BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260421T090512Z
LOCATION:Bldg. 8 - Room B 102
DTSTART;TZID=Europe/Stockholm:20260630T142500
DTEND;TZID=Europe/Stockholm:20260630T150500
UID:submissions.pasc-conference.org_PASC26_sess148_msa243@linklings.com
SUMMARY:Energy-Adjusted Skill Score: A Community Roadmap for Energy-Aware 
 Earth System Science
DESCRIPTION:Ioana Colfescu (University of St Andrews, National Centre for 
 Atmospheric Science); Kara Moraw (University of Edinburgh); Chen Lu (Unive
 rsity of St Andrews, National Centre for Atmospheric Science); and Michael
  Bareford (University of Edinburgh)\n\nEarth system modelling has long opt
 imised for predictive skill alone, treating computational energy as an inv
 isible cost. Model generations improved the performance index by roughly 7
 0% from CMIP1 to CMIP6, yet energy consumption per unit of scientific outp
 ut grew by more than two orders of magnitude over the same period. AI weat
 her forecasters now achieve comparable medium-range skill at inference cos
 ts four to five orders of magnitude lower than a CMIP6 ensemble cycle. Thi
 s divergence makes energy a first-class scientific variable that can no lo
 nger be omitted from evaluation frameworks.\nWe introduce the Energy-Adjus
 ted Skill Score (EASS = χʰ/E, kWh⁻¹), a hardware-agnostic metric unifying 
 skill and energy into a single auditable figure of merit, and present the 
 first published EASS estimates spanning CMIP3 to current AI forecasters. W
 e identify four mutually reinforcing efficiency levers — simulation design
  and reuse, GPU and wafer-scale hardware acceleration, cloud-native data w
 orkflows, and physics-constrained machine learning — that can shift the sk
 ill–energy frontier by orders of magnitude. To operationalise adoption we 
 propose the Net-Zero Readiness Level scale (NZRL 1–9), setting concrete 20
 30 targets for CMIP7 and major operational centres. Reporting EASS alongsi
 de conventional skill metrics should become standard practice, requiring n
 o new infrastructure.\n\nDomain: Computational Methods and Applied Mathema
 tics\n\nSession Chairs: Nick Brown (EPCC); Michele Weiland (EPCC, The  Uni
 versity of Edinburgh); Adrian Jackson (The University of Edinburgh); and R
 ui Apostolo (EPCC)\n\n
END:VEVENT
END:VCALENDAR