BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//www.bayceer.uni-bayreuth.de//NONSGML kigkonsult.se iCalcreator 2.41.92//
CALSCALE:GREGORIAN
METHOD:PUBLISH
UID:66623862-6164-4232-b365-323263366634
X-WR-CALNAME:BayCMS-Export
X-WR-CALDESC:This is a calendar exported from BayCMS
X-WR-TIMEZONE:Europe/Berlin
BEGIN:VTIMEZONE
TZID:Europe/Berlin
TZUNTIL:20070325T010000Z
BEGIN:STANDARD
TZNAME:CET
DTSTART:20041031T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
RDATE:20051030T030000
RDATE:20061029T030000
END:STANDARD
BEGIN:DAYLIGHT
TZNAME:CEST
DTSTART:20050327T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
RDATE:20060326T020000
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:www.bayceer.uni-bayreuth.de-bayceer-t53293id
DTSTAMP:20260805T181454Z
DESCRIPTION:With the global proliferation of wind power\, accurate short-te
 rm\nforecasts of wind resources at wind energy sites are becoming\nparamou
 nt.  Regime-switching space-time (RST) models merge\nmeteorological and st
 atistical expertise to obtain accurate and\ncalibrated\, fully probabilist
 ic forecasts of wind speed and wind\npower.  The model formulation is pars
 imonious\, yet takes account of\nall the salient features of wind speed: a
 lternating atmospheric\nregimes\, temporal and spatial autocorrelation\, d
 iurnal and seasonal\nnon-stationarity\, conditional heteroscedasticity\, a
 nd non-Gaussianity.\nThe RST method identifies forecast regimes at the win
 d energy site and\nfits a conditional predictive model for each regime.  G
 eographically\ndispersed meteorological observations in the vicinity of th
 e wind farm\nare used as off-site predictors.\n\nWe applied the RST techni
 que to 2-hour ahead forecasts of hourly\naverage wind speed near the State
 line wind farm in the US Pacific\nNorthwest.  In July 2003\, for instance\
 , the RST forecasts had\nroot-mean-square error (RMSE) 28.6% less than the
  persistence\nforecasts.  For each month in the test period\, the RST fore
 casts had\nlower RMSE than forecasts using state-of-the-art vector time se
 ries\ntechniques.  The RST method provides probabilistic forecasts in the
 \nform of predictive cumulative distribution functions\, and those were\nw
 ell calibrated and sharp.  The RST prediction intervals were\nsubstantiall
 y shorter on average than prediction intervals derived\nfrom univariate ti
 me series techniques.\n\nThese results suggest that quality meteorological
  data from sites upwind of wind farms can be efficiently used to improve s
 hort-term forecasts of wind resources.  It is anticipated that the RST tec
 hnique can be successfully applied at wind energy sites all over the world
 .\n\nJoint work with Kristin Larson\, Kenneth Westrick\, Marc Genton\nand 
 Eric Aldrich
DTSTART;TZID=Europe/Berlin:20050623T161500
DTEND;TZID=Europe/Berlin:20050623T235959
LOCATION:H6
SUMMARY:Prof. Dr. Tilmann Gneiting\, University of Washington\, Seattle: Ca
 librated probabilistic forecasting at the Stateline wind energy center: Th
 e regime-switching space-time time (RST) technique
TRANSP:TRANSPARENT
END:VEVENT
END:VCALENDAR