Three Years of UAS-Based High Resolution Topographic Surveys of a Coastal Boulder Deposit to Monitor Change and Assess Tsunami vs Storm Wave Deposition in Java, Indonesia

Abstract

Boulder deposits can be important records of tsunami activity; however, distinguishing whether they were formed by tsunami or storm wave activity can be difficult, particularly when they are located in or near the littoral zone. We used uncrewed aerial systems (UAS) and structure-from-motion photogrammetry to generate 4 sets of 1 cm digital elevation models (DEMs) and 0.5 cm orthomosaics of a boulder deposit at Watu Karung (8.241S 110.983E) on the south coast of Java, Indonesia over 3 years as tools to measure change to the deposit and assess its origin. The deposit is ~700 m2, ~2 m asl, and comprises limestone and hardpan boulders ~0.5 to 2 m in length. Except at high tide, the deposit is separated from the sea by a steep (~25-35o) hardpan slope and 0.5-1.0 m tall hardpan berm, which inhibits landward transport of boulders. The deposit was surveyed four times: twice in 2016 (2016a, 2016b), in 2017, and in 2019. Surveys 2016a and 2016b were conducted immediately before and after a significant storm swell event (~4 m) contemporaneous with a spring tide. A larger swell event (~5 m; probably a 10 - 20 year event) occurred in July 2018. UAS used were quadcopters with a 24 MP Sony A5100 camera (2016a and b), a 20 MP DJI camera (2017), or a 12 MP DJI camera (2019). All of the surveys except 2019 included ground control points (GCPs) surveyed with dGNSS. The 2016 surveys used the same GCPs to improve DEM co-registration. The 2019 survey was georeferenced using natural control points that were assigned geographic positions using the 2017 data. DEMs were differenced against each other and hillshades and orthomosaics were compared to reveal locations of boulder erosion and deposition. Between the 2016a and 2016b surveys, ~10% of boulders moved, including one 1.7 x 0.7 x 0.2 m in size that moved 5.6 m. Between 2016 and 2017, approximately 300 of 650 large boulders (1.6 m average length) moved between 10 cm and 30 m. By 2019, most boulders had moved, and at least 150 boulders moved at least ~ 2 m relative to 2016; many of these rotated about horizontal axes. However, only one large (> 1.5 m length) boulder is recognized as having been added to the deposit. Thus, although storm waves extensively modified the deposit, it likely originally was formed by a tsunami, progressive landward erosion of the shoreline and exposure of boulders, or perhaps larger swell events.

Full text

Acknowledgements
Many students and others participated in the eld excursions to Indonesia, including Serena Smith, Gilang Setiadi, Andre
Tompubolon, Bret Young, Bryn Howell, Danielle Spencer, Chad Emmett, Anne Arendt, and Alex Arendt
Funding for eld work was provided by the Utah Valley University Oce of Engaged Learning and the UVU College of Sci-
ence Scholarly Activities Program
We thank NVIDIA for support via an Educational GPU grant
Trimble Navigation supported this work through their educational program
Conclusions
- Results suggest that cli erosion and/or tsunami activity plays a role in boul-
der deposition at the site
- Many boulders rolled (rotated about ~horizontal axes), so models of boulder
transport should account for this transport process
- Storm waves strongly modied deposit and imbrication, so imbrication is
not created by tsunami
- sUAS plus SfM very eective for monitoring geomorphic change at this boul-
der deposit
- Proof of tsunami activity remains challenging
References
Abercromie, R.E., Antolik, M., Felzer, K., Ekstrom, G., 2001, The 1994 Java tsunami earthquake: Slip over a subducting seamount, Journal of Geophysical Research, v 106, n. B4, pp
6595-6607.
Meservy, W.N., 2017, Evaluating the East Java Tsunami Hazard: What Can Newly-Discovered Imbricate Coastal Boulder Accumulations Near Pacitan and at Pantai Papuma, Indonesia Tell
Us?, Unpublished M.S. Thesis, Brigham Young University, 48 pp.
Surine Forecast Team, 2018, Biggest Indo Swell Ever? Here’s the Data, https://www.surine.com/surf-news/biggest-indo-bali-swell-waves-ever-surfed/30832, accessed 1 December,
2019.
Uribe, A.T., Bunds, M.P., Andreini, J.C., Horns, D., Harris, R.A., Prasetyadi, C., Yulianto, E., Putra, P.S., 2017, Using Point Clouds Generated from Unmanned Aerial Vehicles Imagery Processed
with Structure from Motion to Address Tsunami vs Storm Wave Boulder Deposition in Watu Karung, Indonesia, Abstract submitted to the 2017 Fall Meeting of the American Geophysi-
cal Union, New Orleans, LA.
Model 2016a 2016b 2017 2019
Date 7/30-31/2016 8/1/2016 7/12/2017 7/9/2019
sUAS DJI Phantom 2 DJI Phantom 2 DJI Phantom 4 Pro DJI Mavic Pro
Camera / lens 24.3 Mp Sony / 16 mm 24.3 Mp Sony / 16 mm 20 Mp DJI / DJI 8.8 mm 12.3 Mp DJI / DJI 4.7 mm
Flight height (agl, m) 19.5 m 20.5 m 17.5 m 26.6 m
# Photos 262 480 262 263
GSD (mm) 4.4 mm 4.6 mm 4.5 mm 8.1 mm
Point cloud size 108 x 106 210 x 106 118 x 106 70 x 106
Mapped area (m2) 6,480 m2 12,000 m2 6,970 m2 11,300 m2
Point density
(pts/m
2
16,700 pts/m2 17,500 pts/m2 16,900 pts/m2 6,190 pts/m2
DEM pixel size (cm) 1 cm 1 cm 1 cm 1 cm
# GCPs 10 10 12 17
SfM, Point Cloud,
DEM Parameters
Other Large, Local Recorded
Swell Events
(from surine.com)
upper terrace
(stage 5e, ~120 ka?)
lower terrace
with boulders
modern wave
platform
0
2
4
6
0 10 20 30 40 50 60
altitude (msl EGM08, m)
distance along prole (m)
Boulder Deposit and
Shoreline Topography
Motion of hand-placed boulders during
2016 storm wave event.
- green circles = pre-event positions
- red triangles = post-event positions
Every boulder moved, most in direction
of longshore drift
Average displacement = 27.6 m
Minimum displacement = 0.22 m
Maximum displacement = 90.3 m
2016 Storm Event and Hand-Placed Boulder Motion
2016 Storm Event at West Beach: Large Boulder Transport and Horizontal Axis Rotation
Details
Date
Significant
wave height
(m)
Peak period (s)
July 2018
5.0
19
June 2015
4.0
18
Sept 2009
4.9
18
July 1989
4.8
19
June 1983
4.9
18
E. West beach vertical change (dierence map) from pre- and post-
storm wave event, 2016. Green is up, red down (+/- 0.5 m).
C. Hillshade of west beach pre-storm wave event. D. Hillshade of west beach post-storm wave event. Note ~ 5.6 m
displacement of boulder, ~ 1.7 x 0.7 x 0.2 m, ~ 500 kg.
ipped boulder
displaced boulder
displaced boulder
0 m
-0.5
0.5 m
See Background for location
Questions
1. Are near-shore boulder deposits evidence of tsunami(s), or are they primarily storm-wave deposits?
2. Can we eectively track deposition, erosion, and change of the deposit with sUAS-based SfM to detect change to the deposit
and whether storm activity can deposit boulders?
What We Did
1. Built high resolution DEMs of boulder deposit four times, separated by notable storm-driven wave events
Results
1. The deposit probably is not a result of storm-waves alone. Cli erosion and/or tsunami activity plays a role.
Storm waves moved most boulders and rotated many about horizontal axes, but added only one boulder.
Modeling of boulder transport must account for rolling; modern imbrication created in large part by storm waves.
Unusually steep face below deposit may not be representative of other boulder deposits.
2. sUAS-based SfM very eective for tracking change in the deposit
2016a
2016b
2017
2019
Hillshade
Elevation Change from 2016a
(overlain on orthophoto)
Orthophoto
wave event: ~4.2 m swell + ~ 2.5m spring tide
wave event: ~5 m swell + ~ 2.5m spring tide, approximately 10 - 20 year event
miscellaneous undocumented wave events
30 cm
0 cm
-30 cm
Central Points of the Poster
Study site: Is this imbricated boulder deposit at Wa-
takarung evidence of tsunami inundation? See
background section for location.
green = boulder in new location, orange = boulder missing from original location
DJI Phantom 4 Pro DJI Phantom 2, Sony A5100 camera
DJI Phantom 2, Sony A5100 camera
DEM Construction Methods
- Very high resolution (1 cm pixel) DEMs rasterized from point clouds made using SfM processing
(Agisoft) of photographs from sUAS
- UAS platforms: DJI Phantom 2 (customized with 24 Mp Sony A5100 camera, 16 or 20 mm lens);
DJI Phantom 4 Pro (20 Mp DJI camera); DJI Mavic Pro (12.3 Mp DJI camera)
- Point clouds were georeferenced and co-registered with articial and natural ground control
points (GCPs)
- d-GNSS measurement of GCP locatoins, except nal model, which used natural GCPs located
using prior models
- Local GNSS base (JPL APPS solutions)
- SfM processing in Agisoft Photoscan/Metashape
- DEMs dierenced in ArcMAP (shown here) and using iterative closest point (e.g., CloudCom-
pare).
Identifying, let alone dating, unequivocal evidence of past tsunami
activity is very challenging yet important to assessing hazard.
Near shore boulder deposits located along the south coasts of In-
donesian islands on the southwestern edge of the Sunda plate
may be a useful indicator of tsunami activity
Project goals include
Explore whether storm wave activity can be ruled out as the
origin of one of the boulder deposits
Develop methodology to monitor boulder deposit and geomor-
phic change in general with DEMs made with struc-
ture-from-motion (SfM) processing of photographs from
small uncrewed aerial systems (sUAS)
Background
Java and study area in context of
the Sunda Trench, recent tsunami-
genic earthquakes along the relat-
ed Sunda megathrust, and south-
east Asia. Colored areas show esti-
mated rupture areas of earth-
quakes . Basemap from Google
Earth, plate boundaries from USGS,
and earthquake rupture areas from
Caltech Tectonics Observatory and
Abercrombie et al. (2001).
Below: Java island area detail. Pan-
gandaran is site of 2006 Mw 7.8
earthquake and tsunami inunda-
tion in which ~ 700 people per-
ished. Google Earth base map.
Orthophoto of study site generated for study.
Imbricated boulder
deposit
West beach
Sunda Megathrust
52 mm/yr
67 mm/yr
2006 Mw 7.8
1994 Mw 7.9
2000 Mw 7.9
2007 Mw 8.4
2007 Mw 7.9
2005 Mw 8.6
2002 Mw 7.3
2004 Mw 9.2
Watukarung study site
Java
Sumatra
Sulawesi
Borneo
Australia
India
Malaysia
Indochina
S
u
n
d
a
M
e
g
a
t
h
r
u
s
t
The south coast of Java borders the Sunda Trench, which has produced several
tsunami-genic earthquakes in the past three decades. Both seismic and tsuna-
mi hazard are high, and the close proximity of the densely-populated coast to
potential earthquake sources creates large risk.
The study site contains a tombolo and probable 5e marine terrace. The tombolo promontory comprises
Miocene limestone. The imbricated boulder deposit contains limestone and hardpan cemented beach
sediment. However, limestone is ubiquitous in the area so the boulders in the deposit cannot be unequiv-
ocally connected to the nearby promontory.
Introduction / Background /
Methods
additional imbricated
boulder deposits
study site
additional boulder deposits
(from Meservy, 2017)
imbricated boulder deposits
likely from 1994 M7.9
(from Meservy, 2017)
Java
Lombok
Bali
Sumbawa
Jokarta
Jogjakarta
Pangandaran
Sunda Trench
200 km
Anak Krakatao
Three Years of UAS-Based High Resolution Topographic Surveys of a Coastal Boulder Deposit to Monitor Change and Assess
Tsunami vs Storm Wave Deposition in Java, Indonesia
NH43F-1007
1Department of Earth Science, Utah Valley University, Orem, Utah, USA, 2Dept. of Geological Sciences, Brigham Young University, 3School of Geology & Engineering, Universitas Pembangunan Nasional
Michael P. Bunds1 (michael.bunds@uvu.edu), Alexander T. Uribe1, Ron A. Harris2, Bryce Berrett2, Daniel Horns1, Carolus Prasetyadi3, Jeremy C. Andreini1