BowEcho v0.33.0

BowEcho v0.33.0 — "The WRF Laboratory Release"

BowEcho v0.33.0 turns WRF analysis into a first-class part of the application.
It adds a substantially deeper simulated-radar forward operator, gives raw WRF
work its own workspace, and promotes Formula Lab into a dockable tool that can
build diagnostics from WRF and other model runs already stored in BowEcho.

A much deeper WRF simulated radar

Open Windows > WRF and choose WRF simulated radar. Five complete recipes
make the expanded operator approachable without requiring users to understand
every interacting radar setting:

• Storm view (fast) produces readable textured reflectivity and velocity
  loops.
• Clean model truth keeps the model atmosphere free of virtual-instrument
  effects.
• Clean dual-pol adds polarimetric physics and propagation without noise,
  folding or blockage.
• Real radar (balanced) is the recommended practical S-band simulation.
• Maximum fidelity (slow) uses the full 27-point pulse-volume integration
  for a single frame or short loop.

The advanced controls remain independently editable after choosing a recipe.
BowEcho now supports center, 9-point and 27-point pulse-volume sampling in
linear reflectivity; hydrometeor terminal fall speed; pulse-volume spectrum
width; cumulative terrain-horizon blockage; timed rays; configurable beam and
pulse properties; range-dependent sensitivity; velocity folding; calibration
offsets; and optional real-radar presentation texture and clutter.

The scheme-aware bulk S-band dual-polarization path can emit ZH/REF, ZDR,
rhoHV, KDP, PhiDP, Doppler velocity, spectrum width, specific and integrated
attenuation, and intrinsic/corrected reflectivity and differential
reflectivity. It recognizes common Lin/WSM/WDM, Thompson, Morrison,
Milbrandt-Yau and NSSL bulk schemes. Inputs that cannot support an honest
polarimetric interpretation fall back to scalar REF/VEL with an explicit note.

Every generated elevation, radial, gate and moment enters BowEcho's ordinary
radar viewer. Simulated frames therefore use the same loops, readouts,
cross-sections, derived products and velocity tools as observed radar. CfRadial
export now carries the expanded moments plus scan timing, instrument,
calibration, model, microphysics and forward-operator provenance.

Fast radar experimentation

After building one simulated frame or a multi-file loop, use Refresh current
frame(s) to rerun the same WRF source set with the controls as they are set
now. No file picker or re-import is required. A folder selection remains a
stable session snapshot, so experimenting with physics cannot silently add new
files, and the completed refresh replaces the current radar loop in one step.

WRF is now its own workspace

Raw WRF work is no longer mixed into the general Models library. Windows >
WRF provides a focused workspace for quick WRF/NetCDF import, full wrf-rust
diagnostics, GDEX access, simulated radar and Formula Lab handoff. Models
remains the library and output surface for stored fields, plots and soundings.
Both windows share the same model backend and current selection, so results
move between them without duplicate imports or stores.

The WRF workspace keeps extensionless wrfout_* support across every domain,
multi-file and folder processing, size-aware confirmation for large jobs, and
the configurable severe/thermodynamic diagnostics suite. Simulated-radar
controls are organized around the task-oriented recipes first, with radar
location, scan geometry, physics, moments, presentation, instrument and
propagation tuning available below them.

Formula Lab across model workflows

Windows > Formula Lab now opens a first-class floating or docked workspace.
Its editor, source choice and options persist, evaluation runs in the
background, and the window remains usable independently of Models and WRF.

For stored runs, Formula Lab follows the current model/run/time selection and
discovers the fields that actually exist instead of assuming a WRF-only field
list. Model-aware quick starts adapt to compatible variables and units, while
the readiness view explains missing or incompatible inputs before evaluation.
Pointwise formulas therefore work with WRF, HRRR, GFS and other supported model
runs whenever their manifests provide the required fields.

Raw WRF mode remains available for formulas that require native grid metrics,
physical height, vector operations or horizontal calculus. This boundary is
shown directly in the UI: stored-model formulas cannot claim grid-aware
operations when the store does not contain the required geometry.

Formula results install into the shared Models workflow with provenance and
safe units. They can be inspected in the field viewer, placed on the radar map,
sent to the native plotter and styled with BowEcho's model color-table tools.

Scientific scope

This release's dual-pol kernel is a versioned bulk Rayleigh operator, not a
T-matrix solver. Frozen-particle orientation, resonance and scheme-native P3
properties remain future work. Timed scans sample one WRF model instant; they
do not interpolate the atmosphere between forecast files or claim to reproduce
a numbered operational VCP.

Windows, macOS and Linux ship the same WRF, Formula Lab, model import and native
file-dialog workflows. Linux native dialogs continue to require GTK 3.24.