Table 1 Uncertainties towards volume estimates.

Subaerial failure plane gradient from CSK imagery estimated 30–40°, maximum 61°⁵. Failure plane gradient variations may account for uncertainty in volume around 10% of 0.098 km³. The base DEM has a 2 m per pixel resolution. The vertical accuracy of the January 2018 DEM was 3.7 m and the standard error at 2.2 m. The new DEM accounting for 2019 edifice growth provides additional uncertainty increasing from 10% up to 20%.

0.098 km³

20% (up to 30%)

±0.019 km³ (±0.028 km³)

Initial submarine landslide deposit volume estimate

Bathymetry resolution

The spatial resolution of the 1990 bathymetry is 100 m and 5 m in the 2019 bathymetry. The block S of Sertung is resolved in both surveys and is located in precisely the same location; while edges of the block in 1990 bathymetry are smoothed due to low resolution. Height of the same central location on the block is −182 m in gridded 1990 data and −163 m in the 2019 data; however, there is 14 ms TWT (11.9 ± 0.7 m) of drape in 2017 seismic reflection data (line SS-11C). This drape estimate is consistent with estimated 5–10 m of accumulation in the basin between 1990 and 2019. The block height in 2019 is closer to −175 m, implying the resolution difference in the data cause vertical errors of up to 10 m. Other comparative points of lower complexity on the basin margins indicate vertical differences of <5 m.

0.236 km³

~20% conservatively (~15% realistically)

±0.05 km³ (±0.036 km³)

Bathymetry offsets in 2019 data

The full effects of tides and directionality of the 2019 bathymetric survey cannot be fully corrected between survey lines. Vertical differences between survey lines of up to 8.2 m are measured on the high gradient, proximal slopes, 1.9 m over the landslide blocks and 0.8 m over the basin.

~2.7% total (~4% over slope area

~2% over slide area)

±0.013 km³ (±0.008 km³

on slope ±0.005 km³ over slide)

Landslide volume adjustment

Post-event sediment inclusion

Backfilled sediment accumulation landward of the landslide blocks on average 14 m over a slope area of 0.82 km². This accounts for 0.012 km³, with a likely range 0.008–0.016 km³.

0.008–0.016 km³

Up to 25%

±0.004 km³

Pre-event sediment inclusion

The initial landslide deposit volume may include marine sediment deposition between the 1990 baseline and December 2018 landslide. Several methods are used to resolve that 0.035 ± 0.007 km³ of the initial volume may need to be removed:

1) Compare published low-resolution MBES surveys in 2016 (pre-collapse) and 2018¹² (post-collapse) to the 1990 pre-collapse multibeam bathymetry⁹ and new 2019 high-resolution post-collapse multibeam bathymetry. Negligible-to-5 m accumulation implying up to 0.036 km³, but likely less, uncertainty from comparing bathymetries of different resolutions (Supplementary 12);

2) Compare the basin floor in 2017 from pre-collapse seismic reflection profiles to basin floor bathymetry in 1990 and 2019. Negligible-to-8 m accumulation implying up to 0.058 km³ volume (Figs. 7, 8), with uncertainty from velocity model used causing ±1 m uncertainty;

3) Estimate sedimentation rate from pre-collapse 2017 seismic reflection data to project likely 1990–2018 sedimentation. This equates up to 8–10 m accumulation of 0.072 km³, with uncertainty from velocity model used causing ±1 m uncertainty;

4) Compare two-way travel time-converted 1990 bathymetry to our 2017 and 2019 seismic reflection profiles (Figs. 7, 8). This equates <5 m accumulation of 0.036 km³, with uncertainty from velocity model used causing ±1 m uncertainty.

0.036–0.072 km³

1) 20%

2) ~12%

3) ~10%

4) 20%

1) 0.007 km³

2) 0.007 km³

3) 0.007 km³

4) 0.007 km³

Erosion into older strata

Erosion into the pre-collapse stratigraphy is up to 15 m beneath the largest landslide blocks, which decreases to <5 m towards periphery. This equates 0.025 km³ (Fig. 9C), with uncertainty based on velocity model used to covert seismic reflection data equating ±2 m uncertainty. Interpolation across the area presents an additional but unquantified uncertainty.

0.025 km

~15%

0.004 km³

Interpolations have a potential error based upon the velocity model used to depth convert the seismic reflection data. This is estimated to be 1–2 m of the 5–10 m thickness used to calculate the 0.022 km³ volume.

0.022 km³

~25%

0.006 km³

Tidal vertical offset in bathymetry survey lines is 0.8 m, equating a potential error of up to 0.004 km³. Uncertainty in a 1600–1800 m/s velocity model for depth converting surfaces from seismic reflection profiles provides uncertainty up to 2 m.

0.023 km³

~15%

0.004 km³