初始化协作仓库：赛题材料、初步试解与协作说明

- 大文件（附件原始包、data/）由 files.ddwsgood.top 承载，不入库
- 新增 .gitignore/.gitattributes/COLLAB.md

2026_初步评估/B_DeepSeek试解/外部数据/zenodo_22717809_gyirong_precursors/README.md

@@ -0,0 +1,65 @@
+# Langtang Lirung–Gyirong GRL open-research package (v1.2.1)
+
+Derived data, tables, Supporting Information, and figure scripts for:
+
+**Ming, J., and Du, Z. (2026). Optical and radar signals before the 2026 Gyirong cascade. *Geophysical Research Letters* (manuscript 2026GL126799).**
+
+Release **v1.2.1** adds the ASF HyP3 job manifest (`configs/hyp3_jobs.json`,
+33 ASF job IDs; `configs/hyp3_job_status.csv`; `configs/hyp3_download_summary.json`),
+so the ~67 GB interferogram archives referenced in the Availability Statement are
+recoverable from ASF by job ID in addition to the parameterised reconstruction.
+Release **v1.2.0** merged the frozen v1.1.0 package (commit `f57500a`; zip SHA-256
+`17677b8884d77a13d53699e780da03c6a3093dc4b013616c469026f62b856960`) with the
+v1.1.1 InSAR/control-albedo quality-assurance increment and the AGU-compliant
+Supporting Information (title page, table of contents, Tables S1–S4). No derived
+data or scientific content changes relative to v1.2.0.
+
+## Contents
+
+- `si/` — Supporting Information (`Supporting_Information.md`: Text S1–S5, Tables S1–S4) plus machine-readable Tables S1–S4 (CSV)
+- `data/` — area-of-interest (AOI) optical and SAR metrics, full Sentinel-1 RTC pair table, control-glacier albedo comparison, RGI 7.0 near-field glacier inventory and TSI ranks, detachment AOI polygons, topographic summaries, derived TPDC 30 m grids, provisional Shirzaei InSAR annotation metadata and AOI spatial checks
+  - `data/insar_qa/`, `data/optical_qa/` — dry-season HyP3–MintPy SBAS geometry/troposphere/reference-frame QA and glacier-level control-albedo uniqueness QA (the v1.1.1 increment)
+  - `data/precursors/topo_rasters/` — derived 30 m slope, ice thickness, and driving-stress GeoTIFFs
+- `code/` — R and Python figure scripts
+- `scripts/` — SBAS geometry QA, tropospheric comparison, and control-albedo uniqueness scripts
+- `configs/` — ASF HyP3 job manifest (`hyp3_jobs.json`: 33 ASF job IDs; `hyp3_job_status.csv`; `hyp3_download_summary.json`), HyP3 submission parameters (`insar_hyp3.yml`), and MintPy configuration files
+- `figures/` — publication figures (Figure1–Figure3, PDF and PNG) and development panels
+- `notes/` — InSAR claim-level decision notes and the provider GeoTIFF request draft
+- `MANIFEST.json`, `MANIFEST.sha256` — SHA-256 checksums for every file in this package
+
+## Provenance of upstream data
+
+- Sentinel-2 Level-2A and Sentinel-1 RTC: Microsoft Planetary Computer / ESA Copernicus / ASF DAAC (open access).
+- Glacier outlines: RGI 7.0, NSIDC, https://doi.org/10.5067/F6JMOVY5NAVZ.
+- Meteorological fields: ERA5-Land, Copernicus CDS, https://doi.org/10.24381/cds.e2161bac.
+- 30 m DEM/slope/ice-thickness source grids: H. Zhao and T. Zhang, *Glacier topographic and dynamic drivers (Glacier-collapse and debris-flow disasters at Langtang Lirung, Nepal)*, National Tibetan Plateau Data Center (free registration), https://data.tpdc.ac.cn; velocity component ITS_LIVE v2.1.
+- China administrative boundaries: RESDC, Chinese Academy of Sciences, https://doi.org/10.12078/2023010103.
+- Provisional Shirzaei (2026) InSAR annotation: public post metadata only; no provider GeoTIFF is redistributed.
+- Full HyP3 interferogram archives (~67 GB) and MintPy product cubes are not included; they are reproducible from ASF either by the 33 job IDs in `configs/hyp3_jobs.json` or by re-submitting `configs/insar_hyp3.yml` (seed 7341), together with the MintPy configuration files.
+
+## Reproducibility
+
+```bash
+# publication figures
+Rscript --vanilla code/fig_precursor_mechanism_panel.R
+
+# SBAS quality assurance (dry-season stacks)
+uv run --with mintpy --with 'gdal==3.13.3' --with h5py --with numpy \
+  --with pandas --with pyyaml --with pyproj --with scipy \
+  python scripts/insar_sbas_geometry_qa.py
+
+uv run --with mintpy --with 'gdal==3.13.3' --with pyaps3 --with cdsapi \
+  --with h5py --with numpy --with pandas --with pyyaml --with pyproj --with scipy \
+  python scripts/insar_sbas_tropo_test.py
+
+# glacier-level control albedo uniqueness
+uv run --with numpy --with pandas python scripts/control_albedo_uniqueness_qa.py
+```
+
+Random seed: **7341**. CRS for topographic products: **EPSG:32645** (30 m).
+
+## Licence
+
+Analysis code and scripts: MIT (see `LICENSE`). Derived data tables and rasters:
+CC-BY-4.0. Upstream satellite and reanalysis products remain under their
+provider terms (ESA/Copernicus, NSIDC, TPDC, RESDC).

2026_初步评估/B_DeepSeek试解/外部数据/zenodo_22717809_gyirong_precursors/detachment_aoi.geojson

@@ -0,0 +1,27 @@
+{
+  "type": "FeatureCollection",
+  "features": [
+    {
+      "type": "Feature",
+      "properties": {
+        "name": "Langtang Lirung North detachment source area",
+        "name_zh": "朗塘里峰北坡崩解源区",
+        "source": "Petley 2026-08-27; Planet/Shugar ~85.52°E 28.276°N; RGI G-15-05732 vicinity",
+        "event_date": "2026-08-26",
+        "note": "Corrected from earlier GeoLibre viewport (85.37°E); ~14 km east"
+      },
+      "geometry": {
+        "type": "Polygon",
+        "coordinates": [
+          [
+            [85.505, 28.262],
+            [85.540, 28.262],
+            [85.540, 28.292],
+            [85.505, 28.292],
+            [85.505, 28.262]
+          ]
+        ]
+      }
+    }
+  ]
+}

2026_初步评估/B_DeepSeek试解/外部数据/zenodo_22717809_gyirong_precursors/si/Supporting_Information.md

@@ -0,0 +1,176 @@
+Geophysical Research Letters
+
+# Supporting Information for
+
+# Optical and radar signals before the 2026 Gyirong cascade
+
+**Jing Ming¹ and Zhencai Du²**
+
+¹ Beacon Science & Consulting, Tranmere, SA 5073, Australia
+
+² Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China
+
+Corresponding authors: Jing Ming (mingjing@beaconscience.org); Zhencai Du (duzhc@mail.iap.ac.cn)
+
+**Contents of this file**
+
+- Text S1. Optical and SAR processing
+- Text S2. GCSI and TSI definitions (descriptive only)
+- Text S3. Topographic driving stress
+- Text S4. Provisional InSAR LOS rates, open SBAS processing, and downslope projection
+- Text S5. Seismic and precipitation metadata adopted from Guo et al. (2026)
+- Table S1. Sentinel-2 AOI optical metrics for the eight evaluated dates
+- Table S2. Full Sentinel-1 RTC amplitude-offset pair series
+- Table S3. Fifteen highest-ranked near-field glaciers by Topographic Susceptibility Index
+- Table S4. 15 June versus 12 August 2026 albedo for the detachment AOI, source glacier, and five non-failing controls
+
+Machine-readable copies of Tables S1–S4 and all other derived data, scripts, and configurations are archived in the Zenodo dataset accompanying this article (release v1.2.0; the dataset DOI is given in the Availability Statement of the main article).
+
+## Introduction
+
+This Supporting Information documents the processing rules, descriptive indices, and quality-assurance tests behind the main-text figures and tables. It contains (i) Sentinel-2 and Sentinel-1 processing details and the usability rules applied to each series (Text S1); (ii) exact definitions of the descriptive Glacier Collapse Susceptibility Index (GCSI) and Topographic Susceptibility Index (TSI) (Text S2); (iii) the shallow-ice driving-stress calculation and its limits (Text S3); (iv) the provisional third-party InSAR line-of-sight annotation, the open dry-season HyP3–MintPy SBAS reproducibility path, all tropospheric and reference-frame tests, and the line-of-sight to downslope projection (Text S4); and (v) the seismic and precipitation metadata adopted, without re-digitisation, from Guo et al. (2026) (Text S5). Tables S1–S4 embed the full tabular data cited in the main text. Known anomalies (cloud-contaminated optical dates, quantized amplitude offsets, same-sign ascending/descending SBAS velocities) are stated at the point of use. All spatial processing uses CRS EPSG:32645 and random seed 7341 unless otherwise stated.
+
+# Text S1. Optical and SAR processing
+
+## Sentinel-2
+
+- Level-2A surface reflectance accessed through the Microsoft Planetary Computer STAC API (ESA Copernicus Sentinel-2, tile T45RUM).
+- Broadband albedo: mean of the visible/NIR bands as implemented in the project figure and metric scripts (archived in `code/` of the Zenodo release).
+- NDSI, NDWI, and SWIR (B11) texture standard deviation computed per AOI pixel as in `Table_S1_sentinel2_aoi_timeseries.csv` / `sentinel2_detachment_metrics.csv`.
+- **Usable albedo-series rule:** scenes with at least 1000 valid AOI pixels. The 14 August scene is excluded (*n* = 110). The 23 June scene is excluded because the AOI-mean albedo exceeds 1 (1.076), which is physically impossible and indicates cloud or bright-snow contamination.
+- Full date list, scene identifiers, cloud cover, and usability flags: Table S1 (eight evaluated dates; six retained for the albedo series).
+
+## Sentinel-1
+
+- Radiometrically terrain-corrected (RTC) amplitude products (Microsoft Planetary Computer, sourced from ASF DAAC), matched on the same relative orbit; 12-day pairs preferred.
+- Reported fields: mean offset velocity (m d⁻¹), `velocity_std`, number of matched pixels, and integer pixel shifts `dx_m`, `dy_m` (Table S2).
+- **Interpretation limit:** every deposited pair has `velocity_std = 1.0`, indicating coarse, quantized amplitude matching rather than a high-precision velocity time series. Rates are intermittent offset anomalies near a practical detection floor. The full ascending and descending series is reported (Table S2; Fig. 2b), not only the largest late-season descending pair.
+
+# Text S2. GCSI and TSI definitions (descriptive only)
+
+## Glacier Collapse Susceptibility Index (GCSI)
+
+The GCSI is reported only in the archive and this Supporting Information; it is not used in the Abstract or Key Points and is not a calibrated probability.
+
+1. Standardise selected components (Δalbedo and/or albedo *z*, NDWI *z*, SWIR texture *z*, optional land-surface-temperature *z*, optional velocity *z*) across the June–August 2026 sample.
+2. Composite = equal-weight mean of the available component *z*-scores on each date (implementation: `gcsi_timeseries.csv`, `gcsi_v2_timeseries.csv`).
+3. The index is not cross-validated against any ice-avalanche inventory and is sensitive to which dates meet the usability rule in Text S1.
+
+## Topographic Susceptibility Index (TSI)
+
+For each RGI 7.0 glacier within 15 km of the scar centre:
+
+- Normalise mean slope and mean elevation to [0, 1] within the near-field set.
+- TSI = 0.5 × slope_norm + 0.5 × elevation_norm (equal weights; ranking only).
+- Hanging-ice flag: steep, high ice facing the valley (Boolean produced by the project inventory script; slope ≥ 30° and *z*max ≥ 6500 m for the summary in Fig. 3d).
+- Eight flagged glaciers among 138 near-field glaciers quantify the false-positive load of the geometric screen (Table S3 lists the 15 highest TSI glaciers).
+
+# Text S3. Topographic driving stress
+
+Shallow-ice ranking stress τ ≈ ρ g H sin(α), with ρ_ice = 917 kg m⁻³ and an optional mixed-density case ρ_mix = 1500 kg m⁻³. Inputs are the 30 m TPDC DEM, slope, and ice-thickness grids described in Methods §2.5. **Caveat:** steep hanging ice violates the shallow-ice assumptions; the stress maps are spatial ranking fields, not failure criteria (seed 7341). DEM ice-mask pixels inside the AOI: *n* = 9426 at 30 m, distinct from the Sentinel-2 valid-pixel counts in Table S1.
+
+# Text S4. Provisional InSAR LOS rates, open SBAS processing, and downslope projection
+
+**Product (provisional rates used in Table 1).** Shirzaei (2026) annotated InSAR line-of-sight (LOS) rates for 8 January–19 August 2026 (ending 7 days before the 26 August cascade). Reported LOS: **−4 to −6 mm month⁻¹**. The peak annotation falls inside the corrected detachment AOI after an approximately 14 km westward viewport shift (`insar_aoi_validation.json`; `shirzaei_langtang_los_velocity.json`).
+
+**What is archived for the Shirzaei product.** Annotation JSON, JPEG-overlay metadata, and AOI spatial checks. No provider GeoTIFF is deposited; full track, incidence angle, multilook, and DEM metadata from the annotator remain pending. A citable provider product, if released, will be added to the archive.
+
+**Downslope projection (approximate; Shirzaei LOS).** Under a single descending-orbit geometry (heading 193°, incidence 36°), LOS is projected onto a local downslope unit vector from the AOI DEM ice mask (`insar_los_to_downslope.csv`). For median LOS = −5 mm month⁻¹ the projected rate is about 8 mm month⁻¹ (~0.10 m yr⁻¹). Ascending geometry yields a larger |downslope| because the LOS–slope dot product is smaller (less sensitive); the descending projection is quoted in the main text as the more conservative geometric case.
+
+**Open dry-season HyP3–MintPy SBAS (co-location only).** A dry-season Sentinel-1 network (January–April 2026; maximum temporal baseline 24 d) was submitted to ASF HyP3 `INSAR_GAMMA` (33 successful pairs; job manifest `configs/insar_hyp3.yml`) and processed with MintPy SBAS separately for ascending path 85 (11 interferograms after dropping one INT80 grid mismatch; `configs/mintpy_path85_asc_sens.cfg`) and descending path 19 (12 interferograms; `configs/mintpy_path19_des_sens.cfg`), with orbits kept separate. Baseline AOI median MintPy LOS (positive = toward satellite) is about +4.3 mm month⁻¹ (ASC; velocityStd ≈ 3.7) and +7.5 mm month⁻¹ (DESC; ≈ 3.3). Converting HyP3 incidence maps from radians and projecting LOS to local downslope yields same-sign projected rates within a factor of about 2, but ascending and descending MintPy LOS share the same sign, which is inconsistent with pure downslope creep, and ASC |LOS|/velocityStd ≲ 2. QA tables: `data/insar_qa/aoi_sbas_geometry_qa.json`, `aoi_sbas_downslope_qa.csv`, `aoi_sbas_velocity.csv`.
+
+**Tropospheric and reference-frame tests (did not raise the claim).** ERA5 delays via PyAPS and an empirical height–phase correlation were applied to both stacks (`scripts/insar_sbas_tropo_test.py`; `data/insar_qa/aoi_sbas_tropo_compare.*`). ERA5 reduces |LOS| but preserves the ASC/DESC sign and lowers SNR (ASC ~0.66; DESC ~0.95). Height correlation removes nearly all AOI signal. A pre-registered sensitivity with a shared bedrock reference pixel (YX = 241, 220; seed 7341), linear deramp, and minTempCoh = 0.5 (`aoi_sbas_refderamp_preregister.json`, `aoi_sbas_refderamp_compare.*`) raises both AOI SNRs above 2 but still yields same-sign ASC/DESC LOS. The claim level remains **open-SBAS co-location and reproducibility path only**; these rates do not replace the Shirzaei −4 to −6 mm month⁻¹ annotation.
+
+**Relation to Sentinel-1 RTC offsets.** InSAR LOS (annotation or SBAS mean) is a creep-scale, multi-month or seasonal mean (mm month⁻¹). Sentinel-1 RTC amplitude offsets are intermittent pair rates of order 0.1 m d⁻¹ with `velocity_std = 1.0` (Text S1). They are complementary observables, not interchangeable velocity series.
+
+**Use in the Letter.** The Shirzaei rates enter Results §3.3 and Table 1 only as provisional supporting co-location. The open SBAS path is cited at the same claim level and is not calibrated ice velocity, not a failure probability, and not required to reproduce the optical, Sentinel-1 RTC, control-glacier, or hanging-ice results.
+
+# Text S5. Seismic and precipitation metadata adopted from Guo et al. (2026)
+
+The following are adopted as event metadata from Guo et al. (2026) and are not recomputed here: Kakani (KKN) broadband timing; catalogue M ~5–5.7 read as mass-movement radiation rather than a tectonic trigger; about 8.3 mm of source-area precipitation in the final 72 h; and the approximately 0.72 km² detached ice–rock footprint.
+
+# Table S1.
+
+Sentinel-2 Level-2A area-of-interest (AOI) optical metrics for the eight evaluated dates (3 June–24 August 2026), with the usability flag applied to the albedo series. The 23 June scene (AOI-mean albedo 1.076) and the 14 August scene (*n* = 110 valid pixels) are excluded. Full-resolution machine-readable file: `si/Table_S1_sentinel2_aoi_timeseries.csv` in the Zenodo release.
+
+| Date | Sentinel-2 scene ID | Cloud (%) | n pixels | Albedo mean | Albedo SD | NDSI | NDWI | SWIR texture | Usable |
+|---|---|---|---|---|---|---|---|---|---|
+| 2026-06-03 | S2C_MSIL2A_20260603T045701_R119_T45RUM_20260603T100112 | 36.1 | 24844 | 0.842 | 0.239 | 0.606 | 0.036 | 0.021 | yes |
+| 2026-06-05 | S2A_MSIL2A_20260605T050231_R119_T45RUM_20260605T100518 | 41.9 | 23923 | 0.869 | 0.215 | 0.580 | 0.041 | 0.024 | yes |
+| 2026-06-15 | S2B_MSIL2A_20260615T044659_R076_T45RUM_20260615T083228 | 18.4 | 23668 | 0.887 | 0.231 | 0.632 | 0.027 | 0.018 | yes |
+| 2026-06-23 | S2C_MSIL2A_20260623T045701_R119_T45RUM_20260623T100125 | 47.5 | 28735 | 1.076 | 0.251 | 0.395 | 0.064 | 0.041 | excluded: albedo > 1 |
+| 2026-07-03 | S2C_MSIL2A_20260703T045701_R119_T45RUM_20260703T095511 | 44.4 | 18655 | 0.735 | 0.200 | 0.442 | 0.038 | 0.023 | yes |
+| 2026-08-12 | S2C_MSIL2A_20260812T045701_R119_T45RUM_20260812T100317 | 18.7 | 16719 | 0.652 | 0.295 | 0.559 | 0.036 | 0.015 | yes |
+| 2026-08-14 | S2B_MSIL2A_20260814T044659_R076_T45RUM_20260814T083413 | 49.0 | 110 | 0.388 | 0.054 | 0.274 | 0.057 | 0.018 | excluded: n < 1000 |
+| 2026-08-24 | S2B_MSIL2A_20260824T044659_R076_T45RUM_20260824T083447 | 38.6 | 15538 | 0.722 | 0.359 | 0.528 | 0.047 | 0.029 | yes |
+
+# Table S2.
+
+Full Sentinel-1 radiometrically terrain-corrected (RTC) amplitude-offset pair series, ascending path 85 and descending path 19 (May–August 2026). All pairs have `velocity_std = 1.0`. Full-resolution machine-readable file: `si/Table_S2_sentinel1_velocity_pairs.csv`.
+
+| Start | End | Orbit | Rel. orbit | Velocity (m d⁻¹) | velocity std | Pair days | n pixels | dx (m) | dy (m) |
+|---|---|---|---|---|---|---|---|---|---|
+| 2026-05-05 | 2026-05-17 | ascending | 85 | 0.250 | 1.0 | 12 | 58196 | 0.0 | 3.0 |
+| 2026-05-17 | 2026-05-29 | ascending | 85 | 0.083 | 1.0 | 12 | 62374 | −1.0 | 0.0 |
+| 2026-05-29 | 2026-06-10 | ascending | 85 | 0.083 | 1.0 | 12 | 60399 | 1.0 | 0.0 |
+| 2026-06-10 | 2026-06-22 | ascending | 85 | 0.083 | 1.0 | 12 | 58079 | 1.0 | 0.0 |
+| 2026-06-22 | 2026-06-29 | ascending | 85 | 0.143 | 1.0 | 7 | 59779 | 0.0 | −1.0 |
+| 2026-06-29 | 2026-07-11 | ascending | 85 | 0.083 | 1.0 | 12 | 65521 | 0.0 | 1.0 |
+| 2026-07-11 | 2026-07-23 | ascending | 85 | 0.118 | 1.0 | 12 | 59097 | −1.0 | 1.0 |
+| 2026-07-23 | 2026-08-04 | ascending | 85 | 0.083 | 1.0 | 12 | 63447 | 0.0 | −1.0 |
+| 2026-08-04 | 2026-08-16 | ascending | 85 | 0.000 | 1.0 | 12 | 62071 | 0.0 | 0.0 |
+| 2026-08-16 | 2026-08-28 | ascending | 85 | 0.000 | 1.0 | 12 | 59135 | 0.0 | 0.0 |
+| 2026-05-01 | 2026-05-13 | descending | 19 | 0.083 | 1.0 | 12 | 60069 | 0.0 | −1.0 |
+| 2026-05-13 | 2026-06-18 | descending | 19 | 0.028 | 1.0 | 36 | 54012 | 0.0 | 1.0 |
+| 2026-06-18 | 2026-06-25 | descending | 19 | 0.143 | 1.0 | 7 | 59996 | 0.0 | −1.0 |
+| 2026-06-25 | 2026-07-07 | descending | 19 | 0.118 | 1.0 | 12 | 59241 | −1.0 | −1.0 |
+| 2026-07-07 | 2026-07-19 | descending | 19 | 0.083 | 1.0 | 12 | 65333 | 0.0 | 1.0 |
+| 2026-07-19 | 2026-07-31 | descending | 19 | 0.118 | 1.0 | 12 | 61733 | 1.0 | 1.0 |
+| 2026-07-31 | 2026-08-12 | descending | 19 | 0.186 | 1.0 | 12 | 62017 | −1.0 | 2.0 |
+| 2026-08-12 | 2026-08-24 | descending | 19 | 0.083 | 1.0 | 12 | 62341 | −1.0 | 0.0 |
+
+# Table S3.
+
+Fifteen highest-ranked RGI 7.0 glaciers within 15 km of the scar centre by Topographic Susceptibility Index (TSI; Text S2), with mean slope, maximum elevation, and the hanging-ice flag. The failed source glacier G-15-05732 ranks fifth. Full RGI identifiers and fields: `si/Table_S3_surrounding_glacier_tsi_top15.csv`; the complete near-field inventory is `data/precursors/surrounding_glacier_inventory_tsi.csv` in the Zenodo release.
+
+| RGI label | Distance (km) | Mean slope (°) | z max (m) | TSI | Hanging ice |
+|---|---|---|---|---|---|
+| G-15-05733 | 2.25 | 46.81 | 6522.1 | 0.498 | yes |
+| G-15-05746 | 3.00 | 43.58 | 7199.4 | 0.474 | yes |
+| G-15-05747 | 3.21 | 52.16 | 6436.9 | 0.472 | no |
+| G-15-05734 | 3.10 | 49.36 | 6402.9 | 0.458 | no |
+| G-15-05732 | 1.40 | 32.45 | 7199.9 | 0.450 | yes |
+| G-15-05745 | 3.46 | 43.94 | 6998.2 | 0.438 | yes |
+| G-15-05736 | 3.63 | 54.19 | 6053.9 | 0.433 | no |
+| G-15-05751 | 4.89 | 45.11 | 6696.7 | 0.358 | yes |
+| G-15-05737 | 4.65 | 45.83 | 6140.3 | 0.342 | no |
+| G-15-05760 | 5.35 | 52.55 | 5841.5 | 0.330 | no |
+| G-15-05762 | 4.22 | 34.21 | 6920.2 | 0.319 | yes |
+| G-15-05744 | 4.36 | 41.72 | 5934.5 | 0.318 | no |
+| G-15-05739 | 4.24 | 32.49 | 7216.4 | 0.317 | yes |
+| G-15-05731 | 2.91 | 34.75 | 5699.0 | 0.313 | no |
+| G-15-05735 | 3.64 | 34.71 | 6211.6 | 0.312 | no |
+
+# Table S4.
+
+Sentinel-2 broadband albedo on 15 June and 12 August 2026 for the post-event consensus detachment AOI, the failed source glacier polygon (G-15-05732), and five non-failing control glaciers within about 3–12 km selected for steep ice and usable pixel counts on both dates. Glacier-level controls use RGI polygon means; the AOI row uses the scar AOI mask. Full fields (mean elevation, pixel counts, notes): `si/Table_S4_control_glacier_albedo.csv`.
+
+| Glacier / unit | Role | Distance (km) | Area (km²) | Slope (°) | Hanging | Albedo 15 Jun | n (15 Jun) | Albedo 12 Aug | n (12 Aug) | Δ albedo |
+|---|---|---|---|---|---|---|---|---|---|---|
+| detachment AOI | source AOI | 0.00 | 8.483 | 36.42 | yes | 0.887 | 23668 | 0.652 | 16719 | −0.235 |
+| G-15-05732 | source, failed | 1.40 | 9.842 | 32.45 | yes | 0.873 | 54186 | 0.690 | 37178 | −0.183 |
+| G-15-05745 | control, non-failing | 3.46 | 0.744 | 43.94 | yes | 0.841 | 6466 | 0.826 | 6242 | −0.015 |
+| G-15-05757 | control, non-failing | 6.57 | 1.865 | 30.43 | yes | 0.837 | 6790 | 0.654 | 6298 | −0.183 |
+| G-15-05907 | control, non-failing | 11.60 | 0.548 | 33.95 | no | 0.777 | 4859 | 0.822 | 4307 | +0.045 |
+| G-15-05740 | control, non-failing | 3.55 | 0.900 | 29.16 | no | 0.706 | 4076 | 0.475 | 1106 | −0.231 |
+| G-15-05762 | control, non-failing | 4.22 | 3.201 | 34.21 | yes | 0.824 | 13202 | 0.668 | 10683 | −0.156 |
+
+## References from the Supporting Information
+
+Alaska Satellite Facility (ASF) DAAC. *HyP3 (Hummynet Processing Pipeline v3), Sentinel-1 On-Demand InSAR (INSAR_GAMMA) and RTC products.* NASA ASF DAAC, Fairbanks, Alaska. https://hyp3.asf.alaska.edu/ (accessed June–August 2026).
+
+Muñoz Sabater, J. (2019), *ERA5-Land hourly data from 1950 to present*, Copernicus Climate Change Service (C3S) Climate Data Store, https://doi.org/10.24381/cds.e2161bac.
+
+Yunjun, Z., Fattahi, H., and Amelung, F. (2019), Small baseline InSAR time series analysis: Unwrapping error correction and noise reduction, *Computers & Geosciences*, 133, 104331, https://doi.org/10.1016/j.cageo.2019.104331.
+
+Zhao, H., and Zhang, T. (2026), *Glacier topographic and dynamic drivers (Glacier-collapse and debris-flow disasters at Langtang Lirung, Nepal): pre-event 30 m DEM, slope, ice thickness and surface velocity grids*, National Tibetan Plateau Data Center (TPDC), Beijing (free registration required), https://data.tpdc.ac.cn. Surface velocity component: ITS_LIVE v2.1 (NASA MEaSUREs).

2026_初步评估/B_DeepSeek试解/外部数据/zenodo_22730419_bhotekoshi_flowextent/LICENSE.txt

@@ -0,0 +1,27 @@
+DIFFERENT LICENCE FROM THE REST OF THIS RECORD
+==============================================
+
+Files in this folder:
+
+    flow_extent.shp / .shx / .dbf / .prj / .cpg
+
+Content: the mapped debris-flow runout extent of the 26 August 2026 Bhote Koshi ice-rock
+avalanche, Rasuwa District, Nepal. Digitised from post-event WorldView-3 imagery of 27 and
+28 August 2026.
+
+Licence: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
+         https://creativecommons.org/licenses/by-nc/4.0/
+
+Copyright: (c) Vantor 2026. Source imagery released through the Vantor Open Data Program
+           under CC BY-NC 4.0.
+
+The rest of this Zenodo record is licensed CC BY 4.0. That licence does NOT apply to these
+files. Because the source imagery carries a non-commercial restriction, the outline digitised
+from it carries the same restriction.
+
+Required attribution if you reuse these files:
+
+    Debris-flow extent digitised from WorldView-3 imagery (c) Vantor 2026, released through
+    the Vantor Open Data Program under CC BY-NC 4.0.
+
+Geometry: EPSG:4326, 2 features, mapped area 13.21 km2.

2026_初步评估/B_DeepSeek试解/外部数据/zenodo_22730419_bhotekoshi_flowextent/README_upstream.md

@@ -0,0 +1,127 @@
+# Bhote Koshi 2026 ice–rock avalanche: NISAR L2 GOFF site series and footprint-wide offset-tracking screen
+
+Data and code supporting the article:
+
+> Meena, S. R. *NISAR offsets measure pre-failure motion but do not single out the Bhote Koshi
+> avalanche source.*
+
+Author: Sansar Raj Meena, Geophysics Department, SpatioAI Lab, National Institute of Oceanography
+and Applied Geophysics – OGS, Borgo Grotta Gigante 42/C, 34010 Sgonico, Trieste, Italy
+(<smeena@ogs.it>, ORCID 0000-0001-6175-6491).
+
+DOI of this record: <https://doi.org/10.5281/zenodo.22730419>
+
+A preprint of the article is available at <https://www.researchsquare.com/article/rs-11022365/v1>.
+
+---
+
+## Licensing — read this first
+
+This record contains **two components under different licences**.
+
+| Component | Licence |
+|---|---|
+| Everything in this record **except** `extent_CC-BY-NC-4.0/` | **CC BY 4.0** |
+| `extent_CC-BY-NC-4.0/` (mapped debris-flow runout outline) | **CC BY-NC 4.0, © Vantor 2026** |
+
+The debris-flow extent was digitised from WorldView-3 imagery released through the Vantor Open Data
+Program under CC BY-NC 4.0. It is kept in its own folder, with its own `LICENSE.txt`, so the
+non-commercial restriction travels with the file. It is **not** covered by the CC BY 4.0 licence
+applied to the record as a whole.
+
+---
+
+## What this is
+
+The 26 August 2026 ice–rock avalanche detached from the Bhote Koshi headwaters in Rasuwa District,
+Nepal, and generated a debris flow down the Bhote Koshi–Trishuli valley. The failure fell inside the
+first season of routine 12-day NISAR acquisitions, so the source was imaged five times before the
+event and once in the interval spanning it.
+
+The article asks two questions and this record contains everything needed to reproduce both answers:
+
+1. **At the source** — is pre-failure motion resolvable against a calibrated null rather than against
+   a bare background, and does the series show acceleration?
+2. **Across the footprint** — screening the full 345 × 339 km (116,707 km²) NISAR footprint blind, is
+   the source distinctive among coherently moving surfaces?
+
+The headline result is negative: motion at the source is measurable before failure, but it is not
+diagnostic of failure. The blind screen recovers the source among **101 validated zones** of
+downslope motion, 93 of them inventoried glaciers, and does not distinguish it by magnitude,
+persistence or coherence.
+
+No custom SAR processing was performed. The analysis uses only the operational NISAR Level-2 GOFF
+pixel-offset product, so every result can be regenerated from publicly archived granules.
+
+---
+
+## Contents
+
+```
+code/                     analysis and manuscript-build scripts (34 files)
+fields/                   per-pair offset fields over the 6 km source box (5 GeoTIFFs)
+tables/                   every statistic and null distribution quoted in the article,
+                          plus the machine-readable Supplementary Tables S1–S7
+inventory/                the footprint-wide candidate inventory (GeoPackage, shapefile, CSV)
+                          with attribution, controls and sensitivity results
+rasters/                  footprint screen rasters (EPSG:32645, 80 m)
+terrain/                  elevation, slope and downslope azimuth used by the screen
+extent_CC-BY-NC-4.0/      mapped debris-flow runout outline — DIFFERENT LICENCE, see above
+```
+
+### `fields/`
+`offsets_YYYYMMDD_YYYYMMDD.tif` — east, north, magnitude and directional-coherence bands over the
+6 km box around the source, for all five pairs. EPSG:32645, 80 m posting.
+
+### `tables/`
+`canonical_numbers.json` is the single source of truth: every number in the article is rendered from
+it or from one of the other artefacts here, and `code/verify_cee.py` checks that claim against the
+manuscript text. The remaining files hold the per-pair series, the size-matched null distributions,
+the detector variants, the noise floor and percentiles, the interval-length and symmetry tests, and
+the GAMMA cross-check. `exposure_verification.json` is the building/runout intersection behind the
+exposure counts. `TableS1`–`TableS7` are the Supplementary Tables in CSV and JSON;
+`TableS7_all_validated_zones` is Supplementary Data 1 of the article — all 101 validated zones with
+their attributes.
+
+### `inventory/`
+`candidates_attributed.*` is the blind screen output with glacier attribution against the Randolph
+Glacier Inventory 6.0. `canonical_numbers.json` here is the live copy (14,277 bytes), re-staged for
+this deposit; an earlier, shorter dump circulated in working copies and is superseded.
+
+### `terrain/`
+`elev_m.tif`, `slope_deg.tif`, `downslope_az_deg.tif`, derived from the **Copernicus DEM GLO-90**
+(90 m; © DLR e.V. 2010–2014 and © Airbus Defence and Space GmbH 2014–2018, provided under COPERNICUS
+by the European Union and ESA). Redistribution of these derived layers is permitted; the source DEM
+itself is not redistributed here and should be obtained from the Copernicus Data Space Ecosystem.
+
+---
+
+## Deliberately not included
+
+- **Full-footprint per-pair offset rasters** (4230 × 4311 px). Regenerable from the public granules
+  with `code/wide_screen.py`; available from the author on request.
+- **NISAR granules.** Public at the Alaska Satellite Facility DAAC; identifiers are listed in
+  `tables/TableS1_granules.csv` (track T098 ascending, L-SAR frequency A, HH).
+- **Geopera v1.1 stereo DSM / elevation-change products.** Third-party, CC BY-NC; cite the Geopera
+  dataset directly. Only our own stable-ground statistics on it are reported, in the article's
+  Supplementary Fig. 1.
+- **PlanetScope imagery.** Research licence, display only.
+- **Copernicus GLO-90 source tiles.** See `terrain/` above.
+
+## Known gap
+
+`displacement_null.py` was listed in the deposit plan but does not exist as a separate script; its
+output, `tables/displacement_null.json`, is included. The null-distribution machinery used elsewhere
+in the analysis is in `code/polygon_null.py` and `code/sensitivity_and_controls.py`.
+
+---
+
+## Citing
+
+Please cite the article. If you use this record directly, cite it as:
+
+> Meena, S. R. *Bhote Koshi 2026 ice–rock avalanche: NISAR L2 GOFF site series and footprint-wide
+> offset-tracking screen — data and code* (Zenodo, 2026). https://doi.org/10.5281/zenodo.22730419
+
+NISAR data are provided by NASA and ISRO and distributed by the Alaska Satellite Facility DAAC.
+Glacier outlines from the Randolph Glacier Inventory 6.0.

2026_初步评估/B_DeepSeek试解/外部数据/zenodo_22730419_bhotekoshi_flowextent/flow_extent.cpg

@@ -0,0 +1 @@
+UTF-8

2026_初步评估/B_DeepSeek试解/外部数据/数据清单.md

@@ -0,0 +1,63 @@
+# 外部数据清单（B 赛道初步评估）
+
+下载日期：2026-10-23。下载者：DeepSeek-V4.1-Flash（DeepSeek），经 OpenCode。
+总量：**47 个文件，3.5 MB**（限制 300 MB 内）。只含公开、无需登录、CC 许可的数据。
+TPDC 数据不在本清单内（需登录，主线程负责），见 `tpdc_priority.md`。
+
+## 1. Zenodo 10.5281/zenodo.22717809（CC-BY-4.0）
+
+标题：Data and code for: Optical and radar signals before the 2026 Gyirong cascade (v1.2.1)
+作者包：Ming J., Du Z. (2026), *Geophysical Research Letters*（投稿号 2026GL126799）
+下载：https://zenodo.org/api/records/22717809/files/gyirong_grl_open_research_v1.2.1.zip/content
+落盘：`外部数据/zenodo_22717809_gyirong_precursors/`（36 个文件，3.0 MB）
+
+| 关键文件 | 内容 | 用于哪问 |
+|---|---|---|
+| `detachment_aoi.geojson` | 崩解源区 AOI 多边形，bbox 85.505–85.540°E、28.262–28.292°N；属性含 `event_date=2026-08-26`、来源与“从 85.37°E 更正到 85.52°E”的说明 | Q2/Q4 源区定位 |
+| `era5_source_daily.csv` | 源区 ERA5-Land 逐日：2026-08-26 降水 **15.1 mm**（气候值 7.1，距平 +8.0），气温距平 +1.35 °C | Q1/Q2 触发 |
+| `era5_station_vs_source_hourly.csv` | 站点（**28.86°N, 85.29°E**）与源区逐小时 ERA5 对照 | Q1 站点代表性 |
+| `key_dates_comparison.csv` | 影像实际日期与目标日期对照（8/26 目标用 8/24 影像，云量 38.6%） | Q2 影像时相 |
+| `topo_rasters/tpdc_{slope_deg,thickness_m,tau_ice_kpa}.tif` | 源区 30 m 派生栅格（与附件 4 同源） | Q4 释放区 |
+| `topo_pixel_summary.csv` | 源区像元统计：坡度中位 37.2°，冰厚中位 43 m，剪应力 p90 232 kPa | Q4 参数 |
+| `sentinel1_velocity_pairs.csv`、`si/Table_S1–S4` | 事件前 SAR 速度对、S2 时序、周边冰川 TSI | Q2 前兆 |
+| `surrounding_glacier_inventory.csv` | 周边冰川清单（RGI 7.0 + TSI 排序，689 KB） | Q2 空间易发性 |
+
+**已核对**：该包内 `topo_pixel_summary.csv` 的冰厚均值 49.587 m 与附件 4 的 nc 计算值一致 → 说明附件 4 与公开派生栅格同源。
+
+## 2. Zenodo 10.5281/zenodo.22730419（数据 CC-BY-4.0；flow-extent 子集 CC-BY-**NC**-4.0）
+
+标题：Bhote Koshi 2026 ice–rock avalanche: NISAR L2 GOFF site series and footprint-wide offset-tracking screen
+下载：`bhotekoshi2026_flow-extent_CC-BY-NC-4.0.zip`（109 KB）+ `README.md`
+落盘：`外部数据/zenodo_22730419_bhotekoshi_flowextent/`（7 个文件，448 KB）
+
+- `flow_extent.shp/.dbf/.prj/.shx`：同一 8·26 事件在尼泊尔侧的**实测/解译流动范围多边形**，WGS84，bbox 85.147–85.448°E、27.923–28.337°N。多边形面积未在本文计算，可用于 Q3 空间范围与 Q4 淹没面积的**独立对照**。
+- 许可为 **NC**（非商业）：竞赛学术用途可用，但论文与支撑材料中须注明许可，不要用于任何商业演示。
+
+## 3. 只读取元数据、未下载的公开记录（不作为已获得数据）
+
+| DOI | 标题 | 为什么没下 |
+|---|---|---|
+| 10.5281/zenodo.23466792 | Cascading rock–ice avalanche and downstream flooding in the Nepal Himalaya | 记录无公开文件（仅元数据）；描述给出关键量：源区 **0.716 km²**、总地形损失 **110.80 Mm³** |
+| 10.5281/zenodo.22933601 | The 2026 Rasuwa–Trishuli mega rock–ice avalanche and cascading disaster | 文件为 196 MB GeoTIFF + 视频，超单文件预算且非必需 |
+| 10.5281/zenodo.12752901 | Aksaut Rock Ice Avalanche simulation in r.avaflow | 无公开文件，仅记录 |
+
+## 4. 引用与合规
+
+- 使用这些数据时按 DOI 引用；`flow_extent` 注意 CC-BY-NC。
+- 论文/支撑材料中不得声称“已获得 TPDC 登录数据”；TPDC 结果待主线程合并。
+- 数据仅用于本题分析，未上传至任何外部服务。
+
+## 5. 主线程经 TPDC 登录取到的数据（本轮审计对象）
+
+位置：`../../TPDC_补充数据/下载/<metadataUUID>.zip`，19 个 ZIP，138 MB，CRC 校验通过。
+本轮只抽样读了 4 包，未改动任何 ZIP：
+
+| UUID | 内容 | 本轮用法 |
+|---|---|---|
+| 147456c9-f740-4259-a2ea-fd7423613510 | GFS 降水/云/水成物（0.25°，起报 2026-08-25 00Z，f000–f072） | 提取源区/口岸时序，得到 14.6 mm/日、lead time 26.5 h |
+| be1c4b80-f26e-4f43-a54a-cedbfe06203b | TPCSM 5 km WRF（RAINNC、T2，逐小时增量） | 源区 8/26 16.3 mm；事件时刻 0 mm |
+| 65a266d2-c9c8-4f49-803b-cbeff12ff450 | 2021-03-22 色东普冰崩 SAC（7 台×3 分量） | 结构核对 + STA/LTA；无一致初动 |
+| 72aff770-2f75-459d-a5cc-77812f5b32be | 受灾点位 shp（28.2785°N, 85.3766°E） | 与路径终点相距 1.01 km |
+| e7f8d50b-f647-4efe-8d15-a87173c13ab8 | 河网 shp | 路径 7.15 km 后才贴河 |
+
+**撤回**：`10.11888/SolidEar.tpdc.303620` 不是灾害数据（实为地热水化学数据库），已从清单与经验中删除。