What it is good for - and what it is not

The corrected product moves the daily distribution to the gauge but inherits raw satellite timing (the r ≈ 0.35 ceiling, measured against CPC-UNI at daily resolution, native window, per-pixel spatial median then averaged over the 36 dekads). So the useful question is not "is it accurate?" but "does your application need the distribution or the calendar day?" Pick one:

The split tracks day-specific timing, not topic: flood, hydrology and agriculture each appear on both sides. A 25-year flood-frequency analysis is well-served; real-time flood nowcasting on a named date is not - the same field, judged on different axes.