The 132-Page Well Inspection Atlas That Will Not Plot
A state water board publishes a 132-page inspection atlas at 1:24,000, one page per public-supply well buffer. Every page carries the same layers: a hillshade at 40 percent transparency, 20-foot contours, PLSS grid labels, road names, hydrology, and well IDs with a red halo. Inspectors report that in dense subdivisions the well IDs disappear entirely and road names collide with contour labels. Because labels are placed dynamically, a road name near a page edge appears on one page and is absent on its neighbor, so field notes cite a street that is not on the map. Page 47's hillshade sits on top of the contours in one export and beneath them in another. Exporting the full series to PDF takes about 5 hours, each page is roughly 38 MB, and the plotter queue rejects any job over 100 MB. The board wants one PDF per well.
1. Diagnose the disappearing well IDs and the road names that vanish at page edges, then recommend a fix appropriate for a published inspection atlas.
Read the worked answer
Why it happens. Dynamic labels are placed independently for each page extent. A road name near a page boundary is evaluated against whatever else falls inside that page, so both its placement and whether it is drawn at all change from page to page. That is precisely what breaks a field note citing a street.
Fix for a published atlas: annotation. Run Convert Labels to Annotation over the entire atlas extent, not per page, and use feature-linked annotation for roads and wells so edits propagate. Store it in the geodatabase. Annotation carries fixed placement, so every page clips the same annotated text and the street appears consistently wherever it falls.
If labels must stay dynamic:
- Set a reference scale on the map so symbol and label sizes are stable.
- Use the Maplex engine, and raise the well ID label class priority above contours and roads.
- Give well IDs a label buffer and halo; cap contour density with the contour placement options or label every nth.
- Turn off place-overlapping-labels, and set road labels to at least one label per feature part.
What not to do: lower the map scale per page to make labels fit. It changes the atlas contract and hides data that inspectors are required to see.
2. The hillshade flips above and below the contours between pages of the same series. Explain the likely cause and how you would make the stacking order auditable.
Read the worked answer
Diagnosis. Layer order inside one map is not page-dependent, so if the hillshade flips, the atlas is not being produced from one map. Ranked by frequency: the series was assembled from layer files created on top of different base data; two maps exist in the project and different page numbers point at different ones; a group layer's visibility is driven by a page-name expression that removes and re-adds the hillshade; or the hillshade was re-added from a different source and inherited a new position.
Fix. Keep one map, one map series, one layer stack: hillshade at the bottom with transparency applied in Appearance, or a Multiply blend over a neutral ramp, then contours, hydrology, roads, PLSS, wells, then annotation. Save the stack as a layer file and treat it as the template, not as something rebuilt per page.
Make it auditable. Before export, run a short arcpy.mp check that iterates each map's layers (listLayers) and asserts the hillshade's index is 0 and its transparency value matches the standard, failing loudly if not. Run the same check after any data source is repointed; layer order and effects are not preserved reliably when sources are swapped, and that is how page 47 diverged from its neighbors.
3. Propose a plan that reduces the 5-hour export and the 38 MB per page while still delivering one PDF per well inside the 100 MB plotter limit.
Read the worked answer
38 MB per page means the hillshade and contours are being written as heavy content on every page. Attack it in three places.
Export options. In the PDF export settings, enable compression of vector graphics, set image compression to JPEG at quality around 80, and resample to 300 dpi for line work with a lower rate for the hillshade. Uncheck layers and attributes if inspectors do not need feature-level PDF layers; that single option usually removes most of the size. Keep georeferencing only if field devices consume it.
Content. Drop layers no page needs. Pre-render the hillshade once as a mosaic or cached raster instead of projecting it live for 132 pages. Symbolize contours with a single line class rather than elevation-banded classes.
Throughput and limits. Five hours is about 2.3 minutes per page. Partition the page range and run several Pro instances concurrently, each scripted to export its own page subset and write one PDF per well. Per-well files are small, satisfy the queue, and mean one corrupted export does not cost the whole run. Validate the first ten pages visually and check file sizes before committing the batch, and record the export settings in the project so the next edition does not rediscover them.