Hi,
I’m performing bias correction of my SEAS5 dataset by empirical quantile mapping, but I’d like some guidance to understand the process. This is my data:
-My reference is ERA5 (2016-1993)
-My hindcasts, SEAS5 (2016-1993) with 25 ensemble members
-My forecast of interest, SEAS (start month=March 2024) with 51 ensemble members.
-Is SEAS5 distribution calculated for each start month and each lead month?
-Is the bias correction of the hindcast applied to individual ensemble members first, and then, averaged?
-Is the average over the ensemble members of the 25-member-hindcast applied to correct the 51-member-forecast?
-The first 25 ensemble members of the forecast (real-time forecasts) have the same initialization conditions of the 25 ensemble members of the hindcast?. If so, can I apply bias correction member by member from hindcast to forecast? This would impply bias correction of 25 out of the 51 ensemble members of forecast in order to avoid neutralization of extreme information which is inherent in some ensemble members.
Thank you
Angela