Adding species layer to complex area brings in every occurrence record

[timhicks-ala]

When adding a particularly complex area to SP, and then adding a species layer, instead of adding a single species to only the specified area as expected, it instead loads every occurrence record in every location (ignoring the species selection as well as the area). This affects uploaded shapefiles as well as areas already defined in spatial layers.

Issue raised in helpdesk ticket 91645, where a user is combining multiple areas from a layer and attempting to download records from the combined area (https://spatial.ala.org.au/?ss=1598923626401), and received records that fell outside of the defined areas as well. Also related to helpdesk ticket 82236, for which #308 was raised to fix.

As an easy example, this also affects the bushfire extents layer:

• Add the National Indicative Aggregated Fire Extent 2019-2020 area to SP via Gazetteer polygon option
• Add a single species (eg. Osphranter rufus) to SP, restricted to the fire extent areas
• Every occurrence record of every species is loaded, including outside the defined area.

[Tasilee]

Thanks @timhicks-ala. I'd hoped that this was fixed. There is a work-around: Use the defined area for Tools | Area report - interactive, and then download the relevant data - which could be filtered externally for the relevant species? This seems to work for the 37 combined DIWA areas.

[timhicks-ala]

Looks like the area report isn't quite right either, it shows plenty of records outside of those areas: https://biocache.ala.org.au/occurrences/search?q=qid:1598923543842&fq=#tab_mapView

The original reporter seems okay to reduce the number of areas they're working with to work around the issue.

[timhicks-ala]

Additionally raised in 92050

[timhicks-ala]

A user has confirmed that while a particularly complex shape file can trigger this issue, reducing the overall size of the area does work around the issue (as they are able to work with multiple smaller areas), which suggests this issue is entirely based on the complexity of an uploaded area.

[nickdos]

@adam-collins add commits.

[adam-collins]

Partially resolved with https://github.com/AtlasOfLivingAustralia/spatial-hub/commit/267ceb21152999072564f87bcee691c8ee79e7b9

[adam-collins]

related to #344

[Tasilee]

Not working as advertised on test. Mapping Eucalyptus gunnii on area Tasmania (from ASGS States/Territories) maps all occurrence records everywhere. I gave up waiting for it to finish mapping.

[qifeng-bai]

Reproduce this issue: Add to map -> gazzetter area -> Tasmania from ASGS stats/territories Add species: Eucalyptus gunnii

Current implementation is to collect WKT from this layer and upload wkt to biocache. It would cause issue when WKT is too big.

Fix: using layer, for example: cl1023 : Tasmania, instead of wkt

[qifeng-bai]

Another issue is caused by two suspected reasons: 1, oversized wkt 2, intersected polygons （There is no intersections of polygons by visual check on QGIS）

Working case: https://biocache.ala.org.au/ws/webportal/params/details/1616994553941

Find eucalyptus gunnii in two polygons

Broken case: https://biocache.ala.org.au/ws/webportal/params/details/1616715731279 When we load complicated areas, query in BS lost species and wkt. aspis (1).zip

@alexhuang091 found BS collect wkt and q from params, which has de facto limit of 2000 characters

[qifeng-bai]

Depended on: https://github.com/AtlasOfLivingAustralia/biocache-service/issues/579

[qifeng-bai]

Related to https://github.com/AtlasOfLivingAustralia/spatial-service/issues/120

[qifeng-bai]

After updated Tomcat settings, I confirmed that adding a species in a complicated area on Spatial-test and prod works now