Join spatial features with dataframe by id with inconsistent format - r

Question

Hello everyone I was hoping I could get some help with this issue:
I have shapefile with 2347 features that correspond to 3172 units, perhaps when the original file was created there were some duplicated geometries and they decided to arrange them like this:
Feature gis_id
1 "1"
2 "2"
3 "3,4,5"
4 "6,8"
5 "7"
6 "9,10,13"
... like that until the 3172 units and 2347 features
On the other side my data table has 72956 observations (about 16 columns) with data corresponding to the gis_id from the shapefile. However, this table has a unique gis_id per observation
head(hru_ls)
jday mon day yr unit gis_id name sedyld tha sedorgn kgha sedorgp kgha surqno3 kgha lat3no3 kgha
1 365 12 31 1993 1 1 hru0001 0.065 0.861 0.171 0.095 0
2 365 12 31 1993 2 2 hru0002 0.111 1.423 0.122 0.233 0
3 365 12 31 1993 3 3 hru0003 0.024 0.186 0.016 0.071 0
4 365 12 31 1993 4 4 hru0004 6.686 16.298 1.040 0.012 0
5 365 12 31 1993 5 5 hru0005 37.220 114.683 6.740 0.191 0
6 365 12 31 1993 6 6 hru0006 6.597 30.949 1.856 0.021 0
surqsolp kgha usle tons sedmin ---- tileno3 ----
1 0.137 0 0.010 0
2 0.041 0 0.009 0
3 0.014 0 0.001 0
4 0.000 0 0.175 0
5 0.000 0 0.700 0
6 0.000 0 0.227 0
With multiple records for each unit (20 years data)
I would like to merge the geometry data of my shapefile to my data table. I've done this before with sp::merge I think, but with a shapefile that did not have multiple id's per geometry/feature.
Is there a way to condition the merging so it gives each feature from the data table the corresponding geometry according to if it has any of the values present on the gis_id field from the shapefile?

Answer 1

This is a very intriguing question, so I gave it a shot. My answer is probably not the quickest or most concise way of going about this, but it works (at least for your sample data). Notice that this approach is fairly sensitive to the formatting of the data in shapefile$gis_id (see regex).
# your spatial data
shapefile <- data.frame(feature = 1:6, gis_id = c("1", "2", "3,4,5", "6,8", "7", "9,10,13"))
# your tabular data
hru_ls <- data.frame(unit = 1:6, gis_id = paste(1:6))
# loop over all gis_ids in your tabular data
# perhaps this could be vectorized?
gis_ids <- unique(hru_ls$gis_id)
for(id in gis_ids){
# Define regex to match gis_ids
id_regex <- paste0("(,|^)", id, "(,|$)")
# Get row in shapefile that matches regex
searchterm <- lapply(shapefile$gis_id, function(x) grepl(pattern = id_regex, x = x))
rowmatch <- which(searchterm == TRUE)
# Return shapefile feature id that maches tabular gis_id
hru_ls[hru_ls$gis_id == id, "gis_feature_id"] <- shapefile[rowmatch, "feature"]
}
Since you didn't provide the geometry fields in your question, I just matched on Feature in your spatial data. You could either add an additional step that merges based on Feature, or replace "feature" in shapefile[rowmatch, "feature"] with your geometry fields.