Re: Q: Building a Weibull distribution of air temperatures based on local min, max and average?

Thanks for your reply Dean.

Maybe you have suggested better distibution types than the Weibull !!

My mission (impossible) is to reconstruct the set of observations
behind a local min, max and average temperature, without loss of to
much information. However I can live with a minor loss.

During summer and winter the average temperature is often displaced
from the middle, closer to either min or max, so the distribution must
be able to deal with this.

I am aware of the absolute lower limit, however local weather data
always specify a normal minimum temperature which I think of using as
the location parameter.

So if Weibull still is the better distibution, the question is then how
to estimate the shape and scale parameter based on local min, max and
average air temperature and the knowledge that we are dealing with air
temperatures specifically?

Best regard
Jakob Lundholm




dean ford skrev:
I would not recommend Weibull for temperatures. One problem with
Weibull if that it has an absolute lower limit and does not allow a
lower tail that keeps going to infinity.

If you set the minimum life to absolute zero, then it might make sense,
but I don't think that is what you want.
Weibulls are used for wind speed because wind speed cannot be less than
zero.

Would a normal or log normal distribution be a reasonable model?

Dean M. Ford


lundholm.gregersen@xxxxxxxxx wrote:
Hi,

Is it reasonable to believe, that monthly or yearly air temperatures
fit a Gaussian or more likely a Weibull distribution with 3 parameters?

Have anyone worked with the opposite, namely building a Weibull
distribution based on the local min, max and average?

Thanks in advance!

Best Regards
Jakob Lundholm

.

Relevant Pages

  • Re: Predicting using SAS-Survival analysis
    ... The problem with simply using the highest likelihood is that those ... one parameter, Weibull, logistic, log-normal, normal, and log-logistic ... are two-parameter distributions, and gamma has three. ... chi-square distribution with 1 degree of freedom, ...
    (sci.stat.math)
  • Re: Predicting using SAS-Survival analysis
    ... The problem with simply using the highest likelihood is that those ... one parameter, Weibull, logistic, log-normal, normal, and log-logistic ... are two-parameter distributions, and gamma has three. ... chi-square distribution with 1 degree of freedom, ...
    (sci.stat.edu)
  • Re: Distribution fitting
    ... > Peter Perkins wrote: ... >> Jan Van den Bulcke wrote: ... >>> like a distribution. ... >> which, if you plotted them against 1:n, look like a Weibull (or ...
    (comp.soft-sys.matlab)
  • Re: Computing derivative using finite differnce method
    ... Inorder to compute the asympotic covariance of the MLE of Weibull ... I took the data set: ... and the loglikehood function for the Weibull distribution fuction: ...
    (comp.soft-sys.matlab)
  • Computing derivative using finite differnce method
    ... Inorder to compute the asympotic covariance of the MLE of Weibull ... I took the data set: ... and the loglikehood function for the Weibull distribution fuction: ...
    (comp.soft-sys.matlab)