Daafip / eWaterCycle-DA

Data assimilations functions to be used in combination with Hydrological modeling system eWaterCycle
https://ewatercycle-da.readthedocs.io/en/latest
Apache License 2.0
0 stars 0 forks source link

Particle Filter implementation: Observational noise #37

Open Daafip opened 5 months ago

Daafip commented 5 months ago

In the current implementation of the particel filter no noise is added to the observations. This should be the case as the observation also contains errors. In literature this is often sampled from a zero-mean normal distribution with a variance of: $(eta * Q_{obs})^2$ where eta is often chosen as 0.1 or 0.01. See the table obtained from a literature review below.

image

Before using the particel filter for any real use case, this should be added

Daafip commented 5 months ago

This should be the case as the observation also contains errors.

Correction: Could be the case. If real data is used, no noise is added. Only to synthetic runs this is added. The table can be useful so will leave this issue open