Basic HTML Version
Understanding the spatial and temporal occurrence of landslides using satellite and airborne technologies: Papua New Guinea
2
Rainfall-induced landslides in PNG
To understand the relationships between rainfall and landslide occurrence, records of historical landslides needed
to be collated into an inventory. Information from a range of sources, including: (1) technical and site inspection
reports obtained from the PNG Mineral Resource Authority (MRA) and the Department of Mineral Policy and
Geohazards Management archives (Itiogen, 2007; Kuna and Moihoi, 2006; Browne, 1994; Tutton and Buleka, 1993;
Peart, 1991); (2) accessible journal publications (Fookes and Dale, 1992; Griffiths et al., 2004; King et al., 1989); (3)
newspaper records; (4) internet publications; and (5) supplementary archives (i.e. Dartmouth Flood Observatory
database and United States Geological Survey (USGS) National Earthquake Information Centre’s (NEIC) Preliminary
Determination of Epicentres (PDE) database) were used to extract pertinent information (date, location, type, size/
number, possible trigger and impacts) about each landslide. Only those landslides, where both the date and location
could be established with reasonable accuracy, were entered into the database as this information is critical for
analysing the characteristics of potential trigger mechanisms. 126 landslide entries were identified between 1970
and 2009 (Fig. 1(A)). Each entry represents a unique temporal period in which a landslide or cluster of landslides
occurred and/or the unique spatial location of a slide.
Fig. 1| A
Location of landslide-triggering events which occurred between 1970 and 2010.
B
A frequency distribution of monthly landslide
triggering events compared against the regional monthly mean rainfall.
Using this inventory, regional landslide occurrence was assessed relative to the interseasonal and interannual
rainfall patterns that interact with PNG. Given the limited availability of rainfall gauge data, (freely-available) monthly
rainfall data from the Global Precipitation Climatology Centre (GPCC; Adler et al., 2003; http://kunden.dwd.de/GPCC/
Visualizer) were used to compare monthly landslide frequencies with monthly rainfall climatology data (40-year;
1970-2009; Fig. 1(B)). Monthly landslide frequencies were also compared with the monthly rainfall climatologies
associated with different phases of El Niño Southern Oscillation (ENSO; Fig. 2). This analysis confirmed two widely
held assumptions. Firstly, landslide events occur with greater frequency during the wetter season (December to
May) than during the drier season (June to November; Fig. 1(B)) and, secondly, that more landslides are observed
during years associated with La Niña episodes than years associated with either El Niño or ENSO neutral episodes
(Fig. 2). The analysis also drew out a number of key considerations for rainfall-induced landslide occurrence. Of
particular importance was the identification that, particularly for the latter half of the wet season, absolute mean
month-to-month rainfall variability was found to be small, while changes in the monthly frequency of landslides
were relatively large. This indicated that antecedent rainfall in excess of one month was likely to be important for
the initiation of larger-scale landslides, such as those collated in the inventory. Furthermore, the findings hinted
that rainfall driven by different dynamic mechanisms could result in different types of failure, at different times of
the year.