updating metadata for stream forecast #13
Conversation
| @@ -63,16 +63,16 @@ metadata: | |||
| #species in a community model, number of age/size classes in a population model, | |||
| #number of pools in a biogeochemical model. | |||
| initial_conditions: | |||
| status: absent #options: absent, present, data_driven, propagates, assimilates | |||
| complexity: 0 #How many models states need initial conditions | |||
| status: assimilates #options: absent, present, data_driven, propagates, assimilates | |||
There was a problem hiding this comment.
If running this model with EnKF, then this should be assimilates
| status: absent #options: absent, present, data_driven, propagates, assimilates | ||
| complexity: 0 #How many models states need initial conditions | ||
| status: assimilates #options: absent, present, data_driven, propagates, assimilates | ||
| complexity: 1 #How many models states need initial conditions |
There was a problem hiding this comment.
we have one state variable (temperature) with initial conditions for this model
| propagation: | ||
| type: ensemble #How does your model propogate initial conditions (ensemble or MCMC is most common) | ||
| size: 2000. #number of ensemble or MCMC members | ||
| size: 31 #number of ensemble or MCMC members |
There was a problem hiding this comment.
Based on the model output, there are 31 ensemble members, not 2000
| reference: "NA" #reference for assimilation method | ||
| complexity: 4 #number of states that are updated with assimilation | ||
| type: EnKF #description of assimilation method | ||
| reference: https://aslopubs.onlinelibrary.wiley.com/doi/full/10.1002/lom3.10302 #reference for assimilation method |
There was a problem hiding this comment.
If using EnKF to run the model, then the url is an appropriate citation
| @@ -85,17 +85,17 @@ metadata: | |||
| #model, this would be the number of climate inputs (temperature, precip, solar | |||
| #radiation, etc.). | |||
| drivers: | |||
| status: present #options: absent, present, data_driven, propagates, assimilates | |||
| status: propogates #options: absent, present, data_driven, propagates, assimilates | |||
| complexity: 1 #How many drivers are used? | |||
There was a problem hiding this comment.
only using air temperature, correct?
| @@ -85,17 +85,17 @@ metadata: | |||
| #model, this would be the number of climate inputs (temperature, precip, solar | |||
| #radiation, etc.). | |||
| drivers: | |||
| status: present #options: absent, present, data_driven, propagates, assimilates | |||
| status: propogates #options: absent, present, data_driven, propagates, assimilates | |||
There was a problem hiding this comment.
This is changed to propagates because the model incorporates driver uncertainty (the ensembles) into the model predictions of stream temperature
| # propagation: | ||
| # type: ensemble #how does your model propogate parameter uncertainity? | ||
| # size: 2000 | ||
| # #Leave everything below blank UNLESS status = assimilates | ||
| # assimilation: | ||
| # type: refit | ||
| # reference: "none" | ||
| # complexity: 4 |
There was a problem hiding this comment.
don't need these lines since the parameters are present but not drawn from a distribution / propogated into the forecasts. The may be data driven , but I'm not sure if the model was fit to the POSE data before running the forecast. If so, change this to data_driven
| covariance: FALSE #TRUE OR FALSE | ||
| type: EnKF | ||
| reference: https://aslopubs.onlinelibrary.wiley.com/doi/full/10.1002/lom3.10302 | ||
| complexity: 1 |
There was a problem hiding this comment.
should be same size as the number of states we're updating, which is just water temperature
| type: EnKF | ||
| reference: https://aslopubs.onlinelibrary.wiley.com/doi/full/10.1002/lom3.10302 | ||
| complexity: 1 | ||
| covariance: TRUE #TRUE OR FALSE |
There was a problem hiding this comment.
if using the EnKF, then this is TRUE
| @@ -185,16 +185,16 @@ metadata: | |||
| #match the dimensionality of the initial_conditions unless there are state | |||
| #variables where process error is not being estimated or propagated | |||
| obs_error: | |||
| status: absent #options: absent, present, data_driven, propagates, assimilates | |||
| complexity: 2 #Leave blank if status = absent | |||
| status: data_driven #options: absent, present, data_driven, propagates, assimilates | |||
There was a problem hiding this comment.
will either be present or data_driven if using the EnKF. I set to data_driven because I think the observation uncertainty was guided by reported thermistor uncertainty
|
Hey Jake, this is not the latest version of the model. @Muhammed-Shikhani has made some code changes and hasn't updated here yet. |
|
Yes this is related to #12 |
I'm basing these changes on the model used is this one, and it is run by the ensemble Kalman filter here. If there are other models / assimilation methods used, let me know.
See specific comments below.