap any other cluster.
Or probably better with the help of a table:
A list of all overlaps (greater than 200 m2) between pairs of RGI 7.0 and RGI 6.0 glacier outlines, described by the following columns:
Column
Type
Description
rgi7_id
string
RGI 6.0 outline
rgi6_id
string
RGI 7.0 outline
overlap_area_km2
number
Overlap area in km2
rgi7_area_fraction
number
Overlap area (overlap_area_km2) divided by the area of the RGI 7.0 outline (rgi7_id)
rgi6_area_fraction
number
Overlap area (overlap_area_km2) divided by the area of the RGI 60 outline (rgi6_id)
n_rgi7
integer
Total number of RGI 7.0 outlines that overlap the RGI 6.0 outline (rgi6_id)
n_rgi6
integer
Total number of RGI 6.0 outlines that overlap the RGI 7.0 outline (rgi7_id)
cluster_id
integer
Arbitrary cluster identifier, which groups together all overlapping RGI 6.0 and RGI 7.0 outlines such that each cluster does not overlap any other cluster
For example, if an RGI 6 outline perfectly matches an RGI 7 outline, the overlap is a 1:1 relation (n_rgi6: 1, n_rgi7: 1) with 100% coverage (rgi7_area_fraction: 1, rgi6_area_fraction: 1). If an RGI 6 outline divided into two outlines (of equal area) in RGI 7, the two overlaps are part of a 1:2 relation (n_rgi6: 1, n_rgi7: 2) with 50% and 100% coverage (rgi6_area_fraction: 0.5, rgi7_area_fraction: 1). Often the relation between RGI7 and RGI6 is more complex, for example when an outline was remapped in RGI 7 and partially overlaps many in RGI 6.
I feel that the description of the link tables may benefit from a bit more precision about what each column represents. I also would like to correct the equivalence made between n:n relations and clusters, which are in fact not equivalent.
So the current text (https://www.glims.org/rgi_user_guide/products/glacier_product.html?highlight=_links#product-files):
Could become something like:
A list of all overlaps (greater than 200 m2) between RGI 7.0 (
rgi7_id
) and RGI 6.0 (rgi6_id
) glacier outlines. The overlap area (overlap_area_km2
) is divided by the area of the original outlines to also provide area fractions for RGI 7 and RGI 6 (rgi7_area_fraction
andrgi6_area_fraction
), respectively. Columnsn_rgi7
andn_rgi6
indicate whether each overlap is isolated (a 1:1 relation) or part of a more complex 1:n, n:1, or n:n relation. For example, a perfect match between an RGI 7 and RGI 6 outline results in a 1:1 relation with overlap area fractions of 1. If an RGI 6 outline is divided into two outlines (of equal area) in RGI 7, the table will contain two rows withn_rgi6
: 1,n_rgi7
: 2,rgi6_area_fraction
: 0.5,rgi7_area_fraction
: 1. Often the relation between RGI7 and RGI6 is more complex, for example when an outline was remapped in RGI 7 and partially overlaps many in RGI 6. Note thatn_rgi7
andn_rgi6
only count direct relatives – that is, the total number of RGI 7 outlines overlappingrgi6_id
and the total number of RGI 6 outlines overlappingrgi7_id
. Columncluster_id
groups together all overlapping RGI 6 and RGI 7 outlines such that each cluster does not overlap any other cluster.Or probably better with the help of a table:
A list of all overlaps (greater than 200 m2) between pairs of RGI 7.0 and RGI 6.0 glacier outlines, described by the following columns:
rgi7_id
string
rgi6_id
string
overlap_area_km2
number
rgi7_area_fraction
number
overlap_area_km2
) divided by the area of the RGI 7.0 outline (rgi7_id
)rgi6_area_fraction
number
overlap_area_km2
) divided by the area of the RGI 60 outline (rgi6_id
)n_rgi7
integer
rgi6_id
)n_rgi6
integer
rgi7_id
)cluster_id
integer
For example, if an RGI 6 outline perfectly matches an RGI 7 outline, the overlap is a 1:1 relation (
n_rgi6
: 1,n_rgi7
: 1) with 100% coverage (rgi7_area_fraction
: 1,rgi6_area_fraction
: 1). If an RGI 6 outline divided into two outlines (of equal area) in RGI 7, the two overlaps are part of a 1:2 relation (n_rgi6
: 1,n_rgi7
: 2) with 50% and 100% coverage (rgi6_area_fraction
: 0.5,rgi7_area_fraction
: 1). Often the relation between RGI7 and RGI6 is more complex, for example when an outline was remapped in RGI 7 and partially overlaps many in RGI 6.