Re: GEOSTATS: variogram modeling
- Vijay Sathya (Vijay.Sathya@...)
Fri, 19 Sep 1997 11:42:29 +0200 wrote:
>air pollution modelling.A part of the uncertainity comes from themodels
>interpolation process.I've just finished analysing data sets for
>temperature in the domain and the variogram structure during the day is
>very different from that during the night.
>So if I intend to use kriging must I build and use different variogram
>for the night and day periods.??It depends. What would your maps eventually be for? This might
give you a clue as to whether you should combine day and night
measurements. I would gather that temperature distributions
for day and night periods are supposed to be very different anyway.
Would the temperature maps be for a Stealth pilot? A meteorologist?
A science exhibit? Would it be used for making far reaching decisions
concering rainfall predictions or the like? Litigation?
>Is second order stationarity a must in the variogram model used forkriging??
Yes, strict (variogram and covariance exists) or not very strict (intrinsic
hypotheses, or just the variogram). But of course, that does not
necessarily mean that your data are actually stationary. Maybe you think
there is a trend in there somewhere and would rather prefer to work with
(hopefully) stationary residuals. Or maybe you can't live with the bias
and would rather work with an IRF-k technique to derive a spatial
model. Sometimes another meteorologist or geologist would come
along and suggest that you break up the data into different
"stationary" pockets and work on these separately, because combining
them might introduce non-stationarity. There are a lot of factors to take
into account. Maybe you alreay know that there is non-stationarity but
intend to leave the variogram as it is by just using the early lag data.
Perhaps you think you can find a more or less "stationary" area of the
data and derive a variogram from this area. You can then krige with
limited search neighborhoods. Follow your common sense, there is
no hard or fast rule about this.
>I would also like to have some information on softwares for variogramRefer the software FAQ at http://java.ei.jrc.it/rem/gregoire.
*To post a message to the list, send it to ai-geostats@....
*As a general service to list users, please remember to post a summary
of any useful responses to your questions.
*To unsubscribe, send email to majordomo@... with no subject and
"unsubscribe ai-geostats" in the message body.
DO NOT SEND Subscribe/Unsubscribe requests to the list!