CHANGELOG.md

Changelog

All notable changes to the Groundwater Drawdown Tool are recorded here, and these notes are what the tool shows you when it updates itself.

Format inspired by Keep a Changelog — Added, Changed, Fixed, Removed.

[Unreleased]

[0.5.4] — 2026-08-12

Working through the feedback from the end-user testing round. These notes move under a version heading when the release is actually cut.

Added

• Licensed and unlicensed wells are now shown. Every well carries its GWELLS licence status — Licensed, Unlicensed, Historical, or Unknown where GWELLS does not say — in the details table, the map pop-up, the CSV, the PDF, the KML, and the standalone map. On the map, a dark ring around a marker means the well is currently licensed. Licence status is shown for information only; it does not change any well's status or at-risk result.
• An "Other" option in the aquifer picker. If the well is completed in an aquifer that has not been delineated, you can now say so — even when the point sits on or near mapped aquifers. Pick "Other", choose the material, and enter T and S. The results page and the PDF record the choice, and name the nearest mapped aquifer so a reviewer can see what was set aside.
• A "Method, assumptions and limitations" panel on the results page, collapsed until you open it. It explains what the tool does and does not tell you, and what the 30% at-risk threshold means. Guidance on aquifer defaults now sits with the T and S fields on the setup page, and guidance on checking driller's logs sits with the details table. The PDF export carries all of it in one section.
• A 0 m reference line on the distance-drawdown chart, marking the water level before pumping starts, so the drawdown and SAD bars have a datum to be read against.
• Plain-language definitions of s and r under the distance-drawdown chart, and of NPL above the details table.
• The map pop-up now shows the well's aquifer number.
• Aquifer numbers that are not formally delineated are now marked. GWELLS assigns some wells to an aquifer number the provincial aquifer layer has no mapped polygon for — Aquifer 1143 is one. Those now read "1143 (not delineated)" in the details table, the map pop-up, the CSV, the PDF, and the KML, so it is clear the number is not a mapped aquifer. The wells are still included and still analysed; a blank Aquifer ID still means GWELLS assigns the well to no aquifer at all. The check is made against the aquifer layer on every run rather than against a fixed list, so it keeps up as more aquifers are delineated.

Changed

• Aquifers are now identified by number first, with the material in brackets — "Aquifer 199 (Sand and Gravel)" — in the aquifer picker, the results summary, and the PDF, since that is how aquifers are usually referred to.
• Up to five nearby aquifers are suggested instead of three, for areas where several small aquifers sit close together.
• SAD bars on the distance-drawdown chart are now colour-coded. Orange means the well has headroom left; red means predicted drawdown has passed that well's Safe Available Drawdown. A red bar points upward from the well — that is not a fault, it is the tool showing an over-impacted well. A caption under the chart explains it.
• The manual-entry material option is now just "Unconsolidated" rather than "Unconsolidated (sand and gravel)", which wrongly implied every unconsolidated aquifer is sand and gravel.
• The impact chart caption now reads "sorted by magnitude of impact" and no longer describes the threshold line as red — the line is drawn dark on purpose, so it stays distinct from the red at-risk bars.
• Well tag numbers no longer get clipped at the top-right of the distance-drawdown chart.
• Well tag numbers on the distance-drawdown chart now alternate above and below their point, so wells at similar distances from the pumping well stop printing on top of each other. When even that isn't enough, a new "Show well tag numbers on the chart" tickbox turns them off — hovering a point still names the well, and the PDF captures the chart as you left it.
• The Impact % chart now keeps a WTN on every bar. It grows taller as the buffer gets busier instead of squeezing the bars until labels silently dropped out. Past roughly 85 wells there is genuinely no room left, and the caption now says so rather than leaving you to spot the gaps.

Removed

• The "Max drawdown" summary tile, from both the results page and the PDF. It was easily misread as the drawdown at the pumping well when it was actually the largest drawdown at any nearby well. Per-well drawdown is unchanged in the table, charts, and exports.

[0.5.3] — 2026-06-22

Added

• An "internal use only" notice now appears in the footer of every page: the tool is for internal use and must not be shared outside the organization. This notice shows in the tool only — it is deliberately left off exported reports, maps, and KML files, which may form part of a licence file.

Changed

• The results-interpretation disclaimer now names a Qualified Professional alongside the regional hydrogeologist: results must be interpreted by, or in consultation with, a regional hydrogeologist or a Qualified Professional with expertise in hydrogeology. The updated wording appears in the tool and on every export (PDF, interactive map, KML).

• The hints under the pumping-parameter fields were tidied up — clearer, shorter wording under the pumping-rate, duration, and buffer-radius fields.

Fixed

• In Lat / Lon input mode, clicking Place without entering both values now shows a red "Enter both longitude and latitude." message right next to the button, instead of appearing to do nothing. (The message was previously shown below the map, where it was easy to miss.)

[0.5.2] — 2026-06-16

Added

• The "Well tag number" input mode now links to the BC Groundwater Wells and Aquifers map (apps.nrs.gov.bc.ca/gwells). If you don't have a well tag number handy, you can find one by location on that map; the link opens in a new tab.

• A Clear button on the setup page resets all inputs for a fresh analysis without refreshing the page.

Changed

• The maps are larger on bigger monitors — both the setup and results maps now grow with the window while still fitting smaller laptop screens.

• A short note under the pumping-duration field explains that the 90-day default applies to all of BC.

• The page header now uses the official BC Government wordmark, which sits flush in the dark-blue band instead of appearing as a white panel.

• Storativity (S) now reads as a plain decimal (e.g. 0.00003) in the results-page input parameters and the PDF export, instead of scientific notation (3e-05). Transmissivity displays the same way.

• When a results table is empty, the message is now a clearly highlighted notice instead of a faint line of text — and the table's Export CSV button is hidden when there's nothing to export. This covers both "no wells were flagged at risk" and "no wells were found in the buffer". The buffer-empty message also suggests increasing the buffer radius and re-running.

Fixed

• The distance-drawdown chart no longer flips upside down when you press Autoscale after zooming in. The inverted axis (drawdown growing downward) now stays put, and "Reset axes" returns to it reliably.

• Refreshing the Results page no longer re-runs the analysis or discards your edits. Previously, pressing F5 (or restoring the tab) silently re-queried the BC Geographic Warehouse, wiped any per-well values you had overridden in the details table, and recorded the run a second time in the usage statistics. The page now reuses the cached result when the inputs haven't changed, so your overrides survive a refresh.

• The Run Analysis button is now greyed out until you place a pumping point, instead of letting you click it and then showing a "Place a pumping point first" message. Once a point is set (by map click, lat/lon, or well tag number), the button enables and any earlier message clears — so there's no need to scroll back up to check whether a point was placed.

[0.5.1] — 2026-06-01

Changed

• The documentation site now opens in light mode by default. The light/dark toggle in the sidebar still works and still remembers your choice per browser.

Fixed

• The footer's "What's new" panel no longer shows an empty "Unreleased" heading above the current release notes.
• First launch on a fresh install no longer shows a connection-refused page. run.bat now waits for the local server to be ready before opening your browser, instead of opening it after a fixed 4-second delay. On a cold first launch the server can take 10–15 seconds to start (Python and Dash have to load); the browser now opens at the moment the page is actually available, so there is no error page to refresh past. Subsequent launches are unchanged — the wait is essentially zero once everything is warm.

[0.5.0] — 2026-05-25

First broadly-installable release. Phase 5 (visual identity, map overlays, KML/PDF/HTML exports, logging and disclaimers) and Phase 6 (documentation site, auto-update on launch, version footer with changelog modal) shipped. Distributed through GitHub Releases — end users download setup.bat from the latest-release URL and the installer pulls the matching tool zip. A pre-release security audit closed out the open Dependabot advisories.

Pre-release for internal testing by the GIS team; a non-pre-release follows for end users once internal sign-off is in.

Security

• A pre-release dependency-and-code audit was completed; nine advisories were either resolved or formally reviewed. Five were fixed by refreshing transitive dependencies (idna, urllib3) and the development-only test runner (pytest). Four affect components that the tool depends on indirectly (Flask, Werkzeug) and are not exploitable on a localhost-only single-user tool; they will clear automatically when the tool migrates to the next major Dash release.
• The signed-in BCGW username is no longer written to the general application log file when a session starts. It remains in the structured usage log used for monitoring and troubleshooting, so traceability is unchanged.

Phase 6 — Online documentation

Added

• An online documentation site is now available, with a step-by-step User Guide covering installation, BCGW account help, running an analysis, reading the results, exporting them, troubleshooting, and the methods and assumptions behind the calculations: https://bcgov.github.io/groundwater-drawdown-tool/
• A "Documentation" link in the footer of every page opens that site.
• The documentation site supports a light/dark theme toggle (defaults to dark; your choice is remembered per browser).
• When the tool starts, it now checks for a new release on GitHub and updates itself in place if one is available. The check is silent when nothing has changed; if an update is applied you see a short "An update is available…" line and the dependency refresh. Pass --no-update to run.bat to skip the check on a slow network. Update failures are logged to logs\auto-update.log and never block the app from starting.
• The footer on every page now shows the running version and the date it was installed — for example, "Version 0.5.0 — last updated 2026-05-25". Clicking the version opens a "What's new" panel with the most recent release notes, so you can see at a glance what changed (and whether a colleague with a different install is on the same version as you).

Phase 5d — Logging, sign-in messaging, and UI polish

Added

• The tool now keeps a daily log file in the logs\ folder next to the tool (gwdrawdown.log). A new file starts each day and the last 30 days are kept, so if something goes wrong there is a record to look back on.
• Usage logging. Each time an analysis is run, a small summary record (run parameters and headline results — no passwords) is written to a central GeoBC log location, along with sign-in and error events. This helps the team monitor the tool's health, understand how it is used, and troubleshoot issues. If the central location can't be reached (for example, off the government network), logging quietly switches off — it never blocks or slows the tool.
• The sign-in screen now shows a "BCGW account help guide" link when sign-in fails, pointing at the documentation page that covers the common causes (locked or expired account, expired password, network or VPN) and the right next step for each.
• A short note on the sign-in screen confirming that your password is never stored — it is held only in memory for the session and discarded when you sign out.
• A second "← Back to Setup" button at the foot of the results page, so you don't have to scroll back to the top to start a new analysis.

Changed

• The page header now shows the official British Columbia logo in place of the typographic "British Columbia / Government of B.C." text.
• Friendlier sign-in error messages. A failed sign-in now explains the problem in plain language — wrong username or password, locked account, expired password, or a network / connection problem — instead of showing the raw database error. The technical error code is still shown, in small print, in case you need to quote it to support.
• Reworded the sign-in screen subtitle to "Connect to your BC Geographic Warehouse (BCGW) account to use the tool."
• Minor visual polish: the footer no longer repeats the signed-in user name (it is already shown in the header) and its disclaimer is centred; the "Impact % per well" chart caption is smaller so it sits below the section heading rather than competing with it.

Phase 5b — Map layers and overlays

Added

• Basemap switcher on both the setup and results maps. A layers control in the top-right corner switches between OpenStreetMap (the default), a topographic basemap, and satellite imagery.
• Context overlays you can toggle on either map:
  • Aquifers — BC's mapped aquifer polygons, shown by default on the setup map. Appears once you are zoomed in to roughly regional scale.
  • All BC Wells — every registered well in the province, on the setup map. Appears only when zoomed in close, so it doesn't swamp the view.
  • Water Management Districts and Water Management Precincts — the administrative boundaries, each with a name label that tracks the part of the boundary you're looking at and follows you as you pan and zoom.
• A small legend in the bottom-right corner of the map explains the aquifer and well symbology. It appears only while those layers are switched on.
• The setup map shows a crosshair cursor while you are in "Map click" mode — a clearer cue that the map is waiting for a click to place the pumping point.

Changed

• Entering a latitude / longitude or looking up a well tag number now recentres and zooms the map to that point automatically, so you see it in context without panning there yourself.

Phase 5c — Exports

Added

• Download KML button on the results page. Exports the pumping well and every nearby well as a KML file you can open directly in Google Earth. Each well is colour-coded by its status (at-risk, OK, and so on) and sized by its predicted impact — the same scheme as the results-page map — and carries its full result row (distance, drawdown, SAD, impact, and the rest), so you can inspect any well by clicking it in Google Earth.
• Download PDF report button on the results page. Produces a print-ready summary of the whole analysis, one section per page: page 1 — input parameters, a row of summary cards, and a method-and-assumptions note; the two result charts, one per page; then the at-risk wells table; then the full per-well details table. Every page carries a screening-tool disclaimer banner and a footer with the run timestamp, a unique run ID, the tool version, and your username — suitable for attaching to a licence assessment file. Wells outside the Cooper-Jacob validity range are tinted light purple in the per-well table, matching the on-screen view.
• Download interactive map (HTML) button on the results page. Produces a self-contained HTML file that opens in any browser as an interactive Leaflet map — the pumping well, its buffer, and every well with a click-through popup. A handy way to share the result without the full tool.
• All three exports reflect any per-well overrides you have applied, and the PDF charts are captured from exactly what you see on screen.

Changed

• The per-well CSV export gains an "Outside Validity" Yes/No column, so the Cooper-Jacob validity advisory (shown as a purple row tint on screen) survives the export to a format that can't carry cell colour.

Phase 4d — Aquifer selection fallback + manual entry

Added

• Nearby aquifers are now offered on the setup page alongside the aquifer the well directly overlaps. The tool searches a 1000 m radius and lists the three closest aquifers, each labelled with its distance (e.g. "Aquifer 123 — 47 m away") and tagged "(nearby — not directly overlapping)". The aquifer the well sits inside is tagged "directly overlapping" and pre-selected, but you can pick a nearby one instead — useful when a well sits inside one aquifer (e.g. bedrock) but just outside the boundary of the aquifer it should really be associated with (e.g. an unconsolidated aquifer 50 m away).
• Manual-entry mode for remote areas the Province hasn't mapped. A "No mapped aquifer at this location — enter materials manually" option appears at the bottom of the picker in the same fallback list. Choosing it reveals an aquifer-material dropdown (Unconsolidated or Bedrock) and requires you to enter T and S values directly. The same-aquifer filter is disabled in this mode (there's no polygon to filter against), and the results page shows an orange banner above the run summary so reviewers can see at a glance the run was based on user-supplied materials and T/S rather than mapped data.
• If no aquifers are found within 1000 m and none contain the point, the picker shows just the manual-entry option with a note explaining nothing nearby was found, so the workflow is never blocked by missing aquifer coverage.

Phase 5a.3 — Distribution via GitHub Releases

Changed

• The tool now ships through GitHub Releases instead of being handed out as a folder. End users download one file — setup.bat from https://github.com/bcgov/groundwater-drawdown-tool/releases/latest/download/setup.bat — and double-click it. The installer pulls the matching release zip, extracts it to %USERPROFILE%\Tools\groundwater-drawdown-tool\, installs uv + Python 3.13, and runs uv sync. No editing files, no admin rights.
• Re-running the same setup.bat checks for newer releases and updates the install in place. .env, outputs\, logs\, and flask_session\ are preserved across updates so users don't lose exports or have to re-login on every release. Update time is roughly 30 seconds; "already up to date" exits in under a second.
• CLIENT_INSTALL.md rewritten end-to-end for the one-URL flow.

Added

• scripts/publish_release.ps1 — developer-side release script. Verifies a clean tree on main, runs uv run pytest, builds groundwater-drawdown-tool.zip excluding dev cruft, tags the commit, pushes the tag, and creates the GitHub release with the zip and setup.bat as assets. Release notes are extracted from the matching CHANGELOG.md section. -Draft flag for pre-release review; -SkipTests for emergency use.
• README publish-workflow section covering gh CLI setup and the release cut steps.

Phase 5a.2 — Setup-page polish and form styling

Added

• Section icons beside each setup-page section heading (location pin, layers, sliders).
• The setup page's source-aquifer "Override default T / S" and "Filter out wells spatially outside source aquifer" controls are now on/off toggle switches (instead of basic checkboxes) so it's clearer at a glance whether they are on or off.
• The setup-page input-mode selector (Map click / Lat-Lon / Well tag number) is now a segmented control — one row of three buttons rather than radio dots.
• A new m³/yr pumping-rate unit, so multi-year licence-volume estimates can be entered directly without pre-converting.

Changed

• Pumping-rate unit list reordered to lead with m³/d (now default), followed by m³/min, m³/s, m³/yr, L/min, L/s. Imperial GPM and US GPM removed (BC officers don't use them outside the legacy BCGW YIELD column, which still flows through its own conversion). Default Q value is now 200 m³/d (was 3.97 L/s).
• Default pumping duration is 90 days (was 100). Quick-pick presets are now 30 d / 90 d / 180 d / 1 yr / 10 yr.
• Results page now reads as a sequence of clearly-separated sections (Distance-drawdown, Impact %, Map, At-risk, All wells) with a faint divider line above each section heading. Headings are smaller and more compact so the charts and tables get the visual attention.
• "Map" section heading renamed to "Wells in buffer (map view)" so it describes what the map actually shows.
• "Editable columns" line on the per-well details table is now on its own bold line and lists the four columns directly, with the longer usage notes below it.
• The pagination reminder under the at-risk and per-well tables moved from above the table to below — closer to the page controls themselves.
• Setup-page Lat / Lon and Well-tag-number input panels now have compact, fixed-width inputs with the action button on its own row below.

Fixed

• The footer no longer floats above the bottom of the screen on short pages like the login screen — the page wrapper was shortening itself by the header height it contained.

Phase 5a.1 — Visual identity

Added

• The tool now follows the BC government visual identity. Every page has a dark-blue header with the "British Columbia" wordmark, a thin gold stripe, the app title, and (once signed in) your username with a Logout button. Matching dark-blue footer with the version, your username, and a "screening tool" reminder.
• A show / hide password button (eye icon) on the sign-in screen, in case you want to double-check what you typed.

Changed

• The status tiles on the results page (Total wells, At risk, OK, etc.) have been redesigned. White tiles with a coloured left edge and a larger number, in place of the pale pastel backgrounds — easier to read at a glance and consistent with the rest of the new BC theme.
• Buttons, section panels, and link colours updated to match the BC navy and gold palette.

[0.4.0] — 2026-05-14

First milestone release. Phase 4 complete: end-to-end interactive workflow from BCGW login through to a results dashboard with two charts, a colour-coded map, an at-risk summary table, and an editable per-well details table with live recompute. The tool now runs the same canonical example case as the legacy Excel and produces visually-equivalent output (deck slide 21 reference).

Not yet released to end users — the auto-update mechanism (Phase 6) isn't built, so installation is still a manual checkout. Treat 0.4.0 as the milestone version against which Phase 5 polish and exports will land.

Added

• Initial project scaffolding.
• Phase 1 skeleton: src/gwdrawdown/ package layout (core/, data_access/, ui/{pages,components}/), config.py with hardcoded BCGW_DSN and env-overridable defaults, and an app.py stub that launches an empty Dash app on localhost:8050 and logs the tool version on startup.
• Committed uv.lock for reproducible installs.
• Phase 2 core math (no UI): core/units.py (BCGW field conversions plus the CSV-driven pumping-rate unit table), core/crs_utils.py (WGS84 <-> BC Albers via cached pyproj.Transformer), core/drawdown.py (Cooper-Jacob with r->0.1 m fallback, u<0.01 validity check, and a superposition-ready signature accepting a list of pumping sources), core/aquifer_lookup.py (T/S defaults from data/ts_lookup.csv), core/well_classification.py (legacy Impact!R reassigned-material rule), core/sad.py (legacy Impact!U SAD formula with three status branches). All ported from the legacy Excel; verified against the canonical case (Q=3.97 L/s, T=250 m^2/d, S=0.005, t=180 d, r=100 m). 66 pytest tests, all passing.
• Phase 3 data access: data_access/db.py (lazy oracledb thin-mode pool; init_pool called by the future login handler, get_connection context manager that raises PoolNotInitialisedError pre-login, close_pool on logout/shutdown), data_access/queries.py (parameterised SQL templates for the four BCGW queries from DATA_REFERENCE.md §6: nearby wells with optional same-aquifer filter, aquifers-at-point returning a list for stacked polygons, subtype-code lookup, well-by-tag), scripts/smoke_test_db.py (getpass-based developer smoke test). Verified live against BCGW using the Cowichan Bay test point (intersects aquifer 186 sand & gravel and 198 bedrock); all four queries return sensible results. Three new optional config keys: DB_POOL_MIN, DB_POOL_MAX, DB_POOL_INCREMENT (defaults 1/2/1).
• Phase 4a auth shell: multi-page Dash app (use_pages=True), server-side sessions via Flask-Session (filesystem backend at config.SESSION_DIR, lifetime config.SESSION_TIMEOUT_HOURS), /login page that runs SELECT 1 FROM DUAL to verify BCGW credentials before calling data_access.init_pool and redirecting to /setup, Flask /logout route that closes the pool and clears the session, root redirect / -> /setup if authenticated else /login, footer component on every page showing version + signed-in user + Logout link. Setup and results pages are 4a stubs (sub-stages 4b and 4c will fill them in). Verified end-to-end in the browser with valid and invalid BCGW credentials.
• Phase 4c.1 results dashboard (read-only). Replaces the 4b text dump with a real layout: run-summary block (timestamp, BCGW user, source aquifer, T/S used with override marker, Q in m³/day, duration, buffer, filter), seven colour-coded stat cards (Total / OK / At risk / Insufficient / Suspect / Outside / Max drawdown), at-risk summary table (InputValues!B30:E32 parity, descending by Impact %), and a full per-well table with the 17 columns from PROJECT_PLAN.md §4.1. Both tables are sortable / filterable, paginate at 10 rows / page, have a sticky header row, and live inside their own scroll containers so the horizontal scrollbar stays in view. Each table has a custom Export CSV button (no built-in export to avoid dash_table's fixed-columns layout bug); CSV reflects the current sort + filter state via derived_virtual_data. Status cells are colour-coded per WellStatus. New WellResult fields for total depth, stickup, and the per-well top-of-fracture/screen override (populated as None today; 4c.2 turns the latter three into editable cells with per-row live recompute). Auto-loaded assets/styles.css standardises dash_table's filter-row placeholder visibility on hover. New ts_overridden flag on AnalysisInputs so the run-summary can mark T/S as "(override)" when the officer customised them.
• Phase 4c.2 editable per-well overrides + live recompute. The per-well details table on /results gains four editable columns — NPL, finished depth, stickup, and top of fracture/aquifer/screen — and rebuilds SAD, Impact %, status, the at-risk summary table, and the seven stat cards as the officer types. Overrides live in a sessionStorage dcc.Store keyed by WTN; the page-level pipeline callback caches the BCGW result in another Store (analysis-result) so override edits and tab refreshes don't replay the queries. Editing a cell to match the BCGW value reverts the override; clearing a cell does the same. The four editable columns are declared with type="any" (rather than numeric) so dash_table doesn't treat empty input as invalid and silently revert the cell; the trade-off is alphabetical sort on those four columns, which the new "Edited" summary column at the right of the table compensates for. The Edited column lists the field names of any active overrides for each row (e.g. "NPL, Stickup") and is carried through to the CSV export. A "Reset all overrides" button next to Export CSV wipes every per-well override in one click. Rows with active overrides are tinted light yellow. New analysis.recompute_well (pure math, no DB) shares its kernel with _compute_well_result; analysis.apply_overrides rebuilds the AnalysisResult totals from a cached base + per-WTN override map. New analysis.effective_u_threshold centralises the Cooper-Jacob bypass — flipping the bypass back is now a one-line edit shared by the initial pipeline and the override recompute. JSON round-trip on WellResult and AnalysisResult preserves the three enum fields. 110 pytest tests passing (14 new); ruff clean.
• Phase 4c.3 distance-drawdown chart, results map, spatial source- aquifer filter, and advisory Cooper-Jacob validity flag. The /results page now leads with a Plotly distance-drawdown chart (red dots with WTN labels, smooth black Cooper-Jacob curve, vertical orange SAD bars, inverted Y axis) matching the legacy Excel chart from deck slide 21, followed by a dash-leaflet map with the pumping well, a translucent buffer-radius circle, and a CircleMarker per observation well coloured by status and sized by predicted impact. Chart and map are cross-linked through a page-scoped selected-well Store: clicking a chart point or a map marker highlights the matching well in both views.
• Source-aquifer filter is now spatial and default off (confirmed with the client). When enabled, the nearby-wells query filters with an SDO_ANYINTERACT correlated subquery against the source aquifer polygon geometry rather than comparing w.AQUIFER_ID to the source id. This safeguards against stale GWELLS aquifer assignments and against future re-delineation of aquifer boundaries — a well's recorded AQUIFER_ID may drift, but its point geometry is authoritative. Filter label updated to "Filter out wells spatially outside source aquifer".
• Cooper-Jacob u<0.01 validity check is now an advisory rather than a hard status (per client direction). Wells failing the check keep their SAD-based status (AT_RISK, OK, etc.) and the math still computes u_max per row, but the per-well table tints the affected rows light purple so the officer can spot them at a glance. Purple takes precedence over the yellow override tint on rows that trip both. A small legend above the per-well table explains the row tints and reminds users that the table paginates after 10 rows.
• Setup page now restores the last-run inputs on Back-to-Setup. Point, lat/lon, Q + unit, duration, buffer radius, same-aquifer filter, and the source-aquifer pick all replay from analysis-inputs on /setup mount, so iterating on a parameter no longer means re-keying the whole form. T/S override values are deliberately not restored — re-tick "Override default T / S" if you customised them.
• Phase 4b setup page + analysis pipeline. New core/flagging.py combines drawdown + SAD into a WellStatus (OK / AT_RISK / INSUFFICIENT_DATA / SUSPECT_DATA / OUTSIDE_VALIDITY); SUSPECT_DATA fires when GWELLS reports static water level deeper than the well bottom (e.g. WTN 96473), so the UI can flag the baseline record for review rather than the proposed pumping. New analysis.py orchestrates BCGW queries -> per-well SI conversion -> Cooper-Jacob -> SAD -> classification -> flagging, with a pure _compute_well_result function that's unit-tested without a database. The Cooper-Jacob u<0.01 validity check is bypassed at the pipeline level pending client confirmation (math still computes u_max for diagnostics; revert is a one-line change). Real setup page replaces the 4a stub with three input modes (map click / lat-lon / WTN with auto-aquifer), source-aquifer picker for stacked polygons, T/S override checkbox with full-precision float display, pumping-rate dropdown driven by data/unit_conversions.csv, duration presets (30 d / 100 d / 1 yr / 10 yr), buffer radius (default 1000 m), same-aquifer filter (default on per Q12). Run Analysis opens results in a new browser tab via clientside callback so the user can iterate on the setup page without losing earlier results. Results page in 4b is a textual dump of the pipeline output (sub-stage 4c builds the chart, tables, and map). Hardened the auth shell: is_authenticated() now also verifies the BCGW pool is open (clears stale sessions left over from a previous app run); SESSION_USE_SIGNER=True so cookies issued before a SECRET_KEY rotation are rejected. 96 pytest tests passing (16 new flagging + 14 analysis); ruff clean.

Removed

• .env.example. The tool runs without a .env; BCGW credentials are entered through the login UI at runtime. Override variables are documented in README.md.

[0.1.0] — superseded

Initial scaffolding version, never released; rolled into 0.4.0 above when Phase 4 closed out.