[ncl-talk] problem of gamma fit and cumulative probability functions

[Xin Zhou]

Dear NCL users,

I'm twisting the SPI formula since the build in function dim_spi_n() does
not meet my needs. I meet a problem of gamma fit and cumulative probability.

The cumulative probability of gamma distribution that I calculated
(cdfgam_p) based on the output of gamma fit (dim_gamfit_n) does not make
sense. I suspect that the online documentations of these two functions are
not consistent, and the "shape" parameter in one function probably means
"1/shape" in another function. Has anybody run into the same problem?

Also, does anyone know where is the source code of dim_spi_n? My
calculation of SPI does not match the output of dim_spi_n exactly, although
the distribution seems reasonable.

Thank you,
Xin
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20150810/e9c4a25f/attachment.html