Bounded at both ends: the detection window of routine 12-day C-band InSAR at rock–ice avalanche sources
Abstract
A negative result from satellite radar — "no precursory deformation was detected" — carries no information unless the range of signals the measurement could have detected is reported alongside it. We measure that range at two rock–ice avalanche sources, Langtang Lirung, Nepal (2026) and Ronti Peak, India (2021), using a single procedure, and identify four determinants of it that are site-specific and not predictable from the sensor specification. Viewing geometry cannot be selected on usable-pixel fraction, the quantity most easily computed before processing: at Langtang an ascending track covering 100 per cent of the source disc has about four times the variance of a descending track covering 44 per cent, and the failure is invisible on the former (z = −0.52) while the latter registers it at z = +3.43. The limiting seasonal term is variance rather than offset, and so cannot be detrended away: nine same-window February controls at Ronti span 6.95 to 32.11 per cent of the source disc below a coherence threshold, a range wider than any plausible precursor. Control design does not transfer; the published 120-disc scheme admits 46 candidates at Ronti, aspect-limited. And the upper bound is rarely stated at all. Placed against the half-fringe unwrapping limit of a 12-day pair, no published precursory detection we examined lies inside the range a single pair can measure — three sit at or below its floor, three at or above its ceiling, and every one was obtained by multi-epoch inversion accumulated over years rather than from the routine product. Neither event is detectable at its source in routine 12-day C-band products, for opposite reasons: one below the precision floor, one above the unwrapping ceiling. The sensitivity range is bounded at both ends, both bounds are site-specific and differ by an order of magnitude between sites of the same hazard class, and a null reported without them is not interpretable.
Related
- Calibrated detection limits in routine Sentinel-1 products 50.6 hours before the Langtang Lirung rock–ice avalanche, Nepal
- Very-high-resolution optical imagery reveals key clues to the Langtang Lirung ice–rock avalanche
- Multiple portraits of dry snow avalanche slopes near villages in Nunavik, Canada
- A rock-ice avalanche and the 2026 Bhote Koshi-Trishuli flood, Nepal
- Detecting arrest in snow avalanche simulations : interplay between rheology, topography and numerical diffusion
BibTeX
@misc{sah2026bounded,
title = {{Bounded at both ends: the detection window of routine 12-day C-band InSAR at rock–ice avalanche sources}},
author = {Ajay Kumar Sah},
howpublished = {EarthArXiv},
year = {2026},
month = sep,
doi = {10.31223/x58j72},
url = {https://eartharxiv.org/repository/view/14966/},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Bounded at both ends: the detection window of routine 12-day C-band InSAR at rock–ice avalanche sources
AU - Ajay Kumar Sah
PY - 2026
DA - 2026/09/14
PB - EarthArXiv
DO - 10.31223/x58j72
UR - https://eartharxiv.org/repository/view/14966/
AB - A negative result from satellite radar — "no precursory deformation was detected" — carries no information unless the range of signals the measurement could have detected is reported alongside it. We measure that range at two rock–ice avalanche sources, Langtang Lirung, Nepal (2026) and Ronti Peak, India (2021), using a single procedure, and identify four determinants of it that are site-specific and not predictable from the sensor specification. Viewing geometry cannot be selected on usable-pixel fraction, the quantity most easily computed before processing: at Langtang an ascending track covering 100 per cent of the source disc has about four times the variance of a descending track covering 44 per cent, and the failure is invisible on the former (z = −0.52) while the latter registers it at z = +3.43. The limiting seasonal term is variance rather than offset, and so cannot be detrended away: nine same-window February controls at Ronti span 6.95 to 32.11 per cent of the source disc below a coherence threshold, a range wider than any plausible precursor. Control design does not transfer; the published 120-disc scheme admits 46 candidates at Ronti, aspect-limited. And the upper bound is rarely stated at all. Placed against the half-fringe unwrapping limit of a 12-day pair, no published precursory detection we examined lies inside the range a single pair can measure — three sit at or below its floor, three at or above its ceiling, and every one was obtained by multi-epoch inversion accumulated over years rather than from the routine product. Neither event is detectable at its source in routine 12-day C-band products, for opposite reasons: one below the precision floor, one above the unwrapping ceiling. The sensitivity range is bounded at both ends, both bounds are site-specific and differ by an order of magnitude between sites of the same hazard class, and a null reported without them is not interpretable.
ER -