Reduction in soundings should impact all forecasts, but perhaps not equally given different data assimilation methods. Not sure how you’d “normalize” it so to speak, but I think comparing forecast accuracy between winter 24-25 and 25-26 would be a very interesting study.