When Does the Rainfall Signal Emerge from Natural Variability? Time of Emergence and the Uncertainty from Reference Data and Bias Correction for Rainfall Extremes in Andean, Amazonian and Coastal Ecuador
Year: 2026
Type: preprint
Abstract: Planning for rainfall extremes in the tropical Andes requires knowing when projected changes can be distinguished from natural variability, and how far that answer depends on the reference data, the bias correction and the scale of the noise. We combined 64 rain gauges (54 from the Quito water utility and 10 from the national service), three CHIRPS products and 34 NEX-GDDP-CMIP6 models. The increase in annual maximum daily rainfall at the Quito gauges (+12.5 % per decade) is sensitive to record length and is not reproduced by gauges on the Coast, in the Austro and in the Amazon. No CHIRPS version reproduces annual maxima, although all capture totals. Gauge-based quantile mapping reduced the error of the annual maximum from 29 to 9 %, but inflated the projected change of annual totals by 7-11 percentage points. Under SSP5-8.5 the ensemble signal of annual rainfall leaves the ±1σ band of natural variability before 2090 in 67 % of the 404 cells of continental Ecuador, but in only 20-28 % of the cells under stricter criteria. Model cells underestimate by about half the interannual variability of totals observed at the gauges; with gauge-scale or 5-km noise, the emerging share falls to 19-48 %. Absence of emergence does not imply absence of change: design should adopt projected change factors now, with explicit uncertainty.
Source: Preprints.org
Language: English
Subfield: Global and Planetary Change
Field: Environmental Science
Domain: Physical Sciences
SDG: Climate action
Open Access status: green
OpenAlex ID: W7225172808