Avalanche exposure dataset for International Route 115-CH (Pehuenche Pass, Maule Region, Chile, ~36 S). This repository holds the input data, the corridor-scale simulation results and the road records used to assess the avalanche exposure of 60.64 km of road, from DM 97.62 near La Mina to DM 158.26, close to Pehuenche Pass, simulated with AvaFrame com1DFA.
Version 1.1 (see CHANGELOG.md). It accompanies the revised manuscript of the companion paper:
Latorre, I., Schaefer, M., and Oesterle, F.: How exposed is a mountain road to snow avalanches?
A corridor-scale assessment in the southern Andes, Nat. Hazards Earth Syst. Sci. Discuss.
[preprint], https://doi.org/10.5194/egusphere-2026-3951, 2026. Version 1.0 accompanied the
discussion paper.
All vector data use UTM zone 19S (EPSG:32719). Tables are comma-separated UTF-8 files with a header row; empty cells are missing values.
data/
releaseAreas/ the 208 release areas (corrected in 1.1), their scenario fractions and
the partial-release polygons of the reference
route/ road centerline and the 12 128 road points (corrected in 1.1)
reference/ per-case results of the reference simulations, critical friction, peak
pressure along the road and friction sweeps
candidateUnits/ the 356 candidate release units of the automatic delineation, their per-case
results and the tests of the method
observations/ the 35 records of the road authority and the nivology reports, and the
results of the tests with recorded avalanches
supplementTables/ Tables D1, D2, D3, D6, D7 and D8 of the data supplement of the paper
- Road position.
chainage_kmis the distance along the road points ofdata/route/road_points.csv, from 0 at the La Mina end. Official kilometers are DM = chainage + 97.62. A one-kilometer section k spans DM k - 0.38 to k + 0.62 (k = floor(chainage) + 98; 61 sections, km 98 to 158). The last road point is at chainage 60.635 km, DM 158.26. - Scenarios. Moderate, Severe and Extreme correspond to return periods of 10, 30 and 100 yr (T10, T30, T100) of the design new-snow depth of the 2021 study (storms of 2, 2.5 and 3 days). Fracture depths are measured normal to the slope. A case is one sub-path (one release area) in one scenario: 208 x 3 = 624 cases.
- Release conventions.
complete: the polygon released whole.fractions(main reading): the scenario fraction of each polygon (2021 study, medians of 25, 45 and 75 %), released as a contiguous region grown upward from the lowest cell of the polygon.uniform25,uniform50,uniform75: uniform fractions of 25, 50 and 75 %, also from the lowest cell (friction sweeps only). - Simulations. AvaFrame com1DFA 1.8.3 with samosAT friction (tau0 = 0, Rs0 = 0.222, R = 0.05 m, B = 4.13, kappa = 0.43), 5 m mesh, time step 0.1 s, SPH kernel radius 5 m, stop when the kinetic energy falls below 1 % of its peak, release density 250 kg m-3, outputs every 5 s. Terrain: the 12.5 m ALOS PALSAR terrain-corrected elevation layer (from SRTM GL1) resampled to 5 m, clipped to six simulation zones (not included). The flows of four sub-paths (A11-4, A65, A66-2 and A71c) reached the edge of their zone models; their reference runs and their sweeps without the layer come from runs on terrain models extended over the slopes that drain to the road, and their sweeps with the layer stayed on the zone models.
- Reference. Friction mu assigned from the released volume with the values of the standard
samosAT sets in steps of 0.01 (0.22 up to 12 500 m3, 0.17 at 38 730 m3, 0.155 from 60 000 m3,
linear in log10 V between them), plus an
erodible layer of 0.30 m (entrained density 200 kg m-3, default erosion parameters of com1DFA)
outside the release and above a lower edge set 100 m above the road crossing of the case
(column
layer_lower_edge_m). - Friction sweeps. Constant mu, with the erodible layer of the reference (main reading) and, as a control, without it. Complete polygons: mu from 0.13 to 0.42 in steps of 0.01, plus 0.155 (31 values). Scenario fractions: 0.13 to 0.30 in steps of 0.01, plus 0.155, 0.33, 0.36 and 0.42 (22 values). Uniform fractions: those 22 values without the layer and the eight values of the standard range (0.155 and 0.16 to 0.22) with it.
- Crossing. A simulation crosses the road when its peak pressure (ppr) exceeds 10 Pa at any of the 12 128 road points, without buffer (the raster cell that contains each point). The same peak rasters were sampled again when the road axis was corrected; no simulation was repeated for that reason.
- Arrival time. End of the first 5 s output interval in which the flow thickness exceeds 0.01 m at a road point.
The 208 release areas, one polygon per sub-path, identical across scenarios. Polygons were digitized from the georeferenced map sheets of the 2021 study (in version 1.0.2 of this data set).
| Field | Description |
|---|---|
| name | Sub-path (release area) name; the letters and numbers after the path number identify sub-paths of one catchment (71 catchments) |
| mapSheet | Number of the map sheet of the 2021 study; empty for 75 sub-paths (absent from the source table) |
| aspect | Aspect class of the release area (N, NE, E, SE, S, SW, W, NW); empty for 13 sub-paths |
| areaM2 | Planimetric area (m2) |
| dT10, dT30, dT100 | Fracture depth (m, normal to the slope) in the Moderate, Severe and Extreme scenarios, from the 2021 study |
| volT10, volT30, volT100 | areaM2 x fracture depth (m3), rounded to 1 m3. The simulations used the surface area on the 5 m terrain model, which gives volumes larger by a median factor of 1.20 (column volume_m3 of the reference tables) |
The shapefile was written with pyshp to keep the vertex order of each ring. Polygon A24a is stored counter-clockwise, the order used in the simulations: com1DFA 1.8 seeds its particles with an angle heuristic that depends on the vertex order, and with the clockwise ring of version 1.0 it initialized only 72 % of the volume of this narrow concave polygon. GDAL-based tools (QGIS, ogr2ogr, geopandas with the default engine) rewind rings to clockwise when they write a shapefile; use the file as distributed, or a writer that keeps the ring order, to reproduce the simulations.
Fraction of each polygon released in each scenario (release-area table of the 2021 study, Fase 4 Vol. 1 Tomo 3, pp. 98-100; the study is in version 1.0.2 of this data set).
| Column | Description |
|---|---|
| name | Sub-path, as in releaseAreas115ch.shp |
| mopCode | Code of the release area in the table of the 2021 study (AL68f covers A68f1-A68f4 and AL68i covers A68i1-A68i2) |
| fT10, fT30, fT100 | Fraction of the polygon area released in the Moderate, Severe and Extreme scenarios |
The 624 partial-release polygons of the reference with the scenario fractions (one per sub-path and scenario), as simulated, cut from the lowest cell of each polygon (see cutRule). The vertex order is the one used in the simulations (see the note on A24a); keep it when converting the layer.
| Field | Description |
|---|---|
| name | Sub-path |
| scenario, returnPeriod | Scenario (Moderate, Severe, Extreme) and its return period (yr) |
| fTarget | Fraction of the polygon to release (scenario_fractions.csv) |
| cutRule | contour: region grown from the lowest cell over the lightly smoothed terrain model (601 polygons); grow: the same region grown best-first, used where the first cut failed (8); line: straight cut normal to the mean fall line (4); whole: no acceptable cut (target below 40 cells or failed checks), polygon released whole (11) |
| fPlan | Released fraction of the planimetric area |
| fSurf | Released fraction of the surface area on the terrain model |
| depthM | Fracture depth (m, normal to the slope) |
| areaM2 | Planimetric area of the partial polygon (m2) |
| volSurfM3 | Released volume on the terrain model (m3): surface volume of the complete polygon x fSurf; this volume sets mu |
| volPlanM3 | Planimetric released volume (m3): planimetric volume of the complete polygon x fPlan |
| demZone | Simulation zone (1 to 6) |
route115ch.shp: Route 115-CH centerline, two features (field segment): the simulated road and
its extension to Pehuenche Pass, which was not simulated; lengthM is the length of each. The
simulated road is the axis corrected in version 1.1 (see CHANGELOG.md).
road_points.csv: the 12 128 road points every 5 m used for the exposure analysis (x, y in m;
chainage_km).
One row per case and release convention (2 x 624 = 1248 rows): the reference simulations of the revised paper, with the erodible layer. The crossing columns are the fixed crossing point of the case, the same for both conventions.
| Column | Description |
|---|---|
| release | fractions (scenario fractions, lower part; main reading) or complete (complete polygon) |
| name | Sub-path |
| catchment | Catchment of the sub-path (71 catchments) |
| scenario, return_period_yr | Scenario and its return period (yr) |
| fraction_target | Fraction of the polygon to release (1 for complete polygons) |
| fraction_released | Released fraction of the surface area on the terrain model (fSurf; 1 for complete polygons and for the 11 polygons released whole) |
| volume_m3 | Released volume on the terrain model (m3) |
| mu | Coulomb friction coefficient of the reference, from volume_m3 |
| crossing_chainage_km, crossing_dm_km | Fixed crossing point: the road point of largest peak pressure in the complete-polygon sweep at the lowest mu that crosses the road; if no mu of the sweep crosses (19 sub-path scenarios), the road point nearest to the centroid of the polygon |
| km_section | One-kilometer section of the case, from the crossing point |
| crossing_x, crossing_y | Coordinates of the crossing point (m, EPSG:32719) |
| crossing_z_m | Elevation of the crossing point on the terrain model of the zone (m) |
| crossing_source | sweep or centroid (see crossing_chainage_km) |
| layer_lower_edge_m | Lower edge of the erodible layer used in the run (m): crossing elevation + 100 m, set before the final corrections of the road sampling and of the road axis; because they changed the crossing elevation of some cases, it differs from crossing_z_m + 100 by -9 to +17 m in 298 of the 624 cases (by 1 or 2 m in 244 of them) and equals it in the rest |
| reach | 1 if the peak pressure exceeded 10 Pa at a road point |
| max_ppr_kPa | Largest peak pressure on the road (kPa; 0 without crossing) |
| max_ppr_chainage_km | Chainage of that road point (empty without crossing) |
| road_len_m | Road length with peak pressure above 10 Pa (m; 5 m per road point) |
| len_ge3kPa_m, len_ge30kPa_m, len_ge100kPa_m | Road length with peak pressure of at least 3, 30 and 100 kPa (m) |
| arrival_s | Arrival time (s); empty without crossing and for one crossing case (A22, Moderate, fractions) whose flow thickness at the road never exceeded 0.01 m at a saved step; 0 where the release itself covers road points (A20b, A43a, A45a, A45b) |
| mass_gain | Relative mass gain of the run from the erodible layer (final mass / initial mass - 1) |
| mu_crit, mu_first_stop, mu_crit_state, monotonic, standard_range_outcome | Critical friction with the erodible layer, in the sweep of the same release convention (see critical_friction.csv) |
| reach_no_layer, max_ppr_no_layer_kPa, arrival_no_layer_s | Road outcome of the run with the same mu and without the layer, as a control |
| mu_crit_no_layer, mu_first_stop_no_layer, mu_crit_state_no_layer, monotonic_no_layer, standard_range_outcome_no_layer | Critical friction without the layer, as a control |
Critical friction of every case for five release conventions (5 x 624 = 3120 rows). With the erodible layer the main reading, and without it as a control. Grids: see Friction sweeps above. The polygons of the uniform fractions are not included; they were cut with the same rule as the scenario fractions.
| Column | Description |
|---|---|
| release | complete, fractions, uniform25, uniform50 or uniform75 |
| name, scenario, return_period_yr | Case |
| volume_m3 | Released volume on the terrain model (m3) |
| layer_grid | Grid of the sweep with the layer: full (complete polygons and scenario fractions) or standard (the eight values of the standard range, uniform fractions; the critical friction with the layer of these rows only says where the outcome changes inside that range) |
| mu_crit | Largest mu of the sweep with the layer whose flow crossed the road (empty if none crossed) |
| mu_first_stop | Smallest mu of that sweep whose flow did not cross the road (empty if all crossed) |
| mu_crit_state | between (the outcome changes inside the sweep), always (crossed with every mu: with the full grid, critical friction of 0.42 or more) or never (crossed with none: with the full grid, below 0.13) |
| monotonic | 1 if the case crossed with every mu up to mu_crit and with none from mu_first_stop on; 0 otherwise (2 cases with complete polygons and 1 with scenario fractions). For monotonic cases the road outcome of any constant mu can be read from mu_crit |
| standard_range_outcome | Road outcome within the standard friction range, mu = 0.155 to 0.22: always (crossed with every value), never (with none) or changes. Cases that change: 79 of 624 with scenario fractions and 73 of 624 with complete polygons |
| mu_crit_no_layer, mu_first_stop_no_layer, mu_crit_state_no_layer, monotonic_no_layer, standard_range_outcome_no_layer | The same from the sweeps without the layer (22 or 31 values). Cases that change within the standard range: 92 and 80 |
Peak pressure of the reference simulations along the road (gzip-compressed CSV): one row per
crossing case, release convention and road point with peak pressure above 10 Pa (42 366 rows). The
union of idx by convention gives the road reached: 25.655 km with the scenario fractions and
27.895 km with complete polygons.
| Column | Description |
|---|---|
| release, scenario, name | Case |
| idx | Road point (row of road_points.csv, from 0) |
| chainage_km | Chainage of the road point |
| ppr_kPa | Peak pressure (kPa) |
Road outcome of the friction sweeps of complete polygons and scenario fractions (gzip-compressed
CSV; 66 144 rows), from which the critical friction was computed and the road outcome of any
volume-based assignment of mu can be read. A case crosses the road when max_ppr_kPa exceeds 0.01.
| Column | Description |
|---|---|
| release | complete or fractions |
| layer | yes (with the erodible layer of the reference) or no |
| name, scenario | Case |
| mu | Constant Coulomb friction coefficient of the run |
| max_ppr_kPa | Largest peak pressure at the road points (kPa) |
The 356 candidate release units (1805 ha, median 2.3 ha) of the check of the completeness of the
inventory in the revised paper (Sect. S1 of the Supplement). Potential release areas were
delineated with the pixel rules of the extreme scenario of Buhler et al. (2018,
https://doi.org/10.5194/nhess-18-3235-2018) on terrain models extended to the slopes that drain to
the road, without tree and shrub cells of ESA WorldCover 2021, and cut into release areas with an
open implementation of the object-based segmentation of that study: Baatz and Schape region merging
as implemented in the Generic Region Merging application of Orfeo ToolBox 9.1.1, on the aspect
sectors (weighted three times), the filtered slope and the fold, followed by a second, coarser
merging of these objects with the same criterion (Python). The scales were set by running the same
segmentation on swissALTI3D resampled to 5 m at three test sites near Davos and comparing it with the
objects that SLF published there (Buhler and von Rickenbach, 2018); neither the objects nor the
terrain model is redistributed (see NOTICE). Parameters are in Table S1 of the
Supplement. Units are the parts of the objects outside the inventory that pass this
screen: at least 2500 m2, at least half of their cells draining to the road along D8
steepest-descent paths with H/L of at least 0.245, road met outside the road length already reached
by the inventory. No candidate is larger than the largest polygon of the inventory (A34-2, 32.17 ha):
the five units above that size (B10185, B10198, B10214, B10264 and B10305) were cut again, with that
size as the limit, into 12 pieces that replace them. A candidate unit is not a confirmed release
area. Each unit was simulated with the reference configuration on the extended terrain models, released
whole and with the median scenario fraction from its lower part, in the three scenarios: 2094
simulations (1068 whole units and 1026 scenario fractions; the 42 cases whose scenario fraction was
under 40 cells were released whole and not run again). The polygons follow the cell edges of the 5 m
terrain models.
The delineation and the screen ran before the correction of the road axis, on the former axis
(road_reached_inventory.csv.gz below), which equals the current one up to DM 143.855; kmD8 comes
from that axis. The road outcomes were sampled on the current axis.
| Field | Description |
|---|---|
| unitId | Unit identifier: B, the extended terrain model (1, 2, 3, 4 or 56) and a four-digit number in the order of kmD8 within the model |
| demZone | Extended terrain model (1, 2, 3, 4, or 56 for zones 5 and 6 merged) |
| areaM2 | Planimetric area (m2) |
| zMean | Mean elevation (m) |
| hlMax | Largest ratio of elevation drop to D8 path length from a cell of the unit to the road |
| kmD8 | One-kilometer section where most D8 paths of the unit meet the road |
| newSection | 1 if kmD8 is one of the sections without crossings from the inventory (222 units) |
| parent_cid | For the 12 pieces of the size cap, the unit they were cut from (B10185, B10198, B10214, B10264 or B10305); empty for the other units |
| dT10, dT30, dT100 | Fracture depth (m, normal to the slope) from the rule of the 2021 study (mean elevation and slope of the cells released with the whole unit, neutral wind), used in both readings |
| reachFrac, reachComp | 1 if the unit crossed the road in at least one scenario, with the median scenario fraction (Frac) or released whole (Comp) |
| nCrossFrac, nCrossComp | Number of scenarios with a crossing |
| pprMaxFrac, pprMaxComp | Largest peak pressure on the road over the three scenarios (kPa) |
| kmCrossFrac, kmCrossComp | One-kilometer sections with peak pressure above 10 Pa in some scenario (space separated) |
| edgeFrac, edgeComp | 1 if the flow (above 10 Pa) of a simulation of the unit touched the edge of its terrain model |
One row per object-based unit, scenario and release convention (356 x 3 x 2 = 2136 rows), ordered as
data/reference/cases_reference.csv: release, scenario, unit. Crossings, road lengths and arrival
times follow the conventions of the reference tables (12 128 road points, 10 Pa, no buffer). With the
scenario fractions 81 units cross the road, in 173 cases; released whole, 92 units, in 223 cases.
Nine sections without crossings from the inventory receive crossings (km 98 to 101, 103, 104, 108, 111
and 136). Units of zone 1 ran on windows of the extended zone-1 model; the 57 cases whose flow touched
the edge of their window were run again on the whole model (columns rerun_*), and the main columns
keep the window runs, as in the paper.
| Column | Description |
|---|---|
| release | fractions (median scenario fraction from the lower part of the unit) or complete (unit released whole) |
| unit_id | Unit, as unitId in candidateUnitsOBIA115ch.gpkg |
| scenario, return_period_yr | Scenario and its return period (yr) |
| fraction_target | Fraction of the unit to release: 0.25, 0.45 and 0.75 (median scenario fractions of the 2021 study in the Moderate, Severe and Extreme scenarios); 1 for complete units |
| fraction_cells | Released fraction of the cells of the complete release (1 for complete units and for the fractions released whole) |
| cut_rule | Scenario fractions only, as cutRule of partialReleases115ch.gpkg: contour (963 cases), grow (46), line (17) or whole (42: target under 40 cells, unit released whole) |
| result_from_complete | 1 for the 42 whole cases: they were not run again and repeat the result of the complete unit in the same scenario |
| fracture_depth_m | Fracture depth (m, normal to the slope), as dT10, dT30 and dT100 of the unit |
| release_area_m2 | Planimetric area of the released cells (m2); for complete units, the unit with its holes filled |
| volume_m3 | Released volume on the terrain model (m3): surface area of the released cells x fracture depth; this volume sets mu |
| mu | Coulomb friction coefficient of the reference, from volume_m3 |
| layer_lower_edge_m | Lower edge of the erodible layer (m): road elevation at the D8 arrival of the highest cell of the unit that drains to the road, + 100 m |
| terrain_model | Terrain model of the run: extended zone model (2, 3, 4, 56) or window of the extended zone-1 model (1-W1 to 1-W7) |
| reach | 1 if the peak pressure exceeded 10 Pa at a road point |
| max_ppr_kPa | Largest peak pressure on the road (kPa; 0 without crossing) |
| max_ppr_chainage_km | Chainage of that road point (empty without crossing) |
| km_section | One-kilometer section of that road point (empty without crossing) |
| km_sections | One-kilometer sections with peak pressure above 10 Pa at some road point (space separated) |
| km_sections_new | The sections of km_sections without crossings from the inventory, i.e. outside the 48 one-kilometer sections with crossings from the inventory |
| road_len_m | Road length with peak pressure above 10 Pa (m; 5 m per road point) |
| len_ge3kPa_m, len_ge30kPa_m, len_ge100kPa_m | Road length with peak pressure of at least 3, 30 and 100 kPa (m) |
| arrival_s | Arrival time (s); empty without crossing and for four crossing cases whose flow thickness at the road never exceeded 0.01 m at a saved step; 0 where the release covers road points (14 units) |
| edge_contact | 1 if cells with peak pressure above 10 Pa touched the edge of the terrain model of the run |
| rerun_reach, rerun_max_ppr_kPa, rerun_km_sections, rerun_edge_contact | For the 57 re-run cases (37 complete, 20 scenario fractions), reach, largest peak pressure (kPa), sections above 10 Pa and edge contact on the whole extended zone-1 model; empty for the other cases |
Road axis of the delineation and road length already reached by the simulations of the inventory, used to screen the candidate units (gzip-compressed CSV, one row per road point, 12 209 rows). This is the axis of the discussion paper: it equals the current axis up to DM 143.855 and is 403 m longer beyond DM 151.6. It is kept because the screen ran on it; do not use it for the results of this version.
| Column | Description |
|---|---|
| idx | Road point index on this axis (from 0) |
| x, y | Coordinates of the road point (m, EPSG:32719) |
| chainage_km, km | Chainage and one-kilometer section on this axis |
| union_d10_all | Largest peak pressure (kPa) at the point over the friction sweeps of the inventory without the layer, with mu from 0.155 to 0.22 and any release convention (complete polygons, scenario fractions and uniform fractions from the lower and upper part) |
| refw_all | Largest peak pressure (kPa) at the point over the reference simulations of the inventory (both release conventions, three scenarios) |
| reached_amp | Largest of the two (kPa); a point counts as reached above 10 Pa (5866 points). A unit stays a candidate only if less than half of its draining cells meet the road within 25 m of a reached point |
Tests behind the method table of the Supplement (statistics only; no object or terrain model of the test sites).
| File | Description |
|---|---|
| davos_level1.csv | First level of the segmentation against the frequent SLF objects at Davos, by scale and site |
| davos_level2.csv | Second level against the extreme SLF objects, by reading of the region growing, scale and site |
| grm_check.csv | Python merging against the Generic Region Merging application of Orfeo ToolBox from single cells, on windows of the extreme potential release areas |
| corridor_search.csv | Scales of the second level on the corridor against the inventory polygons, with the candidates each one would give |
| depth_rule_summary.csv | Fracture depths of the inventory against the rule of the 2021 study with neutral wind, by reading, scenario and elevation band |
Columns of davos_level1.csv and davos_level2.csv:
| Column | Description |
|---|---|
| site | Test site (Parsenn, Jakobshorn, Rinerhorn), or all for the three pooled |
| s1, min1 | Scale of the first level and smallest object kept (cells) |
| var, par, s2, ds, min2, fmax, dsmax | Second level: reading of the region growing (bs, uf or mrs), its parameters as text, scale, sector tolerance, smallest object kept (cells), largest mean fold across a boundary (deg) and largest difference of mean slope (deg) |
| n, iterations, seconds, mixed, converged | First level: segments, iterations and run time (s) of the merging, segments that mix release and other cells, and whether it stopped before the iteration limit |
| n1_raw, domain_added_cells, n1, seconds_absorb1 | First-level objects before absorbing small ones, cells added by filling holes, objects after absorbing them, and the time of that step (s) |
| n_ours, med_ours_ha, q25_ours_ha, q75_ours_ha, q90_ours_ha, n_ours_all | Our objects: number, median and percentiles of area (ha); in the pooled rows of the second level, the number of objects compared |
| n_slf, med_slf_ha, q25_slf_ha, q75_slf_ha, q90_slf_ha | The same for the SLF objects |
| best_iou_mean, best_iou_median | For each SLF object, the best intersection over union with one of our objects: mean and median |
| pieces80_median | Median number of our objects needed to cover 80 % of the cells of an SLF object |
| oto_rate | Share of SLF objects with a one-to-one match: one of our objects holds at least 50 % of the SLF object and has at least 50 % of its own cells in it |
| cov_mean | Mean share of the cells of an SLF object inside our potential release area |
| ks | Kolmogorov-Smirnov distance between the log areas |
| size_ratio | Median area of our objects / median area of the SLF objects |
Columns of corridor_search.csv:
| Column | Description |
|---|---|
| var, s1, min1, s2, ds, min2 | Reading of the second level (l1, mrs, bs or d8) and its parameters, as above |
| polygons, pod_n | Inventory polygons compared and the share with a one-to-one match |
| iou_mean, iou_median, iou_ge05 | Best intersection over union of a polygon with one segment: mean, median and share of at least 0.5 |
| pieces80_median, pieces80_p75 | Segments needed to cover 80 % of a polygon: median and 75th percentile |
| seg_in_inv_n, seg_in_inv_med_ha | Segments with at least 50 % of their cells in the inventory: number and median area (ha) |
| best_seg_med_ha, poly_med_ha | Median area of the best-matching segments and of the polygons (ha) |
| units, cand | Units and candidate units |
| cand_ha, cand_med_ha, cand_q25_ha, cand_q75_ha, cand_max_ha | Candidate units: total, median, quartiles and largest area (ha) |
| cand_newkm_d8, n_newkm_d8 | One-kilometer sections without inventory crossings met by the D8 lines of the candidates: list and number |
Columns of grm_check.csv:
| Column | Description |
|---|---|
| site, window, pra_cells, scale | Test site or corridor zone, window (rows and columns), cells compared and scale |
| grm_segments, python_segments | Segments from each implementation |
| grm_iterations, python_passes | Iterations of each implementation |
| ari | Adjusted Rand index of the two partitions (1 = identical) |
Columns of depth_rule_summary.csv:
| Column | Description |
|---|---|
| reading | Reading of the elevation terms of the rule: text (constant from 1600 to 2000 m, +6 % per 100 m above) or applied (6 % per 100 m over the whole range) |
| period, band, n | Scenario (t10, t30, t100), elevation band (m) and polygons |
| median, p25, p75, p10, p90 | Ratio of the depth of the study to the depth of the rule |
| share_within_20pct, share_below_1 | Share of polygons with a ratio from 0.8 to 1.25, and below 1 |
| d0_study_median, d0_rule_median | Median depth of the study and of the rule (m) |
The 35 records of the road authority and the nivology reports used in the coverage check of the paper (Sect. 3.5 of the main text and Sect. S4 of the Supplement): avalanches with confirmed arrival on the road, a point on the road axis, location uncertainty of 300 m or less, no rockfall, and a written report. 34 of the 35 are covered with the scenario fractions. Points derived from the unpublished avalanche outlines of Viento Puelche Expediciones, which the paper also uses, are given only in its Supplement and are not part of this data set.
| Column | Description |
|---|---|
| entry_id | Identifier of the record in the merged catalog of the paper: VIA-AVALn-yyyy, register of the 2021 study; VIA-INT14-km, 2014 study; LC-yyyy-nn, nivology reports of the Los Condores hydropower project |
| date, date_start, date_end | Day of the record, or year of its season; first and last day (ISO 8601), where known |
| date_precision | day, range, season (winter known, date unknown) or undated |
| season | Winter of the avalanche (empty if undated) |
| dm_km, chainage_km, section_km | Position of the record on the road (DM = chainage + 97.62) and its one-kilometer section |
| x, y | Road point at that chainage (m, EPSG:32719) |
| dm_posts_km | Position read from the official road posts, as given or implied by the source |
| location_uncertainty_m | Location uncertainty (m) |
| catchments | Catchments of the sub-paths whose simulations reach the road within 100 m of the record |
| source | R, register and records of the 2021 study; I, 2014 study; N, reports of the nivology team of the Los Condores project |
| source_reference | Source in words |
| distance_m_scenario_fractions, distance_m_complete_polygons | Distance along the road to the nearest road point where the reference exceeds 10 Pa, in any scenario (m) |
| covered_scenario_fractions, covered_sf_moderate, covered_sf_severe, covered_sf_extreme, covered_complete_polygons, covered_with_candidate_units | Road reached within 100 m of the record (yes/no), with the scenario fractions (any scenario, or by scenario), with complete polygons, or with the scenario fractions and the candidate units |
| best_documented | yes for the best-documented avalanches of Table 2 of the main text that are in this list |
| documented_event | Name of that avalanche |
| path | Avalanche path named by the source (codes of the 2021 study) |
| reached_road | yes for all records |
| snow | Snow type given by the source: wet, dry, mixed, dry or mixed, cornice fall, unknown |
| note | Short remark |
| description | Short description from the sources, written for this data set |
The tests with recorded avalanches of Sect. S4 of the Supplement, one row per test. Distance: front past the road crossing of the thalweg (reached) or past the observed toe (others), with the reference. Result: combinations in which the reference reaches the road (for deposits that end short of the road, the road is not the criterion). Across the valley, the runs carry a 0.30 m layer over the whole terrain model. The columns hold the values as printed in that table, so ranges are text ("a to b").
| Column | Description |
|---|---|
| avalanche | Identifier of the record and date |
| path | Sub-path or terrain of the test and its one-kilometer section |
| observation | What was observed |
| releases | Release variants x depths (depths in m) |
| volume_m3 | Released volume on the terrain model (m3) |
| mu | Friction assigned by the reference to those volumes |
| result | Combinations in which the reference reaches the road or meets the criterion of the test |
| distance_m | See above (m) |
| critical_mu | Largest mu that still reaches the road or passes the observed limit |
| note | Remarks on the row |
Tables D1, D2, D3, D6, D7 and D8 of the data supplement of the paper, as CSV files, cited by the main text and the Supplement. Tables D4 and D5 hold positions derived from unpublished outlines and are given only in the Supplement.
| File | Description |
|---|---|
| D1_runs_by_variant.csv | Runs of the design of the simulations, by run set and variant |
| D2_consistency_check.csv | Results of the consistency check of the release-area inventory with the automatic delineation |
| D3_new_sections_candidate_units.csv | One-kilometer sections without inventory crossings that the candidate units reached, by release convention, with the units that cross and those whose outcome changes within the standard friction range |
| D6_sensitivity_runs.csv | Sensitivity runs: releases of 10 %, control from the upper part, fracture depth, release position and release-friction grid |
| D7_critical_friction.csv | Critical friction with the erodible layer: censoring, variance, margins, mass and class, Coulomb bound, regressions, beta point, slope at the crossing, and sections km 155 and 128 |
| D8_volume_ladder.csv | Volume ladder of 73 inventory sub-paths: releases of 40 to 640 cells from the lower part of the polygon, with the friction the reference assigns to each volume and its layer |
The results of the discussion paper (data/results/ of version 1.0.2: runout data, peak pressure
along the road and arrival times with two friction classes and complete polygons) are superseded
by data/reference/ and were removed from version 1.1. They remain available in version 1.0.2
of the Zenodo record.
- Release areas, fracture depths, scenario fractions and map sheets derive from the avalanche engineering study commissioned by the Direccion de Vialidad, Ministerio de Obras Publicas, Chile (MOP, 2021; prepared by Ingenieria Cuatro), georeferenced and digitized for this work.
- Road records: register of avalanches 2016-2018 and 2011 record of the 2021 study, avalanches observed in the 2013 field visits of the 2014 study (Direccion de Vialidad; prepared by Intrat Consultores), and nivology reports of Viento Puelche Expediciones for the ENEL Los Condores project (2020 and 2022).
- This version does not redistribute third-party documents. The map sheets and studies of the Direccion de Vialidad and the official closure records of the route remain in version 1.0.2 of this data set; the nivology reports are cited in the paper.
- Land cover for the delineation: ESA WorldCover 10 m 2021 v200 (https://doi.org/10.5281/zenodo.7254221). Terrain: ALOS PALSAR terrain-corrected elevation layers (Alaska Satellite Facility, https://search.asf.alaska.edu/), not redistributed here.
- Object-based segmentation: Generic Region Merging application of Orfeo ToolBox 9.1.1 (https://www.orfeo-toolbox.org/). Its parameters were set with the potential release areas published by SLF for three test sites near Davos (Buhler, Y. and von Rickenbach, D.: Automated Avalanche Release Area (PRA) Delineation Davos, EnviDat, https://doi.org/10.16904/envidat.55, 2018; ODbL) on swissALTI3D (Federal Office of Topography swisstopo, https://www.swisstopo.admin.ch/en/height-model-swissalti3d); neither is redistributed here.
- A video record of the 2011 Campanario avalanche, "Avalancha de nieve cubre tramo de ruta CH-115" (uploaded by Red Maule), is cited as an external source on YouTube (https://www.youtube.com/watch?v=TV_tc86XjaI) and is not redistributed here.
- Simulations performed with AvaFrame (https://avaframe.org/), com1DFA module, version 1.8.3.
The data in data/ and this documentation are released under the
Creative Commons Attribution 4.0 International license (CC BY 4.0); see LICENSE.
Third-party sources are not covered by that license (see NOTICE):
- The Campanario 2011 avalanche video linked above is an external source by Red Maule, not redistributed here; rights remain with the uploader.
Ivan Latorre (ivanlatorre@inh.cl), Instituto Nacional de Hidraulica, MOP, Chile.