Cliff coast collapses driven by nested biological, astronomical and meteorological activity cycles

Direct links between cliff erosion and forcing mechanisms are poorly constrained, largely due to the difficultly of obtaining precise timing information for individual failure events. Here we use two years of seismic records and auxiliary data to precisely detect and locate 81 failure events at the chalk cliff coast of Germany’s largest island, Rügen. The sub-second timing precision allows the linking of individual events to triggers over a wide range of relevant time scales. We show that in the monitoring interval, marine processes were negligible and cliff failure was associated with terrestrial controls on moisture. Failures were mostly triggered when water caused a state transition from solid to liquid. Water content can be changed by i) subsurface flow towards the cliff, ii) rain onto the cliff and iii) condensation of air moisture, leading to clustered events during night time. Failure periodicity is in alignment with the lunar cycle. Seasonal water availability, controlled by plant activity, sets cliff dynamics at the annual scale. Wetter and drier than average years impose a month-long legacy effect for cliff dynamics. This paper is a non-peer reviewed preprint uploaded to EarthArXiv, and submitted to “Journal of Geophysical Research: Earth Surface”. This is the first submitted version of the manuscript. Potsdam, 12 December 2019 manuscript submitted to Journal of Geophysical Research: Earth Surface Cliff coast collapses driven by nested biological, 1 astronomical and meteorological activity cycles 2 M. Dietze, K. L. Cook, L. Illien, O. Rach, S. Puffpaff, I. Stodian, N. 3 Hovius 4 1GFZ German Research Centre for Geosciences, Section 4.6 Geomorphology, Potsdam, Germany 5 2National Park Authority Vorpommern, Research and Monitoring Division, Jasmund, Germany 6

ter content can be changed by i) subsurface flow towards the cliff, ii) rain onto the cliff 23 and iii) condensation of air moisture, leading to clustered events during night time. Fail-24 ure periodicity is in alignment with the lunar cycle. Seasonal water availability, controlled 25 by plant activity, sets cliff dynamics at the annual scale. Wetter and drier than average 26 years impose a month-long legacy effect for cliff dynamics.  , and rainfall and groundwater recharge that lead to gravitational loading, reduced 51 shear strength, increased pore water pressure, and lubrication of discontinuities (Stephensen,52 2014), among others (cf. . 53 Robust attribution of cliff failure to a particular trigger depends on precise knowl- . While these studies have yielded many useful in-60 sights, we suggest that environmental seismology has the potential to give more detailed 61 understanding of links between cliff failure and its drivers.  In this article we explore the drivers and triggers of coast cliff failures on Germany's 76 largest island, Rügen. We use seismic and UAV monitoring to detect, date, locate, ver-77 ify and quantify cliff failures over a period of two years. We analyze the spatial and tem- The Jasmund peninsula on Rügen, where our study is located, comprises weakly 84 cemented Maastrichtian chalk, which has been folded and thrusted by the Scandinavian 85 ice sheet into a sequence of stacked blocks and covered by till. This sequence is exposed 86 along an 8.6 km long stretch of coast containing cliffs that are steep (57 +8 sea level data with minute resolution from the southern limit of the study area (WSV, are supposed to be recorded with a few seconds offset across the network (Fig. 2 f). To 104 identify these discrete events in the continuous stream of seismic data, we used a STA-

105
LTA picker (Allen, 1982). For details on the settings and parameter constraints see SI. 106 We screened these events with a series of automatic rejection criteria, admitting only events 107 that lasted between 1 and 180 s (assuming that shorter events are random signal coin-  along the coast, for events whose 90% confidence interval overlapped with the coast as 123 the only likely area of active mass wasting in the otherwise gently undulating landscape.

124
All detailed processing steps are described in the SI, including annotated R scripts.

125
Seismic noise cross correlation analysis can be used to infer changes in the relative We manually inspected each of the change maps in concert with the before and af-165 ter photographs to identify cliff failures. For each identified failure, we clipped the be-166 fore and after point clouds to the area of measured change and calculated the volume 167 using the 2.5D volume tool in CloudCompare. We calculated each volume three times 168 using the X, Y, and Z reference planes to determine the most appropriate reference plane 169 to use for a given failure and estimate a relative volume uncertainty of 9.7 % on aver-170 age. In addition, we measured the elevation of the center of each failure to give the height 171 above the shoreline and the distance from the cliff top.   Each synthetic data set is generated by randomly assigning event start times for the en-212 tire study period. As test for difference we use the two-sample Kolmogorov-Smirnov test. that fragmented during transport and covered the beach as a flow-like deposit (Fig. 2 d).  ceding manifestation of a potential trigger, and call this the trigger time lag (Fig. 3 a). and 8 pm (Fig. 3 b) At the monthly scale, failures occurred more frequently when the moon was far-270 ther away from the cliff (Fig. 3 c). The lunar distance ranges from 350000 to 410000 km, 271 a 14.4 % difference. Spectral analysis revealed statistically significant periodicity modes 272 between 25 and 29 days for lunar distance, precipitation and cliff failures (Fig. 3 d). The year end 2017/18 (Fig. 3 c, cluster c3 in Fig. 1). That episode, with a total of 12 sub-275 sequent failures, seven of them at nearly the same location, was associated with persis-276 tent precipitation (31 mm in 7 days, compared to a 30 year monthly average of 46 mm).

277
Detected failure occurrence was highly seasonal (Fig. 1 b) with events predominantly 278 happening in winter. In contrast, precipitation was stronger in summer than in winter 279 (331 mm versus 250 mm). This trend is reflected in the seismic velocity data (Fig. 1 e) 280 with high dv/v values during summer decreasing with the onset of autumn. However, 281 the pattern was decoupled from the evolution of the groundwater level ( Fig. 1 d).  During the entire survey period, recorded activity was focused in the central cliff 312 section, between stations "Beloved Peregrine" and "Shrapnel City", with only 7 out of 313 81 outside this reach (Fig. 1). This activity pattern is also expressed in the shape of the   Time lags show two clusters, at 0-3 (n = 19) and 16-20 (n = 20) hours (Fig. 3 a).

324
This suggests that rain may impact the cliff through two different mechanisms. We in- KS test results (Fig. SI 6) and a lack of plausible mechanisms for the measured time lags.

334
Wind time lags plateau between 0-10 h (Fig. 3 a) and within this are not distinct from 335 random. We do not see any plausible mechanistic interpretation of this distribution. Sea

346
Precipitation is an obvious cause of slope failure, but from our data we see another 347 aspect of water in the environment. A salient though not statistically significant feature 348 is that cliff failures occurred more frequently during the night (Fig. 3 b). Rain has a uni-349 form distribution throughout the day, so cannot explain this diurnal pattern of failures.

350
During failure event days, the relative humidity values were systematically higher than 351 during the other days in the winter and especially summer seasons (Fig. 3 b). But most 352 importantly, cliff activity followed the daily relative humidity cycle with a time lag of 353 1-2 hours. Therefore, we propose that relative humidity may contribute to cliff activ-  to a precipitation cycle synchronized with the lunar month. Our spectral data show pre-383 cipitation peaks when lunar distance is greatest and cliff failures tend to happen (Fig. 3 d).

384
Based on our data, we cannot determine the exact nature of the link between the lunar 385 cycle and cliff coast failures on Rügen. However, all mechanisms reviewed here tend to 386 force water availability on and within the cliff. There is an important seasonal effect that drives the Rügen cliff system to the level 389 of instability that is needed for cyclic variations on shorter, lunar (Fig. 3 c) and diur-390 nal (Fig. 3 a-b) time scales to have an effect on cliff failure. We attribute this seasonal 391 pattern to water uptake for respiration by the dense beech forest covering the cliff hin-392 terland. On Rügen, the vegetative season typically starts in early May and ends in October-

393
November. In this season, water uptake by trees leads to progressive drying of the sub-394 surface beyond the recharge capacity of summer rain events. During the subsequent sea-395 son of vegetation dormancy, from November to April, water uptake is limited, and rain 396 storms can recharge groundwater (Fig. SI 4). Hence, we infer that there is a strong veg-  ter 2018, after a dry summer with only 51 % of the average precipitation.

413
As future climate projections for Rügen include generally drier conditions and more 414 variable precipitation events (Frei et al., 2006;Umweltbundesamt, 2015), the chalk cliffs 415 may experience fewer failure events as the declining groundwater input fails to drive the 416 system to a state where rain and relative air humidity can trigger failures. This may re-417 sult in a decreasing sediment supply to the near-shore environment (Stephensen, 2014), 418 with off-site consequences, especially for adjacent sandy shorelines that may suffer from 419 erosion due to sediment starvation. Moreover, the coast cliffs may become increasingly 420 prone to undercutting, as the absence of a sediment apron exposes them to the direct 421 impact of incoming waves. This may eventually lead to less frequent but more catastrophic 422 failures as the entire cliff height will be mobilized. Unlike sandy beaches, cliffs are not 423 able to recover after erosive events by aggradation of new material (Stephensen, 2014).

424
Thus, there is no adjusting response mechanism in such an erosion-only system, which 425 makes estimating the consequences of climate change for cliff coasts even more impor-