PR#89 has added the capability to retrieve observational data from the preBUFR jason3 files while using the metplus pb2nc tool.
Plotting the wind speed results and comparing to the Quality Controlled Jason3 data from AODN source shows the need for some filtering in the data.
Some ideas to consider could be:
Rolling Median (Spike) Filter: Implement a window-based check (e.g., window size 5).Flag/Remove points where $|WSP_i - \text{median}(WSP_{window})| > \text{Threshold}$ (e.g., 5 m/s).
Or
Physical Bounds Clipping: Hard-clip values below 0.5 m/s (sensor noise floor) and above 30 m/s (unless in a known extreme weather event).