shapley implementation for track perturbation

[saeed-moghimi-noaa]

shapley implementation

[WPringle]

See https://github.com/WPringle/Storm-Surge-Mesh-Subsetting/commit/50cb7f9f5e7dcd128bce8796884e019a2a4bdcfc for the along-track perturbation work.

[WPringle]

@zacharyburnettNOAA @saeed-moghimi-noaa Example for along-track perturbed Hurricane Florence 2018:

[WPringle]

@zacharyburnettNOAA Note also a change in write fort.22 routine of adcircpy in the PR. I want it to "round" the floats instead of change to "int". (e.g., I want 34.67N to be 34.7N not 34.6N). But I am getting a conflict with a recent change you made to use a "convert_value" function.

[WPringle]

@zacharyburnettNOAA @saeed-moghimi-noaa I also want to mention that I changed the Rmax behavior to go in the opposite direction. Previously it was adding the error but I think it should subtract the error.

Example: A storm with small forecasted Rmax is more likely to be under predicted in size, i.e., the error in Rmax will be negative most of the time. So for perturbation, most of the time we want to increase the size of the small storm, therefore need to subtract the negative error.

[ghost]

https://github.com/JaimeCalzadaNOAA/adcircpy/pull/60 is now merged

[WPringle]

See this commit for the cross-track perturbation implementation: https://github.com/WPringle/Storm-Surge-Mesh-Subsetting/commit/239421a05bc8398e45684e005efcb30fdeae7822

Example:

[saeed-moghimi-noaa]

Looks very good @WPringle

[ghost]

wow, very good work!