Unrealistically different results

[shadekul]

We tried WRF over same domain with all other configuration keeping same using IFS and ICON data for init and LBC but output rainc+rainnc (24)hrs differ by very large amounts. Would anyone suggest improvement ? Below is an Example


fcst_hour,valid_time,rain_max,rain_mean
ICON
24,2026-07-21 00:00:00,498.5,52.6
48,2026-07-22 00:00:00,496.5,49.9
72,2026-07-23 00:00:00,479.7,51.7
96,2026-07-24 00:00:00,452.1,48.8
120,2026-07-25 00:00:00,480.6,51.2
144,2026-07-26 00:00:00,484,54.2
168,2026-07-27 00:00:00,422.3,54.6
IFS
48,2026-07-21 00:00:00,139.2,19.5
72,2026-07-22 00:00:00,143,18.6
96,2026-07-23 00:00:00,173.9,22.6
120,2026-07-24 00:00:00,162.9,22.5
144,2026-07-25 00:00:00,142.8,18.9
168,2026-07-26 00:00:00,117.3,17.3
192,2026-07-27 00:00:00,129.3,18.5

Any help be appreciated.
Thanks.

 

[meteoadriatic]

Is that IFS Open Data? It has too coarse vertical data resolution at low levels and my theory is that because of that it lacks moisture (yes, generally too dry), thus WRF produces negative precipitation biases. I see this all the time, and I don't know the solution. Maybe someone will help both of us.