24 Sep 2026
| 24 Sep 2026
Status: this preprint is open for discussion and under review for Natural Hazards and Earth System Sciences (NHESS).
Brief communication: NISAR pixel offsets measure but do not statistically resolve pre-failure motion at the 26 August 2026 Bhote Koshi ice–rock avalanche, Nepal
Abstract. We analyse five NASA–ISRO Synthetic Aperture Radar (NISAR) pixel-offset pairs spanning the 26 August 2026 Bhote Koshi ice–rock avalanche, Nepal. The last pre-event pair estimates 0.96 m median displacement over 24 days, with its mean vector 18° from the topographic fall line. No pre-event interval exceeds the 95th percentile of spatial comparison distributions for displacement or directional coherence. We report rankings rather than detections. The co-event field lacks directional organisation in both processing chains. Available interval averages do not establish precursory acceleration and cannot isolate motion during the final week before failure.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Since the pre-failure motion is not resolvable, the work is of limited interest for understanding what happened during the event from a geomorphological perspective, but it is an interesting test of what NISAR is capable of. It is useful to explore how this newly-launched satellite can be applied to hazard management / disaster response in an area of the world that is challenging for InSAR.
Unfortunately, the way the paper is written makes it very hard to follow. The paper was submitted very soon after the event occurred (less than one month) and while carrying out such analyses immediately after an event like this is useful, it would have been worth proof-reading the manuscript again before submission. In the acknowledgements, the author states that they used AI to review the paper and propose wording. This does not seem to have worked well enough. There are undefined acronyms and symbols in the paper. There are also a lot of very long sentences that are difficult to read. For example: “Along-track (AT) and slant-range (SR) offsets from layer2 were retained where signal-to-noise ratio ≥ 5 and the projected horizontal magnitude did not exceed 10 m, which discards gross correlation blunders rather than clipping them, and projected to horizontal displacement using the product convention, in which losUnitVector is defined from the target to the sensor, so a positive slant-range offset is motion away from the sensor (Eqs. 1 and 2):” This should be broken into several shorter sentences and layer2 and losUnitVector should be defined.
Overall, I am recommending major revisions, because while the analysis itself seems fine, the paper will be very hard to follow for people who are not already acquainted with the methods and processing softwares used. This is especially important because you are presenting products from a brand new satellite and most people reading won’t have worked with data from it yet.
The following are my specific comments:
Line 27 “ Retrospective studies also carry a selection effect that is rarely stated: after a disaster the intervals processed first are those expected to be informative, so a sub-sampled series is structurally more likely to resemble a ramp than a complete one” no idea what it means
Lines 30-32 “Here we report a pixel-offset series covering all available intervals across the 26 August Bhote Koshi event, framed explicitly around what the data resolve and what they do not — because in our own work the interpretation changed when the intervening available pairs were included” The second part of this sentence is unclear to me. What do you mean intervening available pairs? Pairs of satellite images?
Lines 40-41 “26 Jul–19 Aug, twice the repeat interval because the intervening acquisition is unavailable; it is the last pre-event interval and closes seven days before failure” I suggest rephrasing this as “the interval of this final, pre-event image pair is 24 rather than 12 days because the intervening image was unavailable.”. Throughout the paper, there are many examples like this where the odd sentence structure makes the meaning unclear. If you want to state that this interval ends seven days before the failure, it should be a separate sentence.
Line 44 “Along-track and slant range offsets from layer2” what is “layer2”? You have never defined this.
Line 51, if you are assuming that vertical motion is 0, so everything you detect is horizontal motion, does it mean that you are obtaining a maximum possible value for the horizontal motion?
Line 61 “To separate organised motion from correlation noise we use the mean resultant length R of unit displacement vectors in a 5 × 5 window, where R → 1 indicates organised flow.” Is this a commonly used method? Can you cite another study that has used it?
Line 71 “ the crown is computed on the cropped polygon” I cannot see the crown polygon in figure 1b
Line 94-100 this part is very hard to follow, please make it clearer.
Line 94-100 From what I do understand, you are saying that it is difficult to compare between a 24-day image pair and a 12-day image pair. In that case, why not just use 24 image pairs instead of 12?
Line 113 “GAMMA offset-tracking” you have never referenced GAMMA before, so you should state that it is a software package
Line 114 “6 km background box” but the box in Figure 1b is much smaller than 6 km. Should I be looking at the box in Figure 1a? In that case, what does the dashed box in Figure 1b represent?
Equations 1 and 2 you do not define anywhere what most of the symbols mean so the equations are impossible to understand for those who are not already very familiar with the methods.
Figure 2, the upper panels are too small to really see what is happening with the signal.
Line 163-164 “Nothing we measure here establishes that the source area was distinctive”. What do you mean? You mean that what you observe here was not exceptional and so could not have been used to give a warning that the event would happen?
Line 180 “Limitations” should this be a section heading?