Dear All,
I have a number of environmental time series which appear to have some
problems with homogeneity, either in mean, variance, or both over the
length of the record. Some of this may well be the result of station
movements, changes in instrumentation, etc, and it is these change points
I need to detect. Whilst in principle it is better to contact the data
providers directly, in practice, this approach is often less than
fruitful. What I wish to obtain is a reference/set of references to
procedures/tests which may be considered standard/state of the art and
any drawbacks that I should be aware of. At the moment, I am using the
Mann-Kendall test which I am supplementing with regression based dummy
variable procedures to identify changes, together with significance.
Any suggestions would be gratefully appreciated.
Best regards
Paul
**********************************************
Dr. Paul Rockett
Benfield Greig Tel: +44 (0)20 7522 4027
E-mail: [log in to unmask]
|