MOMFIT
: Moment-based calibration of point process models for single-site rainfall sequencesThis software is designed to fit stochastic single-site rainfall
models to data using a method of moments. Specifically, given a set of empirical properties of some observed rainfall time series, models can be calibrated so as to reproduce these properties as closely as possible.
This calibration is often a challenging numerical optimisation problem; this software provides a reaosnably stable solution, building on the optimisation routines available within the R
environment (see below). It also provides assessments of parameter uncertainty in the fitted models. For an overview of the types of models for which the software is designed, along with the methods that are usually used to calibrate them, see the following references:
The software is written to run in the R
environment (version 1.7.0), which is freely available from http://www.R-project.org/. Installations are available for
most operating systems, and most R
code is completely portable between systems. We have used the software on Windows
and Linux
platforms. The routines run substantially faster (a factor of 2 or 3) under Linux
, but otherwise we have not found
any differences between platforms.
The package consists
of a zipped archive containing the R code, a a user guide and some relevant technical reports in PDF
format, a specimen data set and some example scripts illustrating the use of the software. If using a Unix
machine, I suggest you download the file to
an empty directory, saved under the name momfit.zip
, and then
type
unzip momfitto extract the files from the archive.
Windows
users can unzip the file directly into an empty directory, for example by double-clinking in Windows Explorer
. Once you have unzipped the archive, start with the
manual (file manual.pdf
, although if you needed to be told that then probably you shouldn't be attempting this ...).
To download the software, click on the screenshot: