BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Persistence in Solar Physics - Karen Meyer (Dundee)
DTSTART:20241014T130000Z
DTEND:20241014T140000Z
UID:TALK221914@talks.cam.ac.uk
CONTACT:Roger Dufresne
DESCRIPTION:Persistence\, or long memory\, is of longstanding interest in 
 solar physics\, having first been identified in time series of sunspot num
 bers in the seminal paper by Mandelbrot and Wallis (1969): “Some long‐
 run properties of geophysical records”. They used a method called Rescal
 ed Range Analysis (R/S) to determine a Hurst exponent\, H=0.93\, which is 
 indicative of strong persistence. It has since been suggested that for sun
 spot numbers\, and indeed most times series of solar quantities\, R/S is n
 ot an appropriate method for estimating persistence due to the non-station
 ary nature of the time series. Detrended fluctuation analysis (DFA) has be
 en proposed as a more suitable method for estimating persistence\, and has
  since been widely used in the analysis of solar and geo-physical time ser
 ies. However\, DFA is known to introduce uncontrolled bias and is in fact 
 inappropriate for non-stationary processes (Bryce & Sprague\, 2012).\n\nHe
 re\, we assume an alternative class of long-memory models\, more commonly 
 found in statistics and econometrics: fractionally integrated processes. W
 e revisit solar physics time series such as sunspot number and total solar
  irradiance with more robust estimators\, and identify higher persistence 
 than previous studies\, as well as persistence over timescales significant
 ly shorter than previously identified.\n\nWe also consider persistence in 
 time series of quantities derived from solar physics simulations\, demonst
 rating that these simulations capture the memory structure that is present
  in the observational input data. Further\, we provide an algorithm for th
 e quantitative assessment of simulation burn-in: the time after which a qu
 antity has evolved away from its arbitrary initial condition to a physical
 ly more realistic state.
LOCATION:MR14 DAMTP and online
END:VEVENT
END:VCALENDAR