The dataset corresponds to summer temperature maxima taken over the period from August to April inclusive, recorded between 1961 and 2010 at 90 stations arranged on a 0.15 degree grid in a 9 by 10 formation.
The first maximum is taken over the August 1961 to April 1962 period, and the last maximum is taken over the August 2010 to April 2011 period.
The object heatdata contains the core of the data:
A \(50 \times 90\) matrix containing the \(50\) summer maxima at the \(90\) locations.
A \(90 \times 2\) matrix containing the site locations in latitude-longitude, recentered (means subtracted).
A \(90 \times 2\) matrix containing the site locations in eastings-northings, recentered (means subtracted).
A \(50 \times 90\) integer matrix indicating the ``heatwave'' number associated with each summer maximum. Locations on the same row with the same integer indicate maxima arising from the same heatwave, defined over a three-day window.
A \(90 \times 2\) matrix containing the site locations in latitude-longitude, on the original scale.
A \(90 \times 2\) matrix containing the site locations in eastings-northings, on the original scale.
A \(50 \times 90\) matrix containing the \(50\) summer maxima at the \(90\) locations, standardized to the unit Frechet scale.
Standardisation to the unit Frechet scale is performed as in Beranger et al. (2021) by fitting the GEV distribution marginally, using unconstrained location and scale parameters and a shape parameter specified as a linear function of eastings and northings (in 100 km units). The resulting estimates are stored in the objects locgrid, scalegrid, and shapegrid, which are \(10 \times 9\) matrices.
Details about the study region are given in mellat and mellon, vectors of length \(10\) and \(11\), which provide the latitude and longitude coordinates of the grid.
Beranger, B., Stephenson, A. G. and Sisson, S. A. (2021). High-dimensional inference using the extremal skew-\(t\) process. Extremes, 24, 653-685.
image(x = mellon, y = mellat, z = locgrid)
points(heatdata$sitesLLO, pch = 16)
Run the code above in your browser using DataLab