Was the Pre-Event SWIR Anomaly at Tsangbu Ri a Genuine Glacier-Collapse Precursor? A Calibrated Multi-Glacier Test of the 2026 Case

Prakash Pradhan 

On 26 August 2026 a compound bedrock-and-glacier failure on the northern flank of the Langtang Lirung–Tsangbu Ri (Kimshung) massif, Nepal–China border, generated a rock–ice avalanche and a catastrophic downstream flood along the Bhotekoshi–Trishuli river system. A pre-event Sentinel-2 shortwave-infrared (SWIR2–NIR–Red) composite (12 August 2026) shows a sharply bounded, strongly saturated blue-to-cyan patch on the debris-mantled glacier tongue below the slope that failed fourteen days later — a real, reproducible visual feature that a naive reading might treat as a wetness precursor. That reading is tested here against two independent, calibrated baselines built from a 27-year (2000–2026), multi-sensor Landsat/Sentinel-2 reconstruction in Google Earth Engine, using a Shortwave-Infrared Wetness Index, SWI = (NIR − SWIR2)/(NIR + SWIR2).
First, a within-glacier climatological baseline, computed directly against Tsangbu Ri’s own official RGI 7.0 outline (RGI2000-v7.0-G-15-05732): the 2026 ablation-season SWI (0.758) against the catchment’s own 2000–2025 record (0.773 ± 0.084) gives z = −0.19 — a small negative departure, not the positive anomaly the visual composite suggests, and modest even in within-glacier terms alone. Second, and more decisively, an across-glacier false-positive calibration that the remote-sensing glacier-hazard literature has generally lacked: the same pipeline applied to 17 matched, non-collapsed control glaciers in the same RGI region (South Asia East), selected programmatically by elevation, area, and aspect similarity to Tsangbu Ri’s own confirmed polygon, with sensor cross-calibration and a fixed observation-density correction applied identically to target and controls. Against this empirical null distribution, Tsangbu Ri’s 2026 departure sits at the 59th percentile (Figure 6; 95% CI 33–82%, Clopper-Pearson) — ten of seventeen independently selected control glaciers show an equal-or-more-negative departure, and eleven of seventeen (65%) independently show a negative 2026 departure of their own, consistent with a regionally shared dry pattern rather than a site-specific signal. Twelve of seventeen controls (71%) show a larger absolute departure, in either direction, than Tsangbu Ri itself.
The SWIR mechanism itself — sensitivity to standing/ponded liquid water and to melt-metamorphosed, coarse-grained wet snow or firn via strong SWIR2 absorption — is interpreted here as physically well founded and literature-consistent. For the Tsangbu Ri case specifically, however, neither the within-glacier nor the calibrated across-glacier evidence supports a SWIR-based wetness precursor at the failure site. This is reported as a methodologically load-bearing finding: a visually compelling single-date spectral anomaly, and even an uncalibrated within-glacier z-score, are not sufficient grounds for a precursor claim without a matched-control false-positive rate — and the calibrated pipeline, now run directly against Tsangbu Ri’s own confirmed RGI polygon and validated for reproducibility, is set out here as one that would extend this test to a larger control sample and to other High Mountain Asia compound failures.