The idea is to transfer information through all nodes at the same level in the tree structure. However, the question is which dataset will be the referent to transfer the posterior information on the parameters to the other datasets.
A good idea is to estimate the Kolmogorov-Smirnov empiric distance to compare if the CDF are similar.
Two possible solutions for this :
Referent CDF :
A value for each water level. The CDF will be a straight line.
All gauging data before any segmentation.
Both of them have advantages and disadvantages.
The CDF used to compare will be taken by the data after segmentation. The Kolmogorov-Smirnov empiric distance will be the indicator to compare among all nodes at the same level to get the best dataset.
The idea is to transfer information through all nodes at the same level in the tree structure. However, the question is which dataset will be the referent to transfer the posterior information on the parameters to the other datasets.
A good idea is to estimate the Kolmogorov-Smirnov empiric distance to compare if the CDF are similar.
Two possible solutions for this :
Referent CDF :
Both of them have advantages and disadvantages.
The CDF used to compare will be taken by the data after segmentation. The Kolmogorov-Smirnov empiric distance will be the indicator to compare among all nodes at the same level to get the best dataset.