Milestones
List view
Transform Adapt into a complete desktop GUI application where all pipeline execution, monitoring, and configuration are controlled visually without terminal use. ### Scope - [ ] Add **Run Dashboard** to start, stop, pause, resume, and monitor live pipelines. - [ ] GUI becomes primary consumer of logs, status events, warnings, queue lag, and progress metrics. - [ ] Alternative to CLI workflows with GUI controls for radar, mode, output paths, schedules, modules, and thresholds. - [ ] Add named **Run Profiles** (saved configs) selectable from dashboard. - [ ] Selecting profile auto-loads behavior, modules, thresholds, paths, plotting, runtime mode. - [ ] Add **New Run Wizard** to collect user inputs step-by-step and generate config automatically. - [ ] Provide cookie-cutter templates (Optional). - [ ] Add editable config manager: create, clone, rename, export, import, delete profiles. - [ ] Add live health panels: downloader state, processor state, queue size, lag, files processed, errors. - [ ] Add integrated repository browser. - [ ] Add notifications for failures, backlog, missing data, stalled pipeline. - [ ] Support one-click rerun using previous profile. - [ ] Keep CLI optional for advanced users, GUI as default experience. ### Outcome Adapt can be installed and operated as a standalone scientific GUI platform with zero terminal interaction.
No due date•2/2 issues closedExtend Adapt analysis beyond basic radar-variable statistics (Z, Vr, σv, ZDR, ΦDP, KDP, ρHV) to include 3D volumetric and derived convective cell diagnostics. Scope: - Compute 3D cell structure (volume, vertical extent, core geometry) - Derive integrated quantities (VIL, echo-top/base, column depth) - Add vertical profiles and gradients (∂Z/∂z, ZDR columns) - Standardize derived products outputs - This will be done with extension modules.
Due by October 31, 2027•0/1 issues closedAdapt currently tracks storms but does not selects scan targets with a configurable science logic. This milestone delivers a configuration-driven rule engine that evaluates tracked cells against user-defined criteria and ranks candidate targets for adaptive scanning. Exit criteria Rule engine Rules defined entirely in configuration — no code changes to add, remove, or modify a rule Supported criteria include at minimum: storm size, intensity, growth rate, proximity to site Multiple rules composable with AND/OR logic in configuration Rules evaluated deterministically — identical inputs produce identical rankings Target selection Ranked candidate target list produced at each scan cycle Manual override of rule-selected targets supported Targets below all rule thresholds explicitly recorded as rejected, not silently dropped Logging and reproducibility Full record written at each cycle: evaluated cells, applied rules, scores, selected target, rationale Record format supports post-event analysis and downstream scientific workflows Integration Rule engine implemented as a module against a contract in contracts/ Output target JSON compatible with the radar_control module input contract Zero changes to existing pipeline modules Dependencies Tracking module producing stable cell identity across scans radar_control milestone for end-to-end closed-loop testing Out of scope Radar API delivery (see Radar API Integration — Adaptive Scan Target Delivery milestone) Dashboard visualization of candidate targets Priority High — required for science-driven adaptive scanning
No due dateAdapt currently produces scan target recommendations but has no mechanism to deliver them to an ARM radar. This milestone adds a radar_control module that sends selected scan targets to the ARM radar control API, closing the adaptive scanning loop.
No due date•0/2 issues closedAdapt's architecture must support pluggable modules and extensions via registration, but this is undocumented and untested end-to-end. This milestone produces living documentation that is validated by actually building and registering a working extension. Any friction or breakage encountered during that exercise is fixed before the milestone closes. Exit criteria Architecture and design documented for developers (contracts, module boundaries, registration pattern) Step-by-step guide for writing a new module or extension A reference extension implemented, registered, and tested end-to-end Any refactoring required to make the extension work cleanly is completed CI passes with the reference extension included
Due by May 15, 2027•1/4 issues closedAt least 60% unit test coverage for major pipeline modules and other components. Total coverage ~80% Coverage target: all public methods, edge cases, failure paths, and run() contract validation for every module. Exit criteria All items in issue tests: add missing unit test coverage across all pipeline modules are resolved No test uses stored binary fixtures All integration tests are labelled test_integration_* and excluded from default CI CI passes with coverage report enabled Priority Low — complete before beta release, not required for alpha
Due by February 28, 2027•3/3 issues closedTo establish a trustworthy CI baseline before any architectural work. No more feature development until tests are reliable and coverage is measured. Exit Criteria: - CI passes on every push to main with zero skipped tests - Codecov badge shows ≥ 60% coverage - Publish to PyPI and conda-forge
Due by December 15, 2026•10/10 issues closed**Goals:** 1. Assign a stable identity to each convective cell across radar scans, 2. Record how cells are born, evolve, split, merge, and die, and expose that lifecycle to the GUI. 3. Add any cell property needed to compute tracks in analysis module and save it in the repository. 4. create nomenclature for the cell_tables and properties and use them in API and GUI calls. 5. All tracking should work in time (not in frame counts). Consider missing scans and changing intervals between the scans. **Scope:** segmentation and projection are completed. Basic analysis is done, GUI is working. Tracking begins after all this is ready to use.
No due date•2/2 issues closed