When I look at the output frequency analysis .csv, the results make sense, and extend all the way to the 1/500 AEP storm.
However, when I look through the .nc files for the transposed storms in each realization, they only extend through year 50.
I have not yet been able to figure out what is causing this, but I did save the array "whichx" to disk and found that it had shape (22, 500, 20, 1) whereas I assume it should have shape (220, 500, 20, 1). The 10x reduction in the first dimension would explain the 10x reduction in the number of storms written to disk.
I've been doing test runs for RainyDay with the following characteristics:
Storm duration: 24 hours Catalog size: 220 storms Years: 500 Realizations: 20 Storms per year: 5
When I look at the output frequency analysis .csv, the results make sense, and extend all the way to the 1/500 AEP storm.
However, when I look through the .nc files for the transposed storms in each realization, they only extend through year 50. I have not yet been able to figure out what is causing this, but I did save the array "whichx" to disk and found that it had shape (22, 500, 20, 1) whereas I assume it should have shape (220, 500, 20, 1). The 10x reduction in the first dimension would explain the 10x reduction in the number of storms written to disk.
I've attached a .yml of my virtual environment (as .txt), my source code files (I had to make some minor adjustments to get RainyDay to run, e.g. dealing with deprecation errors; attached as .txt), the .json input file, and a subset of my storm catalog. StormCatalog.zip RainyDayInputs.json RainyDay-env.txt RainyDay_functions.txt RainyDay_Py3.txt Baer_SST_Inputs.zip