0

I am trying to forecast hourly wind speed (HWS) data in Trinidad and Tobago and I have read in the literature that "Direct application of stochastic models (ARMA & ARIMA models) to HWS series is not possible due to their non- Gaussian distribution. To resolve this problem, and with the aim of simulating HAWS, Daniel and Chen (1991) have proposed a transformation of HAWS using the Dubey method (1967). In fact, Dubey showed that, for shape parameters close to 3.6, the Weibull distribution has a shape similar to the Gaussian one. This approach is reasonable because a Weibull random variable to the power of x is also a Weibull random variable." the review says that x=k/3.6 where k is the form factor of the weibull distribution of the observed hourly wind speed.

Can anybody assist me in how to go about this? My data is hourly wind speeds for 7 years, so how do I determine k?

oneloop
  • 598
  • 3
  • 14
L N
  • 43
  • 1
  • 7
  • Could you please link Daniel and Chen (1991)? It's not clear to me from your description what they're proposing exactly. – oneloop Mar 30 '18 at 17:44
  • 1
    I believe here's the answer: https://stats.stackexchange.com/a/60530/188779 – oneloop Mar 30 '18 at 17:49
  • @oneloop - IIRC, they estimate the shape parameter ("form factor" in the OP) $k$ of the Weibull, then raise the observed wind speeds to the $3.6/k$ power, which transforms them to a Weibull with shape parameter $=3.6$, which is close to a Normal distribution. – jbowman Mar 30 '18 at 18:58

0 Answers0