Portrayal Catalogue assumes single spatial association per feature.
Closed this issue · 36 comments
the current (v1.5.0) assumes only one spatial association per feature. This isn't a constraint placed by S-100 (part 10a) or S-101 PS itself. Some recent test datasets show at least one implementer uses multiple spatial associations (in the SPAS record) for features which can cause unpredictable results.
PC modifications would be required to ensure datasets can be reliably loaded/portrayed.
I have attached the S101Cell that brought this issue to light.
Please feel free to use this cell for further development...
101AU00413130.zip
this is the cell I made when we were looking at this initially. It's a very simple test dataset with two pairs of spatially disjoint features. There are issues with the current PC in displaying QoBD - I will update this cell with something which shows the issue in more detail but a more complete review of the PC may well show up other areas where a single spatial association is assumed.
this is the cell I made when we were looking at this initially. It's a very simple test dataset with two pairs of spatially disjoint features. There are issues with the current PC in displaying QoBD - I will update this cell with something which shows the issue in more detail but a more complete review of the PC may well show up other areas where a single spatial association is assumed.
Error: [DepthArea:Feature|7] reference to [Surface|4] is invalid. Outer boundary [Curve|3:Forward] must have clockwise orientation.
Also, would be good to have spatial quality applied to some of the curves.
@DavidGrant-NIWC. In the attached dataset I have altered the spatial quality of a number of depth contours. The contours are located around 12-01.49s 130-46.28e. The dataset is complied to FC1.2.0.
101AU00413130.zip
Could you also apply spatial quality to a point and a sounding (multipoint)? Just want to verify I'm not breaking anything...
Unpatched
Patched
@DavidGrant-NIWC. Please see the attached for the requested revisions.
Sounding group near 12-00-58.42s 130-52-37.45e and LateralBuoy 12-05-25.26s 130-55-52.02e
This image shows the location of the amended soundings
i have a test data spec from Tom R which I'm looking at today which needs to have a representative set of data too. I'll post it when I'm done with it and we can amend if we need to add more cases.
Thanks Matt. I should have been clearer; the point should be something like a wreck which has a sounding value.
No problems at all @DavidGrant-NIWC. I have inserted a wreck at 12-01.03s 130-46.83e as requested above.
101AU00413130.zip
another test cell. This has a few different combinations of features, each with multiple spatials (all same primitive) with each pair having one spatial association and one not. Feedback? I believe it loads cleanly but probably needs a bit of tidying up and some extra test cases. I got the spec from Tom R and can feedback if more cases are needed....
updated with proper dataset,......
It's adding the low accuracy symbol to both geometries but only one is approximate:
- There doesn't appear to be any spatial quality associated with either of the bottom
Wreckgeometries.
Missing contour label (I think this is a bug in our testbed):
Confirmed as bug in test bed implementation, here is portrayal after patching:
- Would be better visually if the curve obstructions didn't look like surfaces.
- The PSWG should consider whether the sounding / isolated danger symbol should be drawn on the visible portion(s) of the curve(s). Currently symbols are drawn half-way along the geographic length of each curve and the symbols may not be visible when a portion of the dataset extends beyond the edge of the screen.
sounds good - I can modify....
I'll square the polygons up too.
do we need any other features?
Ideally, we would check all the features associated with Surface or Curve geometry which call GetFlattenedSpatialAssociations(). This includes all the features which call DEPARE03, DEPCNT03, OBSTRN07, QUALIN02, QUAPNT02, SLCONS04, and WRECKS05.
| CSP | Callers |
|---|---|
DEPARE03 |
DepthArea, DredgedArea |
DEPCNT03 |
DepthContour |
OBSTRN07 |
Obstruction, UnderwaterAwashRock |
QUALIN02 |
QUAPOS01 |
QUAPOS01 |
Coastline, LandArea |
QUAPNT02 |
OBSTRN07, QUAPOS01, SLCONS04, WRECKS05 |
SLCONS04 |
ShorelineConstruction |
WRECKS05 |
Wreck |
So: DepthArea, DredgedArea, DepthContour, Obstruction, UnderwaterAwashRock, Coastline, LandArea, ShorelineConstruction, and Wreck. The attribution needs to be set appropriately to ensure all the code paths are tested.
It would also be good to spot-check a few other random features.
To do: evaluate / update remaining callers to QUAPNT02: OBSTRN07, QUAPOS01, SLCONS04, WRECKS05
Test data is portrayed correctly after updating OBSTRN07->QUAPNT02 and support routines in PortrayalAPI:
With the exception of WRECKS05, all other callers to QUAPNT02 are from point geometries. Since S-101 doesn't permit multiple SPAS to point geometries there shouldn't be any further changes required to OBSTRN07, QUAPOS01, SLCONS04 or their callers.
WRECKS05 continuation B requires updating and should be tested.
@kusala9 I believe one of the two Wreck geometries in the bottom row should have spatial quality, but it doesn't appear to be present. Might want to also move the spatial quality on the Obstruction area to the left feature so spatial quality is present on all the left-hand geometries but not on the right-hand geometries.
PsWG meeting on 03/12/24
It was decided to resolve this shortfall in PC 2.0.0. For this, the PsWG was advised by the IHO representative that a change proposal to PC 2.0.0 has to be submitted via a MS as part of the consultation process under IHO CL39/2004.
- Australia submit a change proposal before the closing date (13/12/24).
sorry this took a while to process. included in this version
- curve features look like curve features
- two point features with two points each, one with spatial association, one without. Wreck and UnderwaterAwashRock
- fix to the surface wreck geometries
what attribution do you want on the point features, it's the bare minimum but is easy to change.
I will also add LandArea with coastline and shoreline construction and post another version.
fixes duplicate foid in the Point features.
- I think one of the two wreck geometries has spatial quality but let me know if not.
- I just saw the notes about forbidding multiple point associations. I will remove those from the dataset 👍
- Also, just adding land area with coastline and shoreline construction.
With the exception of WRECKS05, all other callers to QUAPNT02 are from point geometries. Since S-101 doesn't permit multiple SPAS to point geometries there shouldn't be any further changes required to OBSTRN07, QUAPOS01, SLCONS04 or their callers.
WRECKS05 continuation B requires updating and should be tested.
@kusala9 I believe one of the two Wreck geometries in the bottom row should have spatial quality, but it doesn't appear to be present. Might want to also move the spatial quality on the Obstruction area to the left feature so spatial quality is present on all the left-hand geometries but not on the right-hand geometries.
busy afternoon 🥇 - took out point geometries. New Land Area and Coastline. Now looking at shoreline construction. Please ignore the other iterations of this dataset...
Can the dataset be encoded as 2.0.0?
The land area doesn't appear to have spatial quality on either geometry.
I think one of the two wreck geometries has spatial quality but let me know if not.
It does, but it's not currently portrayed correctly. I'll address this today. Recommend moving the spatial quality to the left-hand geometry.
After patching WRECKS05
can do, i'll post fixes in the morning tomorrow - apart from Shorelineconstruction is there anything else required?
can do, i'll post fixes in the morning tomorrow - apart from Shorelineconstruction is there anything else required?
- Add
ShorelineConstruction - Add
DredgedArea - The left-hand
Coastlinegeometry should be approximate
- Land Area. Should the spatial association be on the Surface itself, or on the curve components forming the exterior of the surface? Or are these separate test cases?
- I'll add a dredged Area
- and make coastline approximate.
- Land Area. Should the spatial association be on the Surface itself, or on the curve components forming the exterior of the surface? Or are these separate test cases?
- I'll add a dredged Area
- and make coastline approximate.
- 2.0.0 (and edition number increased)
- dredged area
- shoreline construction
- coastline should be associated with spatial quality with quality of horizontal measurement approximate
Some have spatial quality on the surface, some on the curves which make up the surface, I can do either or both? Caris seems to render it ok (and the spatial quality ones are on the left) - how do I see in the NIWC viewer which ones have the quality associated with them? I can't seem to query it? Is the wreck on the right or the left? They can be changed.
Some have spatial quality on the surface
Spatial quality isn't allowed on surfaces per the DCEG:
Summary
- I believe all these features are now portrayed correctly.
- Currently the viewer doesn't show spatial quality; you need to use the ShoreECDIS to see the quality information. We'll need to get you an updated testbed because there have been testbed corrections made to support multiple SPAS.
- This isn't a problem, but you may want to move the spatial quality of the Wreck geometry to the left-hand geometry.
- Shoreline construction and Coastline appear correct.
- The dredged area also looks correct. It's interesting that dredged areas don't portray spatial quality. They always display as a dashed line. If the boundary coincides with the safety contour, then the safety contour will reflect the approximate nature of the shared curve (see last pic below).
Overview
Zoomed in
Note that the low-accuracy symbol may not be visible when zoomed out since the center of the symbol may be outside of the area it is annotating. Zooming in will cause the symbol to appear:
Shoreline Construction / Coastline
Dredged Area
looks good. I've taken out the spatial quality on the surfaces and swopped the wreck ones around. let me know if any other cases are needed.
I think it's good (plus we're out of time!) Thanks for all the updates!
| SC=13 | SC=9 | SC=5 |
|---|---|---|
 |
 |
 |