anmn_temp_gridded_product.py - Use nominal depth when no depth available

[ggalibert]

Sometimes it is not possible to compute any DEPTH variable for a dataset that contains valuable TEMP data. In this case the script should be able to consider NOMINAL_DEPTH instead.

[mhidas]

Is that a good idea? It may be ok if you're just looking at a single timeseries and you keep in mind that the depth is only nominal, i.e. at times it may be totally off!

However, if you then combine the temperatures from this instrument into a gridded product where some input values are "placed" in the grid based on nominal depths, while others using measured depths, you may introduce artifacts that are hard to identify and hard to flag as being potentially wrong.

[bpasquer]

That's why originally these data were excluded. But additional checks can be made to ensure that the product for this deployments is consistently using temperature at nominal depth.

[ggalibert]

This is just to be able to accept a deployment that doesn't have any P sensor on the mooring. Nominal depth is better than nothing and best we can say. Still better than having no gridded data available in my opinion.

The vertical accuracy lost with using nominal depth might be negligeable/comparable to the vertical bin size?

As soon as there is at least 1 P sensor, there should be an inferred DEPTH for any other instrument.

[mhidas]

👍 It would be good to include some of this information in the output file's metadata so that the user has some idea about the expected accuracy of the depth values.

[lbesnard]

I guess we can close this now @ggalibert

[ggalibert]

Not yet, let's keep of record of this for now.

[ocehugo]

Code deprecated, close!? @ggalibert @lbesnard

[lbesnard]

I'd love to