boshek / limnotools Goto Github PK

Encountering some instances where the depth of the shallowest segment is shallower that the specified mean depth meaning the algorithm is extending beyond the specified depth range. Will add data to make this clear.

I think we can remove the crufty old FORTRAN history from the R code. I'll take this on and submit a pull request, I guess.

I want to write a fake data generator for testing and I want some advice on the parameters. I presume that the water stratifies into three zones: mixing zone, thermocline, and whatever is below that. The mixing zone has zero or small temperature change, the thermocline occurs at some depth and has a large temperature change across its extent, then the temperature changes more slowly down to the deepest sample. Here are the proposed parameters:

• surface temperature
• temperature at the center of the thermocline
• slope of the thermocline in degrees/meter
• slope of the temperature in the deeper water.
• change of depth between samples
• depth of deepest sample
  From these parameters I can work out the entire curve. Then I can add various random variabilities, like noise on the temperature and depth, slow fluctuations in the values using additional parameters to control them.

So that's what I think - how would you, the actual limnologist here, like to see it?

The TODO suggests replacing an error signal return of -99 with a crash but I'm not sure that this is the right thing to do. I don't like the way it works now because -99 has no intrinsic meaning, it just shows up on the report and you the human reader are just supposed to intuit that "something went wrong" and ignore the rest of the report. That is seriously bogus, IMHO. What I think should actually happen here is that the report should just be replaced with a detailed explanation of why the report could not be produced. In order to accomplish this by_s_m needs to be modified to allow an exit that conveys the error message, whatever it is. Perhaps it should always return a list containing an error code, error explanation, and data, if no errors.

Run styler on the code

Use @export to define which are export. Probable functions available to the user: wtr_layer() and wtr_segment().

original getxxnorm fortran code normalizes by the top and bottom values in the profile. This should likely be normalized by the max() and min() values of the profile removing the possibility of negative normalization coefficients.

Need some isothermal looking data to test

change nn in higher functions from desired length of output to match the length of the input. The length of the output does not appear after a length of approximately 300. This would making any of these higher level functions throw an error is nn is lower than 300.