Sensitivity of Consecutive‐Dry‐Day Trends to Trace Loss in Matched Precipitation Archives

ABSTRACT

Consecutive dry days (CDD) is widely used to summarize drought‐duration change, but the index depends on the sequence of dry, wet and missing observation states. We test whether provider versions of the same gauge preserve that sequence using 8020 matched station‐version pairs from 24 countries. Pairs represent provider versions of the same physical gauge, so a difference points to archive processing rather than station sampling, siting or instrumentation. Most matched versions agree, but the Spanish AEMET, ECA&D and GHCN‐Daily lineages reveal a localized sensitivity linked to loss of valid trace precipitation. Among 205 AEMET‐involving pairs evaluated over identical years, deliberately degrading valid AEMET traces to missing reduces median from 0.0763 to 0.0089 days yr. ⁻¹ , identifying trace loss followed by spell‐breaking gaps as the dominant mechanism. The effect is largest where trace‐policy exposure coincides with long dry spells: countries with longer dry spells show no detectable discrepancy in their available matched versions, while within Spain the driest stations change trend class in 41.2% of pairs against 17.6% in the wettest, and summer more than winter. A separate within‐archive GHCN‐Daily counterfactual shows that degrading documented trace flags to missing changes CDD trend classification at 5.96% of stations globally. Critically, trace reporting is not stationary: 60.9% of trace‐exposed stations show a significant trend in trace‐day frequency, 72.3% declining, so the encoding choice imprints a directional shift on fitted CDD trends reaching ±0.24 days yr. ⁻¹ in the most exposed networks. Reporting history can therefore be mistaken for drought change. Trace preservation and missing‐day treatment are components of observational uncertainty in precipitation‐duration climatology.