Signal Construction and Effective Climate Memory in Arctic, Atlantic, and Pacific Climate States (Peer Review)

Tatlı H.

PHYSICS AND CHEMISTRY OF THE EARTH, cilt.1, sa.1, ss.1-20, 2027 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 1 Sayı: 1
  • Basım Tarihi: 2027
  • Dergi Adı: PHYSICS AND CHEMISTRY OF THE EARTH
  • Derginin Tarandığı İndeksler: Scopus, Science Citation Index Expanded (SCI-EXPANDED), Chimica, Compendex, Geobase, INSPEC
  • Sayfa Sayıları: ss.1-20
  • Çanakkale Onsekiz Mart Üniversitesi Adresli: Evet

Özet

synoptic variability, trends, seasonality, and slowly evolving ocean–atmosphere

dynamics. This study investigates how signal construction, particularly the treatment of

synoptic variability and detrending, conditions fractional-order memory parameters and

the diagnostic performance of stochastic climate models. A continuous-time statespace

framework is developed for Arctic sea ice, Subpolar Atlantic heat content, and

the Equatorial Pacific ENSO state (SST–thermocline), with slowly varying state

variables derived from ORAS5 and atmospheric forcings obtained from ERA5.

Regularized Padé logarithmic approximation is used to derive continuous-time

operators for four signal constructions: raw, deseasonalised, detrended, and highfrequency

anomaly. Model behavior is evaluated using a multi-constraint diagnostic

framework combining the Unscented Kalman Filter, autocorrelation structure, power

spectra, and variance retention, comparing a classical stochastic climate model with a

Diethelm-type fractional formulation. The inferred fractional orders vary substantially

(0.50–0.94) across variables and signal constructions. Equatorial Pacific thermocline

depth exhibits stronger fractional dependence than SST, indicating that distinct

memory representations may be required for coupled ENSO components. The

classical formulation generally provides a more balanced representation of temporal

structure and variance, whereas the fractional model offers a limited advantage for

deseasonalised SST but systematically suppresses synoptic-scale variance,

particularly in Arctic sea ice. These results indicate that fractional climate memory is

not an intrinsic, preprocessing-invariant property, but depends on the interaction

between slow-state dynamics, atmospheric forcing, signal construction, and diagnostic

criteria.