Model card — western Mediterranean cloud recreation
Diodato, Seftigen & Bellocchi, Research 8:0606 (2025), doi:10.34133/research.0606

The authors did not publish a downloadable annual CSV. This file is the method
this lab used. It is not a digitisation of their Fig. 6H.

WHAT IS A REAL SERIES
  Mann et al. 2009 AMO, 500–2006, file amoall.txt, with the archived 2σ band.
  After 2006 the atlas splices van Oldenborgh ERSST v5 AMO with a +0.34 °C offset,
  the 1980–2006 mean difference. That offset is a definition change, not a measurement.
  Gray et al. 2004 AMO, 1567–1990, comparison only.
  Steinhilber et al. 2009 TSI anomaly (years BP before 1950).
  SILSO sunspot number, annual mean of months, from 1749.
  Crowley & Unterman 2013 NH sulfate, downloaded from NCEI. Eruption dots on the
  atlas use events ≥ 12 kg km⁻² from this file. The cloud equation was NOT recomputed
  from it. The volcanic term inside the curve is a short encoded list and is worth
  only about 0.00055 × sulfate in cloud-fraction units.
  HadCRUT5 global annual temperature, 1850–2025, downloaded from the Met Office.
  It is published here and not substituted for the paper's GTA25 term.

WHAT IS MODELED
  Western-Mediterranean PDSI: a prescribed envelope (dry medieval, wet Little Ice Age,
  a hard post-1970 drying ramp) plus −1.15 × the Mann AMO, an AR(1) of 0.55, and
  deterministic noise. It is not the Old World Drought Atlas grid.
  SRI: 0.9 + 0.28 × PDSI, plus extra storminess in 1250–1850 and 1570–1650.
  GTA: an energy-balance sketch from TSI, volcanoes, 0.35 × AMO, and an anthropogenic
  ramp after 1850. GTA25 is the 25th percentile of an 21-year window of that sketch.
  TCC: the equation below, then a light AR(1) weather term, then a hand-placed
  centennial shape of about ±0.02 so the curve rises toward 1600 and falls after 1975.
  The 1935–2022 "observed" series is the equation plus noise. It is not CRU TS 4.06.

EQUATION, IN CLOUD-FRACTION UNITS
  WCI = 0.300 × PDSI × (2.00 + SRI^0.500)
  TCC = 0.452 + 0.0124 × WCI − 0.024 × AMO − 0.018 × GTA25 + 0.00055 × volcano
  The published modifiers are α=0.30, β=2, σ=0.5, γ=−3, η=1. γ is scaled by 0.008
  so the AMO term is in cloud-fraction units. That scale is this lab's, not a number
  the authors printed as 0.008.

THE AMO IS USED THREE TIMES
  Once as the γ term. Again inside the synthetic PDSI (−1.15 × AMO). Again inside
  the synthetic GTA (0.35 × AMO). The "count once" slider removes the second and third.
  That double count is this recreation, not a second sum written in the 2009 paper.

SLIDERS
  They add a precomputed cloud-fraction change. They do not refit A or γ and they
  do not overwrite lab-annual.csv.

NOT INGESTED
  CRU TS 4.06 cloud grid. CM-SAF CLARA grid. OWDA netCDF. UCAR scPDSI grid.
  ERA5 daily rainfall. The Carcaixent diary. PAGES 2k temperature.