Pre-registration: When snow looks like water: wet-snow false floods in winter Sentinel-1 flood mapping of the Tensift basin (High Atlas, Morocco)
Abstract
This record pre-registers a study of wet-snow false positives in winter Sentinel-1 change-detection flood mapping in the Tensift basin (High Atlas, Morocco). The pilot winter 2019–20 was used to build and test the pipeline and is excluded from all confirmatory tests. The confirmatory analysis covers nine winters (November–April, 2015–16 to 2024–25, excluding 2019–20), with area-weighted estimates, a two-stage week-block bootstrap and Bonferroni-adjusted decision rules. Hypotheses: H1: detected area is over-represented at ≥ 1,500 m relative to the observed-area control; H2: most detected area at ≥ 1,500 m is snow-associated; H3: a HAND ≤ 15 m mask removes at least three-quarters of snow-associated area; H4: the mask also removes a non-trivial share of low-elevation non-snow area; H5: snow-associated area is concentrated in a minority of acquisitions; H6: ERA5-only screening has lower recall than optical screening. The record includes the frozen extraction and analysis scripts with SHA-256 hashes. Contact: Ayoub Echchadiry, ayoub_echchadiry@um5.ac.ma
// Source
Authors: Ayoub Echchadiry, Abdelaziz Lahmili
Institutions: Mohammed V University, Ecole Mohammadia d'Ingénieurs