[GIS] Polygon geometry suddenly disappeared/deleted from shapefile in QGIS

corruptqgisqgis-3shapefile

I was editing a polygon shapefile in QGIS (3.0.1) today when suddenly 966 out of 5382 polygons disappeared from the map. I had been changing attribute information in the table, cutting polygons, etc as I was going along and saving my shapefile regularly. The polygons are still in the attribute table (if I select all my polygons using a spatial selection and then inverse the selection, the lines with the missing polygons show up) but an error message (zoom failed, feature has no geometry, "le zoom vers l'entité a échouché, l'entité n'a pas de géométrie") appears when I attempt to find those polygons. There was no spatial clumping to the lost polygons nor a common attribute.

Comparison of yesterday's file (purple underlay) with today's corrupted file (black hatched)

I can't share the file because it is proprietary so I realize that it is limited what can be troubleshooted just on my description. I have the back-up in our server from end of last workday so I will have only lost one day of work but it is still frustrating.

Best Answer

Not sure if this is any help, but have you got the polygons categorised into colours etc in the properties panel. It happened on mine where i changed attribute but never had the new attribute in the category so manually added it and then it showed up.

Related Solutions

[GIS] Merging and dissolving shapefile in R

I think we should do this bit by bit. It's good to keep in mind that the simpler and more minimal you keep the example, the more likely you are to get responses. Here's my first shot. I don't see anything obviously wrong with it so far - it produces a plot. However, because you've provided only a subset of the data, the plot is not contiguous (see below). Why not try this with the full data set and see whether it works for you.

spanish counties

Note that the fortified data frame looks okay as well.

> str(foo)
'data.frame':   7876 obs. of  7 variables:
 $ long : num  168544 169128 169380 170315 170700 ...
 $ lat  : num  4814368 4814211 4813874 4814704 4816283 ...
 $ order: int  1 2 3 4 5 6 7 8 9 10 ...
 $ hole : logi  FALSE FALSE FALSE FALSE FALSE FALSE ...
 $ piece: Factor w/ 49 levels "1","2","3","4",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ group: Factor w/ 138 levels "2003.1","2003.2",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ id   : chr  "2003" "2003" "2003" "2003" ...

Here's the code.

library(rgeos)
library(ggplot2)
work.dir <- "your_dir_no_trailing_slash"
zones.sp <- readOGR(dsn = work.dir, layer = "SP0111")
zones.data <- read.csv("statsz.csv")
zones.spwd <- zones.sp
zones.spwd@data <- merge(zones.sp@data, zones.data, by = "PROVMUN", all = TRUE)
zones.IDs <- as.numeric(zones.spwd[["CODER"]])
# note that we only have a subset of the data so only a few zones...
which(!is.na(zones.IDs))
# this seems to work
regions.sp <- gUnaryUnion(zones.spwd, id = zones.IDs)
str(regions.sp)
plot(regions.sp)
foo <- fortify(regions.sp)

QGIS – Updating Point Shapefile Attributes from Polygon Shapefile Using QGIS

Excellent answers provided by @ChrisW and @CAccesible, both describe techniques which should do what you want. Just to add on what @ChrisW mentioned, a spatial join tool called Join by location would provide your points layer with another attribute column containing the names of the polygon it resides in, regardless of how many points there are in a polygon. I use this function many times with the following parameters to obtain results similar to what you want:

Join by location

Related Question