Where the sixfold gap comes from
The quasi-global 6-month Standardized Precipitation Evapotranspiration Index, computed for 1981-2024 with three Penman-type and three energy-based estimates of potential evapotranspiration, gives a drying trend of -0.0084 per year in the first family and -0.0014 per year in the second, and the confidence intervals do not overlap. The trend in the share of land under drought splits the same way: 0.2406 per cent a year against 0.031 per cent a year, a ratio of 7.76. Within each family the individual formulas cluster together; between the families they separate.[1]
That pattern places the source of the disagreement in the structure of the formula, beyond any single implementation. Penman-type estimates compute evaporation from a surface assumed to be wet while reading the atmosphere as observed, so when soil dries they leave out the chain in which suppressed actual evapotranspiration warms the air and inflates the vapour pressure deficit. The exponential temperature dependence of that deficit then turns steady warming into demand that appears to accelerate. Another reading is available: energy-based formulas constrain demand by surface energy and could understate it where energy availability is not the binding limit, which would move the bias to the other side. What the clustering settles is that the choice between the families carries the sixfold gap.[1]
What the acceleration test actually tested
Acceleration is a second-order claim, and here it is measured with a modified Mann-Kendall test that accounts for temporal autocorrelation, the detection power of the framework validated through synthetic Monte Carlo simulations before it was applied. Across about 99 per cent of quasi-global land, neither family of index shows significant acceleration, and the result remains robust when the series is extended to 1950 and then to 1901. Both families still show drying over 1981-2024; the disagreement is about the magnitude, and about whether the rate itself is rising.[1]
The acceleration conclusion this study takes up rested on comparing two arbitrarily selected periods, 1981-2017 against 2018-2022. Such a comparison is sensitive to interannual variability and to where the changepoint is placed, the same dependence an earlier column traced when what counts as a paper decides how much disruption declined. There the sample definition set the magnitude of the trend; here the evaporative-demand formula sets it. If drought assessments move to the paired reporting the authors argue for, a Penman-type drought trend will be published alongside its energy-based counterpart, and the reader will see the spread rather than one number drawn from it.[1], [2]