Process — a radar analysis workbench that outlives its operating system

The problem. Vesta | Proceso is the processing and visualisation program of the Radar Centre at Cuba's Institute of Meteorology, written in Delphi/VCL by Wilfredo Pozas. Forecasters have used it for years, and it works — but it is tied to Windows and to a toolchain whose licence and future are outside the institution's control. A tool that outlives its platform has to be rewritten before the platform forces the issue, not after.

The constraint that is not negotiable. It must keep ingesting raw radar data. Deriving products from already-derived Level III output would be cheaper to build and would quietly cap what the tool can ever answer: once a product is computed, the choices behind it — thresholds, interpolation, the elevation it came from — are no longer yours to revisit.

The decision. One codebase, two targets. The same Pascal core compiles to a desktop application and runs in the browser, and the browser build carries no backend at all — the volume is read client-side. Nothing to deploy, nothing to keep running, nothing to pay for while it is idle. Reach, without an operations bill.

Why it is not part of the LAMULA chain. It reads the same archive the chain writes, but it is a separate tool, not a stage. The chain ends at a viewer that consumes Level III products generated upstream; Process starts from the raw volume and computes its own. Shown as a stage it would suggest a dependency that does not exist in either direction.

What it is not. An analytical instrument rather than a display: it opens empty, and shows something once an observation is loaded. The screenshot below is a real Level-II volume, named in the caption — the same file can be pulled from NOAA's archive and compared.