1
ESA Alcantara No. 5: InSAR and Landslides in Peru
Contribution of Earth Observation to
landslide early warning
White Paper
University of Zurich / Gamma Remote Sensing / National Water Authority, Peru / Czech Academy of
Sciences
Alcantara Study Reference No.: 15/P25
Study Type: Pilot Study
Contract Number: 4000117655/16/F/MOS
This contract was carried out within ESA’s General Studies Programme and funded by the European
Space Agency. The view expressed herein can in no way be taken to reflect the official opinion of the
European Space Agency
2
Authors:
Christian Huggel1, Holger Frey1, Jan Klimeš2, Tazio Strozzi3, Rafael Caduff3, Alejo Cochachin4
1Department of Geography, University of Zurich, Winterthurerstrasse 190,
8057 Zurich, Switzerland 2Institute of Rock Structure and Mechanics, Academy of Sciences, V Holešovičkách 41,
Prague 8 182 09, Czech Republic 3Gamma Remote Sensing, Worbstrasse 225, 3073 Gümligen, Switzerland 4Unidad de Glaciologia y Recursos Hidricos (UGRH), National Water Authority (ANA), Av. Confraternidad
Internacional Oeste 167, Independencia – Huaraz, Peru
3
Table of Contents
0 Scope and context of this document .......................................................................................................... 4
1 Early warning systems ................................................................................................................................ 5
1.1 International EWS standards ............................................................................................................... 5
1.2 Landslide monitoring and landslide EWS – state of knowledge ......................................................... 5
2 Available and potentially feasible EO systems ......................................................................................... 10
3 Understanding of systems and risks ......................................................................................................... 14
3.1 Landslide hazard analysis .................................................................................................................. 14
3.2 Exposure and vulnerability analysis .................................................................................................. 17
3.3 Past events and impacts .................................................................................................................... 18
3.4 Generation of landslide hazard and risk maps .................................................................................. 20
4 Monitoring and warning service .............................................................................................................. 22
4.1 Landslide precursor processes and parameters ................................................................................ 22
4.2 Landslide trigger and flow processes and parameters ...................................................................... 24
5 Landslide EWS design and operation ....................................................................................................... 25
5.1 Local landslide EWS ........................................................................................................................... 25
5.2 Regional landslide EWS ..................................................................................................................... 28
6 Perspectives and recommendations ........................................................................................................ 33
References ................................................................................................................................................... 34
4
0 Scope and context of this document
This White Paper has been developed within the framework of the Alcantara initiative of the European
Space Agency (ESA) as part of the project ‘Integrating EO information for cascading landslide and flood
hazard monitoring and early warning in the Cordillera Blanca, Peru’ (Contract No. 4000117655/16/F/
MOS). This Alcantara project was preceded by S:GLA:MO (Service for Glacial Lake Monitoring, 2014-
2016), a service for the assessment of GLOF-related hazards based on EO derived products, in-situ data
and information, and modelling results, developed in the framework of an ESA project (ECO-PROJ-EOPS-
MM-16-0033). The service was developed in close collaboration with local users and investigated critical
glacial lakes in Greenland, Tajikistan, Switzerland, and Peru. The analyses of the InSAR products for the
Peruvian cases revealed a series of interesting potential applications in the region, such as different
landslide processes at lower elevation bands (cf. Frey et al. 2016). This project demonstrated the basic
potential of EO information, in particular InSAR products, for the detection and analysis of different
types of terrain motions in support of first-order GLOF hazard assessments and as input information for
mass movement modeling for the generation of GLOF hazard maps. These achievements of the
predecessor project have been applied and put in value in the present project.
The aim of the present Alcantara project is to develop and test the potential of Earth Observation (EO) –
mainly InSAR – in landslide monitoring, landslide and flood hazard mapping, and Early Warning Systems
(EWS). First, the potential of InSAR for the detection and interpretation of different types of terrain
motions was evaluated by comparing InSAR data products to landslide inventories based on optical EO
information and geomorphological mapping. Then, InSAR derived slope deformations were compared to
geotechnical model results and ground-based survey data. For a high mountain test site, different
aspects of a landslide monitoring system have been evaluated and tested, combining EO and in-situ
information, including advanced Structure from Motion (SfM) analyses. Information from InSAR analysis,
geotechnical modeling, and on-site investigations have been used for supporting numerical mass flow
modelling in view of GLOF hazard assessments and mapping. Finally, an evaluation of the potential of the
integration of EO information in landslide EWS is a major goal of this projects.
Over the course of the S:GLA:MO and Alcantara project activities and other studies and long-term
expertise of the project consortium it has become evident that (i) much research and efforts have been
dedicated to developing EO based methods to monitor landslide processes, and (ii) a very limited
amount of experience and expertise exists with respect to integration of EO related information into
operational landslide early warning systems (EWS) although many studies emphasize the potential for it.
In line with this, virtually no official documents nor research papers exist to guide the integration of EO
information into operational landslide EWS. Such guiding documents would, however, be extremely
useful and important to set EO information into specific value for landslide EWS, in particular also in view
of the international policy agenda, e.g. the Sendai Framework for Disaster Risk Reduction (UN, 2015) and
the high priority of EWS in these strategies and policies.
Therefore, within the framework of the Alcantara initiative, we address this major gap and undertake an
effort to develop a first White Paper that lays out basic aspects of EO based information for landslide
5
EWS. By clarifying terms and concepts, and structuring techniques, methods and processes it provides a
first basis to guide future initiatives towards designing landslide EWS that implement EO information. To
allow for a relatively wide application the document has a rather generic character but repeatedly makes
reference to specific cases for illustration.
1 Early warning systems
1.1 International EWS standards
Early warning systems have become an increasingly important tool for disaster risk reduction. They have
been developed for a number of different hazards such as tsunamis, volcanoes, floods, avalanches, or
landslides.
The United Nations International Strategy for Disaster Reduction (ISDR) has defined international EWS
standards (UN 2006). Accordingly four different components of an EWS are distinguished:
1) Understanding of system and risks
2) Monitoring and warning service
3) Communication
4) Response
An appropriate EWS needs to include all four components, and is therefore a highly complex system.
It may be useful to further distinguish technical, institutional and social aspects of an EWS. EO primarily
contributes to technical aspects, however, EO may contribute to all four components of an EWS even
though its greatest potential is for components 1) and 2).
We do not distinguish here between warning, alarm and alert systems as it is done in some other studies
(e.g. Sättele et al. 2016, Stähli et al. 2015).
1.2 Landslide monitoring and landslide EWS – state of knowledge
Landslide terminology
The term landslide encompasses a rather broad range of slope deformation and mass movement
processes of the Earth’s surface. Many different disciplines are involved in landslide research and
practice and therefore the definition of landslide can vary quite substantially. A landslide can be defined
as a downslope movement of rock, soil or both, along a rupture which can have different characteristics
(e.g. rotational or translational slides) (Highland and Brobowsky, 2008). Several movement types can be
distinguished: fall, topple, slide, spread or flow. Landslide processes may include more than one type of
movement, e.g. slides that evolve into flows. Classical landslide type classifications can be found in
Varnes (1978), Cruden and Varnes (1996) and more recent updates such as from Hungr et al. (2014).
Here we use the term landslide in a broad sense that encompasses various types of slope deformation
and mass flow processes.
6
Landslide monitoring
There exists an extensive literature on landslide monitoring methods and systems, and the scientific and
practitioner communities are generally well aware of these. We broadly categorize them here as follows:
Topographic and geodetic methods
Techniques focusing on monitoring surface and subsurface characteristics
Methods monitoring meteorological parameters.
In addition there are specific methods for detecting landslide flow (slow or fast flowing) such as
geophones, seismometers, radar, video cameras or other sensors detecting landslide flow (see also Table
1).
Several key aspects need to be considered and defined for landslide monitoring:
Scale: landslide monitoring ranges from local to regional scale, i.e. from single landslides to
multiple landslides. It may be point based or may integrate information of larger areas.
Surface/sub-surface: monitoring techniques can be dedicated to measuring parameters at the
Earth surface (e.g. slope deformation at the surface, precipitations, vegetation effects on water
infiltration) or at the sub-surface (e.g. slope deformation at depth, water saturation).
Measured parameters: the range of measured parameters is wide and includes surface and sub-
surface deformation, properties of the material (such as rock, soil, ice), meteorological
parameters (rainfall, temperature, etc), hydrological or hydrogeological parameters.
Table 1 provides an overview of some landslide relevant parameters and corresponding techniques to
measure them.
7
Table 1: Landslide relevant parameters observed in current landslide monitoring and early warning
systems, and corresponding technologies applied (adapted from Stähli et al., 2015).
Observed parameter Technology
Precipitation Sum, intensity Rain gauge
Precipitation radar
Snow cover Depth
Wetness
Soil moisture Water content TDR
Water suction / pressure Tensiometer
Groundwater table Piezometer
Rock /Soil surface Precursor of failure Acoustic sensors
Displacement Trigger line
Extensometer, total stations
Inclinometer
Ground-based radar interferometry
Satellite-based radar interferometry
Triggered mass movement Vibration Geophone
Seismometer
Flow surface height Radar
Flow characteristics Video
New technologies and model integration
Terrestrial laser scanning (TLS) has seen an impressive development over the past years and has found
widespread application in landslide monitoring. The potential and limitations of TLS has been reviewed
and assessed in a number of scientific studies (Jaboyedeff et al. 2010). The high precision and repeat
monitoring capabilities represent important advantages while geometry, range and access can be
limitations depending on the landslide to be monitored.
The use of unmanned aerieal vehicles (UAV), or drones, has seen an enormous rise over the past few
years in geoscience and other disciplines. The number of applications to landslide monitoring are still
somewhat limited but on the rise as well. Typically, a Structure for Motion (SfM) procedure is applied to
generate digital terrain models. Single overflights can provide important high-resolution details on
landslide surface characteristics. Repeat overflights and DEMs are able to provide information on
landslide motion (Tanteri et al. 2017). First experiences reported in the literature are largely positive,
indicating that UAV may be a more cost-effective tool than TSL to regularly monitor landslide areas
(Rossi et al. 2016), but still involves a substantial amount of human resources and field logistics. The
combination of different techniques and data is a promising approach that deserves further
development.
8
In addition to satellite based SAR, ground-based SAR has also been increasingly applied for landslide
monitoring (Montserrat et al. 2014; Caduff et al. 2015). A fair number of ground based SAR systems are
available on the market and are regularly applied in landslide monitoring as an effective and high
precision (up to sub-millimeter) technique to detect slope deformation rates from safe distance.
Valuable experiences are also available from combining satellite and ground based SAR systems (Strozzi
et al. 2014, Frodella et al. 2016). Ground based SAR is potentially feasible and effective for landslide
EWS, in particular for progressive slope deformation processes eventually leading to failure (Atzeni et al.
2015). Experiences in operational EWS are nevertheless still limited or poorly documented.
The integration of slope stability, hydrological or combined types of models into landslide monitoring
and EWS activities is an important development (Thiebes and Glade 2016). Actually, only in very few
cases numerical models have been integrated into operational landslide EWS, mostly due to the
challenge and limited experience to run such models in an operational mode, limited confidence in
model results or limited utility of model results for EWS tasks. An example is the integration of a
deterministic hydrological and slope stability model (Combined Hydrology and Slope Stability Model,
CHASM) in a landslide EWS in southwestern Germany (Thiebes 2012). Another example is the La Saxe
landslide in Courmayeur in the western Alps of Italy where a rockslide of ca. 8 million m3 is unstable.
Monitored displacement rates are directly integrated into an inverse velocity model allowing for
estimating timing of potential final slope failure (Manconi and Giordan 2015). It resulted successful in
predicting partial slope failures about 10 hours before failure.
Landslide EWS
Landslide EWS have been reported from several locations around the world, especially in the western
US, central Europe and south-east Asia (Table 1, Figure 1). Probably the first operational landslide EWS
was developed and deployed in the San Francisco Bay Region as early as the 1980’s, however it has been
discontinued later. If the reported EWS are analyzed in more detail it can be recognized that many of
these systems do not fully integrate all four components of an EWS (as they were specified above), or
are not fully operational, or have been operational only for a limited amount of time (e.g. Bulmer and
Farquhar 2010). Maintaining an operational landslide EWS over extended periods of time is a challenging
task. Especially challenging are the (quasi-) real-time monitoring capabilities that are typically required
for operational EWS (Casagli et al. 2010). Experiences that incorporate EO information in operational
landslide EWS are particularly rare to date.
9
Table 2: Experiences of worldwide landslide EWS (from Tiebes and Glade, 2016).
10
Figure 1: Map of selected landslide EWS sites worldwide, as reported in the literature (from Stähli et al.,
2015). Note that this list is not comprehensive nor complete, see e.g. Fathani et al. 2016 or Yin et al. 2010
for additional sites of landslide EWS.
2 Available and potentially feasible EO systems
We make a basic distinction between ground-, air- and space based systems. Here we focus mainly on
space-based remote sensing systems.
We further distinguish between optical and radar systems.
The following characteristics of EO systems are relevant for landslide EWS:
Spatial resolution
Temporal resolution, i.e. revisiting time
Spectral bands/wavelength
Spatial coverage / swath
Method and time of data processing
11
A recent publication (Casagli et al. 2017) compiled a list of EO system for potential use for landslide EWS,
as given in Figure 2.
Figure 2: Available EO system and their characteristics (from Casagli et al., 2017).
In the following we analyse potentially feasible EO systems and sensors for landslide EWS. This list is not
intended to be complete nor comprehensive but should provide some guidance and overview.
12
Table 3: ESA EO systems feasible for landslide monitoring and EWS.
System Spatial resol. Revisiting
frequency
Wavelengths Spatial
coverage
(swath)
Operational period Observed landslide parameter
ERS-1/2 ~20m 35 days 2.8 cm 100 km 1991-2011 Slope/surface deformation
Vegetation
Land-cover/surface changes
Landslide occurrence/impacts
Envisat ~20m 35 days 2.8 cm 100 km 2002-2012
Sentinel-1 ~20m 12 days 2.8 cm 250 km 2014-
Sentinel-2 MSI 10-20m 6 days 0.46-2.3 µm 290 km 2016-
13
Table 4: ESA Third-Party Missions with capacities for landslide monitoring and EWS.
System Spatial resol. Revisiting
frequency
Wavelengths Spatial
coverage
(swath)
Operational period Observed landslide parameter
JERS-1 ~20m 44 days 11.8 cm 70 km 1992-1998
Slope/surface deformation
Vegetation
Land-cover/surface changes
Landslide occurrence/impacts
ALOS-1 PALSAR-
1
~10m 46 days 11.8 cm 70 km 2007-2011
ALOS-2 PALSAR-
2
~10m 14 days 11.8 cm 70 km 2014-
Radarsat-1 ~20m 24 days 2.8 cm 100 km 1995-2013
Radarsat-2 ~5m 24 days 2.8 cm 100 km 2008-
SPOT4-7 1.5-20m 3 days 0.51-1.75
µm
60 km 1998-
Pleiades 0.5-2m 1 day 0.48-0.95
µm
20 km 2011-
IKONOS 2 0.8/3.3m 3-4 days 0.45-0.93
µm
11 km 1999-2015
QuickBird 0.66/244m 1-4 days 0.45-0.90
µm
16.5 km 2000-2015
RapidEye 6.5m 1 day 0.44-0.85
µm
77 km 2008-
WorldView-
1/2/3
0.5m 2-3 days 16.4 km 2007-
Terra SAR-X ~3m 11 days 2.8 cm 10-100 km 2008-
TanDEM-X ~3m 11 days 2.8 cm 10-100 km 2010-
Cosmo-SkyMed
1/2/3/4
~3m 1 to 16 days 2.8 cm 10-50 km 2008-
14
3 Understanding of systems and risks
Developing an understanding of a landslide system is essential for any warning activity. In general, the
better a landslide system is understood the better an EWS can be designed and operated. An
appropriate understanding typically implies the analysis of observations of the past, and the EO record
can represent an essential source for this purpose. Yet, any past events and processes need to be
analysed carefully because conditioning and trigger characteristics may change over time, in particular in
highly dynamic high-mountain regions.
There are several technical aspects that are relevant and to which EO can contribute (see below sub-
sections).
A landslide EWS not only requires understanding of the landslide system but also of the associated risks.
Risk is thereby defined as a function of hazard, exposure and vulnerability (e.g. IPCC 2014). Hazard is
typically a function of the magnitude and probability of occurrence of a landslide. Exposure involves any
type of assets, from people to structures or non-economic material. Vulnerability refers to different
dimensions of the assets of objects at risk, such as their physical resistance against landslide processes,
or their degree of preparedness, capacity to recover etc. for people. EO typically contributes to
understanding of the hazard component of risk but may also be highly valuable to assess exposure.
3.1 Landslide hazard analysis
Landslide inventories
A primary purpose of landslide inventories is the identification of landslide occurrence in space
and time. Ideally, the inventory also provides information on landslide type and processes.
EO has been extensively used for, and contributed to compiling landslide inventories. Both
optical and radar technologies are relevant.
15
Figure 3: An InSAR-based slope-instability inventory on the eastern slope of the Santa River near Carhuaz,
Peru (for location see insert in a). a) ALOS-1 PALSAR-1 PSI for the time period 2007-2011, b) ALOS-1
PALSAR-1 DinSAR of the time period 2007.07.12-2007.08.27, c) ALOS-1 PALSAR-1 DinSAR of the time
period 2009.07.17-2009.09.01, d) ALOS-2 PALSAR-2 DinSAR of the time period 2016.02.21-2016.10.02.
From Stozzi et al., (in prep.).
16
Identification of stable slopes
The identification of stable slopes is not necessarily a typical task for a landslide EWS but it is
considered here important because it allows the responsible authorities to identify reasonably
safe locations.
Figure 4: Persistent Scatter Interferometry (PSI) in the surrounding of Lake 513 (red arrow), Cordillera
Blanca, Peru, based on 19 ALOS PALSAR scenes from January 2007 until March 2011. No terrain
deformations are observed in the surrounding of the lake outside the glacierized area. Courtesy of the
S:GLA:MO Project (ESRIN/RFQ/3-13731/12/I-BG).
Surface deformation and landslide process understanding
The magnitude and rate of surface deformation is an important parameter for characterization
and understanding of landslide types and processes, and hazard. Information on slope
deformation in 2.5 to 3 dimensions can be particularly useful to identify and understand
processes and extent of landslides.
Repeat EO data analysis, both optical and radar, can be applied to this purpose.
This information provides an important link to causal factors of landslide activity, i.e. EO data
crossing with hydro-meteorological data, other possibly anthropogenic factors (e.g. undercutting
of slopes).
Glacier outlines RGI
17
This type of information can also be used to support landslide dynamic models (including mass
flow models). EO derived digital elevation data (DEM) can be of important value for numerical
models that help assessing the landslide hazard, including also timing. As such they can be
integrated into EWS frameworks as specified above.
Figure 5: InSAR slope deformation profiles and glacier ice height changes at the Moosfluh slope above the
terminus of Aletsch Glacier (Switzerland). Annual displacements based on satellite InSAR ERS1/2, JERS,
Envisat, and TerraSAR-X (TSX) for the period 1992 to 2012. Glacier height change for the years 1926,
1957, 1980, 1999, 2005, 2011, and 2015 from digital elevation data. Rock slope failure events at the
landslide toe are shown on top, with respect to the number of events and their estimated volume (from
Kos et al. 2016).
3.2 Exposure and vulnerability analysis
The exposure of people and assets is a key element to determine existing landslide risks.
In rapidly developing and changing areas (e.g. urban areas in developing countries) EO can be
the only means to achieve an up-to-date assessment of exposed assets.
Optical satellite imagery, in particular high-resolution data, is particularly well suited for this
purpose.
18
The first-order analysis of physical vulnerability of structures based on high-resolution EO data.
Figure 6: Tsunami exposure map of the Khuek Kak Tambon in the in the Phang-Nga province, Thailand,
based on IKONOS imagery. Römer et al. (2012).
3.3 Past events and impacts
Past landslide events and related impacts represent important evidence to advance the
corresponding understanding.
Numerous landslide studies have made use of EO data to analyse past events and impacts of
landslides. EO has become an invaluable source of information both for research and risk
management practice.
Both radar and optical EO can be very useful to analyse past events and impacts, depending on
the characteristics of the site, weather conditions and purpose of the study.
An important recent case study where past event analysis and early warning has been combined
is described in Figure 7.
19
Figure 7: In 2016 two exceptional landslide events occurred in Tibet. Two glaciers collapsed in the Aru Co
region, one in July 2016, the other one in September 2016. The glacier collapses produced enormous
avalanches on the order of 60-80 million m3 that killed 9 herders and hundreds of animals. These events
have then been studied in detail by various types of EO data, including Sentinel-1 and Sentinel-2 data by a
international team of experts under the umbrella of GAPHAZ (the Glacier and Permafros t Hazards in
Mountains), a Scientific Standing Group of the International Association of Cryospheric Sciences (IACS)
and the International Permafrost Association (IPA). The four figure panels show the situation before, after
the first and after the second avalanche, plus a simulation of an avalanche model. While analyzing the
first glacier collapse and avalanche the international team recognized similar patterns of instability at the
neighbouring glacier and alerted them about a possible upcoming second glacier collapse. In an ad-hoc
coordinated way, facilitated through GAPHAZ, the information was used to alert Chinese colleagues, who
then informed the local government. The fact that remote observations based on a large number of very
different satellite data could be carried out with a delay of only few hours or a day between data
acquisition from space and analyses on the ground, involving scientists and satellite teams from several
nations, in fact is substantial progress for early warning capabilities related to natural hazards in remote
regions (modified from GAPHAZ 2016).
20
3.4 Generation of landslide hazard and risk maps
The identification and selection of EWS sites should rely on knowledge of hazard and risks.
Accordingly, the first EWS component consists of the understanding of hazard and risks, and
ideally should produce hazard and risks maps. However, this is not yet common standard in
many regions of the world.
Generation of hazard and risk maps typically relies on numerical landslide (mass flow) models, in
combination with field work and EO data.
The above aspects all contribute to hazard and risk maps. Depending on the national guideline (if
existing) a hazard map may imply the definition of different hazard scenarios and EO may
contribute to define these scenarios which are then typically modeled to assess the impacts of a
landslide.
As a next step, the EWS component on understanding the system and risks could be used to
update hazard and risk maps if already existing.
Figure 8: Geomorphodynamic analysis of glacier changes and dynamics in the south face of Mount
Hualcán, Cordillera Blanca, Peru, based on a high-resolution Pléiades (optical) scene. This analysis was
used for the definition of ice avalanche scenarios and subsequent modeling of the process chain involved
in outburst scenarios of Lake 513, located below this face (cf. lower left corner of the figure). Schaub et al.
(2016).
21
Figure 9: Flow heights of the ice avalanche and glacier lake outburst flow (GLOF) of the 2010 outburst of
Lake 513, Peru, as modelled with the avalanche model RAMMS. For such modelling, the quality of the
DEM is fundamental, in this case a DEM of 8m spatial resolution (shown in the background) was derived
from 2012 (optical) WorldView imagery. Based on such model results, the final hazard map for the
catchment and the city of Carhuaz (located at the left margin of the figure) was produced (cf. Schneider
et al. 2014).
Figure 10: Hazard map for the city of Carhuaz, Peru, visualized over a high-resolution satellite image. The
map relates to hazards from ice and rock avalanches and glacier lake outburst flows (GLOF) from Lake
513 (as above) and distinguishes five levels of hazards. For more details refer to Schneider et al. (2014)
(Glaciares project / University of Zurich).
22
4 Monitoring and warning service
The EO contribution to the central EWS component of monitoring and warning service depends on the
landslide type and process. In terms of deformation and movement speed the following main landslide
types are distinguished for the purpose of defining EO contributions:
Slow-moving landslides (mm to cm per year). These landslide types often imply deep-seated
slope deformation (meters to decameters deep), but also shallow landslide types can be of slow-
moving nature (e.g. in bedrock or soil, including in frozen state (permafrost), or glacier ice). The
landslide can originate in different types of material, including bedrock, soil, or any type of
sedimentary material. In high mountains landslides involving glacier ice are common but it needs
to be recognized that glaciers per nature are moving according to the material properties of ice
and surface inclination. Glacier speed can vary across orders of magnitude but typically their
range is in the order of 100 to 102 m/year. In sedimentary and bedrock material causes for
landslide motion are often related to geologic structure, variations in water content, or
topography changes such as when glacier recede and stress fields change within the affected
landslide mass, erosional processes, or seismic shaking. EO is ideally suited to monitor slow-
moving landslides.
Fast-moving landslides and mass flows (meters per second). Such landslides include a range of
different processes with different terminology, including debris flows, debris, rock or ice
avalanches, etc. Fast-moving landslides typically involve varying degrees of solid and liquid
contents. The liquid content and the material properties of the solid content (lithology, grain size
distribution) determine the flow behavior. Due to the rapid nature of this type of landslides
there is limited possibility for EO to detect the landslide flow. However, such processes are
typically preceded by slow-moving deformation and/or precursory events such as smaller rock
fall processes. EO may thus well support the early recognition of these processes.
Combinations of slow and fast-moving landslides. It is not uncommon that combinations of
slow and fast-moving landslides are observed, e.g. slow-moving landslides that at some point
involve complete failure and disintegration of the landslide mass, thus resulting in rock or debris
avalanches.
4.1 Landslide precursor processes and parameters
Monitoring of landslide precursor processes and parameters provides an important opportunity for EO,
essentially because the relevant time periods involved (weeks to months or years) correspond to typical
revisit frequency of most EO systems. We distinguish here the following main processes and parameters:
Slope deformation rate
The monitoring and measurement of slope deformation over time is one of the central elements of a
landslide EWS. In fact, a transient history of landslide and slope deformation processes is especially
useful and important to better understand the dynamics of the landslide beyond categories of slow and
23
fast moving landslides (see also section 3.1). For instance, correlation of slope deformation rates with
potential driving factors such as rainfall, snow melt or mechanical redistribution of stress due to glacier
downwasting, provides important insights to forecast and alert when landslide processes are increasing
and may become hazardous.
Figure 11: Acceleration of slope deformations at the Moosfluh landslide above Aletsch Glacier,
Switzerland, as observed by InSAR from different Sensors, dGPS and Aerial Digital Photogrammetry (ADP)
(Strozzi et al. 2010).
Antecedent precipitation and soil moisture
Antecedent precipitation belongs to the important landslide trigger parameters that are often monitored
for different types of EWS. Typically information is derived from meteorological stations in the vicinity of
the site of interest (Huggel et al. 2010). In high mountain areas the density of meteorological stations is
low, and the complex topography additionally introduces challenges and uncertainties (Salzmann et al.
2014). EO based monitoring of precipitation has the potential to support pre-event analysis of
cumulative precipitation. There exist a good number of studies that analyzed data from the Tropical
Rainfall Measurement Mission (TRMM) that later was transformed into the NASA Global Precipitation
Measurement (GPM) Missions. The scale of analysis is usually regional, continental or global (Hong and
Adler 2007, Kirschbaum et al, 2010). In Java Island, Indonesia, a prototype landslide EWS has been
developed that integrates TRMM precipitation data. This prototype system is furthermore based on
landslide inventory information, weather forecasts and a physically based rainfall induced landslide
model (SLIDE) (Liao et al. 2010).
24
Soil moisture is a similarly important parameter for landslide monitoring and early warning as it
determines, or relates to several slope stability parameters. It is typically monitored using ground based
instrumentation. Soil moisture monitoring from EO is challenging, in particular in mountain terrain with
complex topography. The ESA SMOS satellite (Soil Moisture and Ocean Salinity) is a targeted instrument
for this purpose but its potential for landslide application is very limited due to the coarse spatial
resolution and additional topography related issues.
4.2 Landslide trigger and flow processes and parameters
In contrast to landslide precursor processes landslide triggers pose a much higher challenge to EO based
monitoring. An important reason are the short time scales involved. For instance, a typical landslide
triggering rainstorm may take a couple of hours and will thus go unnoticed by virtually any EO system.
Earthquakes are acting on even shorter time scales. Relevant time scales thus may range from seconds
to minutes and hours. In addition to the times scales the detection of the relevant physical processes
may equally pose important challenges to EO system, such as reasonably accurate (in time, space and
magnitude) rainfall detection.
Precipitation
Precipitation, typically rainstorms, is among the most common triggers of landslides, especially fast-
moving landslides such as debris flows. Precipitation intensity and duration are key parameters that are
monitored and used for early warning. Rainfall is commonly monitored by ground based meteorological
stations, i.e. rain gauges, and/or ground based radar instruments. Possibilities for (quasi-) real-time EO
based monitoring are limited at the current state of technology but ex-post EO based analyses of rainfall
duration and intensity (e.g. using TRMM or GPM) can be very helpful to determine critical landslide
triggering rainfall thresholds.
Earthquakes
Earthquakes, depending on magnitude and energy direction and dissipation, have triggered hundreds to
thousands of landslides in single events. The potential of EO for detecting earthquake triggers for the
purpose of early warning is not feasible. However, as for precipitation mentioned above, ex-post analysis
of earthquake impacts on landslide processes, distribution and characteristics may be very useful and
able to inform early warning research as demonstrated in several studies (e.g. Kargel et al. 2016).
Landslide flow processes
For slow-moving landslides EO is able to provide important process related information serving early
warning purposes. For fast-moving landslides involved time scales are too short but EO based analysis of
past landslides can be important to support early warning (see also section 3.3). There exists a large
25
number of studies that used EO data to analyze landslide flow parameters such as landslide flow
geometries (width, length, runout, drop height to runout ratio, etc.), landslide flow depth, landslide
speed related parameters (e.g. curve superelevations), material properties and others. Both high-
resolution optical EO data as well as SAR data can be valuable, depending on the context and objectives.
5 Landslide EWS design and operation
5.1 Local landslide EWS
Landslide EWS termed local here are typically concerned with single (or some few) slopes, with single
catchments with clearly defined sites and structures potentially affected. The following aspects should
be considered:
The design of a local landslide EWS involves the identification/monitoring of:
1. Precursor activity: permanent or repeated monitoring (slope deformation)
2. Triggers: meteo data, landslides impacting water bodies (camera, geophone,
extensometers, trigger line, level/pressure sensor)
3. Ongoing mass flows (in case of fast landslides): cable based sensors, level/pressure
sensors, camera, geophone, rip cords)
For local level EWS it is particularly important that the design and implementation is done in
concert with all actors, from responsible authorities and institutions to affected people. Cultural
and social aspects, depending on the country/region of the EWS, can be important.
All EWS components need to be designed in an integrative manner.
For the technical components initial warning thresholds can be defined based on past events,
experiences and physical/mathematical process understanding. A calibration and validation
phase of the warning thresholds, and the EWS in general, is indispensable and may take a year or
more. This needs to be clearly discussed with local authorities and affected people.
26
Figure 12: Basic concept of a design of a landslide EWS, and more specifically the development of
warning levels and thresholds. The potential contribution of EO is highlighted.
For the communication and warning component of an EWS clear structures, processes and procedures
are crucial. In practice a protocol is commonly developed which indicates the chain of responsibilities
and decisions on an escalating scale across different levels of warning. The role and responsibilities of
different bodies and institutions needs to be clearly defined and responsible people train on it. Issues of
fluctuations of personnel and institutional instability need to be considered and addressed in order to
ensure long-term sustainability and operability of the EWS. Figure 13 shows an example of a protocol for
a glacier lake outburst flood EWS in Carhuaz, Peru.
27
Figure 13: Example of a EWS protocol for a glacier lake outburst flood EWS in Carhuaz, Peru (developed
by Crealp and University of Zurich within Proyecto Glaciares).
Furthermore, it is worth mentioning that the experiences of an increasing number of EWS now have
started to result in more standardized applications. Specifically, the wide application of EWS at the
community scale in SW Asia resulted in an effort for standardization of the methodology and approach
for community-based EWS. The prepared standard contains the concept of people-centered EWS as
defined by UN-ISDR and will be developed by ISO/TC 292 Security and Resilience, with the participation
28
of 43 countries in the committee’s work and another 14 countries as observers (Fathani et al. 2017). The
standard contains seven subsystems:
1. Risk assessment;
2. Dissemination and communication;
3. Establishment of disaster preparedness and response teams;
4. Development of evacuation routes and maps;
5. Development of standard operating procedures;
6. Monitoring, early warning, and evacuation drills;
7. Commitment of the local government and community to the operation and maintenance of the whole
system.
The last point is crucial to ensure long-term sustainability and operation of the EWS as the community
understanding and acceptance will not only have direct effect on the system functionality but will affect
community response to issued warnings thus efficiency of the system.
5.2 Regional landslide EWS
Regional landslide EWS are understood as systems that cover larger areas, .e.g. multiple catchments,
provinces, i.e. operate on a national or sub-national scale (exceptionally also over several countries). The
following aspects need to be considered for landslide EWS over regional scales:
For regional landslide EWS all four components of the EWS are designed, and operated in a
different way than local level EWS. The level of detail, accuracy and precision is typically less
than for a local landslide EWS.
The first EWS component of understanding the risks may be addressed by analyzing GIS based
landslide susceptibility, combined with population and infrastructure data. EO data, both optical
and SAR can substantially contribute to this analysis by providing area-wide information on slope
deformation, landslide reconnaissance, vegetation, soil and surface parameters and exposed
assets.
The second EWS component (monitoring and warning service) can, for instance, be developed by
combining GIS service providing basic landslide susceptibility information with a network of real-
time meteorological rainfall information, potentially also soil moisture information, and possibly
weather forecasts. Additionally, so-called climate services, i.e. easily accessible weather data and
weather forecasts, seasonal forecasts, warnings of severe weather, agrometeorological
information, are currently fostered and provided by different international organizations and
have a potential for being integrated in regional landslide EWS. As for local landslide EWS
warning thresholds needs to be carefully defined and calibrated and validated during a test
phase. EO information can be an important source for establishing the thresholds and for (ex-
post) validation (e.g. with InSAR analysis). Integrating EO information into the operational
29
aspects of threshold definition and changes is challenging and depends, among other, on the
duration of the respective warning levels and the time for EO data availability and processing.
Corresponding experiences are hardly available so far. This important potential, not yet
sufficiently exploited is emphasized by the value of integrating EO information into landslide
EWS to provide additional robustness to the EWS (not only for regional, also for local EWS). EO
technology is capable of supplying necessary input data regardless of ground conditions which
may possibly destroy or temporarily disable field monitoring and management systems e.g. due
to energy shortage, damage caused by earthquakes or human interference. The fact that the EO
data may be processed elsewhere without necessity to access local or regional monitoring or
EWS networks can in some cases represent a significant advantage as well.
The third and fourth EWS components (communication and response) for a regional landslide
EWS operate more on an institutional level involving, for instance, authorities and responsible
institution (such as Civil Defense) for the respective scale of the EWS (e.g. national or provincial
scale). These institutions are then responsible to take the necessary actions on a more local level
by communicating the warning level to local disaster prevention and emergency bodies.
At the current state of technology and research existing regional landslide EWS have the
character of forecasting systems rather than fully developed EWS.
In the following we present a few examples of existing regional landslide EWS, specifically from Colombia
and Japan, both countries that are heavily affected by landslide processes and disasters (Figures 14-16).
30
Figure 14: Example of a landslide alert and forecast bulletin for Colombia, based on a regional scale type
EWS operated by the national meteorological and hydrological service (IDEAM).
31
Figure 15: Example of an online, real-time, landslide alert and forecast for Colombia, based on a regional
scale type EWS operated by the national meteorological and hydrological service (IDEAM). Colors refer to
different warning levels. See also http://www.ideam.gov.co/web/pronosticos-y-alertas/alertas.
32
Figure 16: An example of landslide EWS providing warning levels based on 60-min cumulative rainfall and
calculated soil-water index operational from 2007 (Osani et al. 2010, see also
https://www.jma.go.jp/en/doshamesh/).
33
6 Perspectives and recommendations
The potential of EO for (operational) landslide EWS is currently not sufficiently exploited.
The great majority of research focuses on monitoring aspects of EO for landslides, and very little
has so far actually been concerned with early warning issues. Similarly, there are limited
experiences yet when it comes to integration of EO into operational landslide EWS. One of the
reasons includes constraints concerning delayed availability of EO information and complex and
time-consuming processing procedures necessary to analyze the EO data for their use in the EWS
applications.
A co-design of the EWS with local people and authorities is essential for the longer term success
of the system.
Repeat periods, data availability (time lag and costs), preprocessing, and characteristics of EO
data are crucial parameters for an operational integration into EWS.
Design, implementation and calibration of the EWS needs time. In particular the calibration
phase is critical and needs to be well discussed with responsible authorities and affected people
to avoid false expectations.
Continuous maintenance of the EWS needs to be considered and planned (including in terms of
budget) from the beginning. An important advantage of EO based landslide EWS is that it is
largely independent from structures and instruments operating on the ground which are
susceptible to malfunction or damage.
In-depth training of the local operating people is an important element of success of an EWS.
Overall, in view of the high potential of EO information for landslide EWS, contrasted with the
currently limited use in operational EWS, it is highly recommended to strengthen the efforts
beyond EO based landslide monitoring to the integration in operational EWS.
Pilot studies and cases can be a feasible means to test EO information in operational EWS, gain
further real-world experience, and demonstrate the EO capabilities to a wide range of
stakeholders, authorities and institutions.
This White Paper is the first of its kind and is expected to support and guide future initiatives on
EO integration into landslide EWS.
34
References
Atzeni, C., Barla, M., Pieraccini, M., Antolini, F., 2015. Early Warning Monitoring of Natural and
Engineered Slopes with Ground-Based Synthetic-Aperture Radar. Rock Mech. Rock Eng. 48, 235–246.
doi:10.1007/s00603-014-0554-4
Bulmer, M., H., Farquha, T., 2010. Design and installation of a Prototype Geohazard Monitoring System
near Machu Picchu, Peru. Nat. Hazards Earth Syst. Sci. 10, 2031–2038. doi:10.5194/nhess-10-2031-2010
Caduff, R., Schlunegger, F., Kos, A., Wiesmann, A., 2015. A review of terrestrial radar interferometry for
measuring surface change in the geosciences. Earth Surf. Process. Landf. 40, 208–228.
doi:10.1002/esp.3656
Casagli, N., Catani, F., Ventisette, C.D., Luzi, G., 2010. Monitoring, prediction, and early warning using
ground-based radar interferometry. Landslides 7, 291–301. doi:10.1007/s10346-010-0215-y
Casagli, N., Frodella, W., Morelli, S., Tofani, V., Ciampalini, A., Intrieri, E., Raspini, F., Rossi, G., Tanteri, L.,
Lu, P., 2017. Spaceborne, UAV and ground-based remote sensing techniques for landslide mapping,
monitoring and early warning. Geoenvironmental Disasters 4, 9. doi:10.1186/s40677-017-0073-1
Cruden, D.M., Varnes, D.J., 1996. Landslide types and processes. Landslides Investig. Mitig. 36–75.
Fathani, T.F., Karnawati, D., Wilopo, W., 2016. An integrated methodology to develop a standard for
landslide early warning systems. Nat. Hazards Earth Syst. Sci. 16, 2123–2135. doi:10.5194/nhess-16-
2123-2016
Fathani, T.F., Karnawati, D., Wilopo, W., 2017. Promoting a Global Standard for Community-Based
Landslide Early Warning Systems (WCoE 2014–2017, IPL-158, IPL-165), in: SpringerLink. Presented at the
Workshop on World Landslide Forum, Springer, Cham, pp. 355–361. doi:10.1007/978-3-319-59469-9_30
Frey, H., Huggel, C., Bühler, Y., Buis, D., Burga, M.D., Choquevilca, W., Fernandez, F., Hernández, J.G.,
Giráldez, C., Loarte, E., Masias, P., Portocarrero, C., Vicuña, L., Walser, M., 2016. A robust debris-flow and
GLOF risk management strategy for a data-scarce catchment in Santa Teresa, Peru. Landslides 13, 1493–
1507. doi:10.1007/s10346-015-0669-z
Frodella, W., Ciampalini, A., Gigli, G., Lombardi, L., Raspini, F., Nocentini, M., Scardigli, C., Casagli, N.,
2016. Synergic use of satellite and ground based remote sensing methods for monitoring the San Leo
rock cliff (Northern Italy). Geomorphology 264, 80–94. doi:10.1016/j.geomorph.2016.04.008
GAPHAZ, 2016. Twin glacier collapse in Tibet puzzles scientists and triggers rapid international
collaboration. Glacier and Permafros t Hazards in Mountains (GAPHAZ) A Scientific Standing Group of the
International Association of Cryospheric Sciences (IACS) and the International Permafrost Association
(IPA).
Highland, L.M., Bobrowsky, B., 2008. The landslide handbook. A guide to understanding landslides. U.S.
Geological Survey Circular 1325, Reston, Virginia.
Hong, Y., Adler, R.F., 2007. Towards an early-warning system for global landslides triggered by rainfall
and earthquake. Int. J. Remote Sens. 28, 3713–3719.
35
Huggel, C., Khabarov, N., Obersteiner, M., Ramírez, J.M., 2010. Implementation and integrated numerical
modeling of a landslide early warning system: a pilot study in Colombia. Nat. Hazards 52, 501–518.
doi:10.1007/s11069-009-9393-0
Hungr, O., Leroueil, S., Picarelli, L., 2014. The Varnes classification of landslide types, an update.
Landslides 11, 167–194. doi:10.1007/s10346-013-0436-y
Kargel, J.S., Leonard, G.J., Shugar, D.H., Haritashya, U.K., Bevington, A., Fielding, E.J., Fujita, K.,
Geertsema, M., Miles, E.S., Steiner, J., Anderson, E., Bajracharya, S., Bawden, G.W., Breashears, D.F.,
Byers, A., Collins, B., Dhital, M.R., Donnellan, A., Evans, T.L., Geai, M.L., Glasscoe, M.T., Green, D.,
Gurung, D.R., Heijenk, R., Hilborn, A., Hudnut, K., Huyck, C., Immerzeel, W.W., Liming, J., Jibson, R., Kääb,
A., Khanal, N.R., Kirschbaum, D., Kraaijenbrink, P.D.A., Lamsal, D., Shiyin, L., Mingyang, L., McKinney, D.,
Nahirnick, N.K., Zhuotong, N., Ojha, S., Olsenholler, J., Painter, T.H., Pleasants, M., Pratima, K.C., Yuan,
Q.I., Raup, B.H., Regmi, D., Rounce, D.R., Sakai, A., Donghui, S., Shea, J.M., Shrestha, A.B., Shukla, A.,
Stumm, D., Kooij, M. van der, Voss, K., Xin, W., Weihs, B., Wolfe, D., Lizong, W., Xiaojun, Y., Yoder, M.R.,
Young, N., 2016. Geomorphic and geologic controls of geohazards induced by Nepal’s 2015 Gorkha
earthquake. Science 351, aac8353. doi:10.1126/science.aac8353
Kirschbaum, D.B., Adler, R., Hong, Y., Kumar, S., Peters-Lidard, C., Lerner-Lam, A., 2010. Advances in
landslide nowcasting: evaluation of a global and regional modeling approach. Environ. Earth Sci. 1–14.
Kos, A., Amann, F., Strozzi, T., Delaloye, R., von Ruette, J., Springman, S., 2016. Contemporary glacier
retreat triggers a rapid landslide response, Great Aletsch Glacier, Switzerland. Geophys. Res. Lett. 43,
2016GL071708. doi:10.1002/2016GL071708
Liao, Z., Hong, Y., Wang, J., Fukuoka, H., Sassa, K., Karnawati, D., Fathani, F., 2010. Prototyping an
experimental early warning system for rainfall-induced landslides in Indonesia using satellite remote
sensing and geospatial datasets. Landslides 7, 317–324.
Manconi, A., Giordan, D., 2015. Landslide early warning based on failure forecast models: the example of
the Mt. de La Saxe rockslide, northern Italy. Nat. Hazards Earth Syst. Sci. 15, 1639–1644.
Monserrat, O., Crosetto, M., Luzi, G., 2014. A review of ground-based SAR interferometry for
deformation measurement. ISPRS J. Photogramm. Remote Sens. 93, 40–48.
doi:10.1016/j.isprsjprs.2014.04.001
Römer, H., Willroth, P., Kaiser, G., Vafeidis, A.T., Ludwig, R., Sterr, H., Revilla Diez, J., 2012. Potential of
remote sensing techniques for tsunami hazard and vulnerability analysis – a case study from Phang-Nga
province, Thailand. Nat Hazards Earth Syst Sci 12, 2103–2126. doi:10.5194/nhess-12-2103-2012
Rossi, G., Nocentini, M., Lombardi, L., Vannocci, P., Tanteri, L., Dotta, G., Bicocchi, G., Scaduto, G.,
Salvatici, T., Tofani, V., others, 2016. Integration of multicopter drone measurements and ground-based
data for landslide monitoring. Landslides Eng. Slopes Exp. Theory Pract. Assoc. Geotec. Ital. Rome 1745–
1750.
Salzmann, N., Huggel, C., Rohrer, M., Stoffel, M., 2014. Data and knowledge gaps in glacier, snow and
related runoff research – A climate change adaptation perspective. J. Hydrol. 518, Part B, 225–234.
doi:10.1016/j.jhydrol.2014.05.058
Sättele, M., Bründl, M., Straub, D., 2016. Quantifying the effectiveness of early warning systems for
natural hazards. Nat. Hazards Earth Syst. Sci. 16, 149–166. doi:10.5194/nhess-16-149-2016
36
Schaub, Y., Huggel, C., Cochachin, A., 2016. Ice-avalanche scenario elaboration and uncertainty
propagation in numerical simulation of rock-/ice-avalanche-induced impact waves at Mount Hualcán and
Lake 513, Peru. Landslides 13, 1445–1459. doi:10.1007/s10346-015-0658-2
Schneider, D., Huggel, C., Cochachin, A., Guillén, S., García, J., 2014. Mapping hazards from glacier lake
outburst floods based on modelling of process cascades at Lake 513, Carhuaz, Peru. Adv Geosci 35, 145–
155. doi:10.5194/adgeo-35-145-2014
Stähli, M., Sättele, M., Huggel, C., McArdell, B.W., Lehmann, P., Van Herwijnen, A., Berne, A., Schleiss,
M., Ferrari, A., Kos, A., Or, D., Springman, S.M., 2015. Monitoring and prediction in early warning systems
for rapid mass movements. Nat Hazards Earth Syst Sci 15, 905–917. doi:10.5194/nhess-15-905-2015
Strozzi, T., Delaloye, R., Kääb, A., Ambrosi, C., Perruchoud, E., Wegmüller, U., 2010. Combined
observations of rock mass movements using satellite SAR interferometry, differential GPS, airborne
digital photogrammetry, and airborne photography interpretation. J. Geophys. Res. 115, F01014.
doi:10.1029/2009JF001311
T. Strozzi, H. Raetzo, U. Wegmüller, J. Papke, R. Caduff, C. Werner and A. Wiesmann, Satellite and
Terrestrial Radar Interferometry for the Measurement of Slope Deformation, in G. Lollino et al. (eds.),
Engineering Geology for Society and Territory – Volume 5, DOI: 10.1007/978-3-319-09048-1_32, Springer
International Publishing Switzerland 2014
Tanteri, L., Rossi, G., Tofani, V., Vannocci, P., Moretti, S., Casagli, N., 2017. Multitemporal UAV Survey for
Mass Movement Detection and Monitoring, in: Mikos, M., Tiwari, B., Yin, Y., Sassa, K. (Eds.), Advancing
Culture of Living with Landslides. Springer International Publishing, Cham, pp. 153–161.
doi:10.1007/978-3-319-53498-5_18
Thiebes, B., 2012. Landslide Analysis and Early Warning Systems: Local and Regional Case Study in the
Swabian Alb, Germany. Springer Science & Business Media.
Thiebes, B., Glade, T., 2016. Landslide early warning systems–fundamental concepts and innovative
application, in: Landslides and Engineered Slopes: Experience, Theory and Practice, Edited by: Aversa, S.,
Cascini, L., Picarelli, L., and Scavia, C., Proceedings of the 12th International Symposium on Landslides,
Napoli, Italy. pp. 12–19.
Varnes, D.J., 1978. Slope movement types and processes. Spec. Rep. 176, 11–33.
Yin, Y., Wang, H., Gao, Y., Li, X., 2010. Real-time monitoring and early warning of landslides at relocated
Wushan Town, the Three Gorges Reservoir, China. Landslides 7, 339–349. doi: 10.1007/s10346-010-
0220-1
Top Related