Don't lose sight of "what's good enough" in the
thicket of detail accuracy.
I've gotten a lot of mileage out of simple current
sources to model photocurrent, when these are
an "abnormal condition" to be tested and designed
against. But I also recognize that in other technologies
some assumptions are not as valid, as where I was
working at the time.
If this diode is to be a critical element of the signal
chain and you have multiple dimensions of care-
abouts which involve the photoresponse, then you
will be responsible for due care and eventual
accuracy. Can you do it? Sure, with enough study
and time and presuming adequate comprehension.
But I recommend you budget that time according
to what matters most (and take the advice to find
vendor models, anybody's models, to use as a
start or a substitute or as a component selection
factor (this last, to beat vendors' applications support
people upside, to induce cooperation).