Investigation about rainfall-induced shallow landslides in CYL and TWR watersheds, Taiwan

Abstract

Landslides identified from aerial photos and satellite images between 1988 and 2009 in Chenyulan (CYL) and Tsengwen Reservoir (TWR) watersheds in Taiwan were investigated. These watersheds have similar area, but different percentage of landslide area due to their own watershed characteristics and extreme triggers (earthquakes and typhoons). The Chi–Chi earthquake in 1999 increased about 3 % of landslide area in CYL, had obvious influence on rainfall-induced landslides in subsequent 5 years, while the Typhoon Morakot in 2009 increased 2.53 % of landslide area in TWR. The incremental landslide area by rainfalls, especially brought by typhoons indicates that the rainfall-induced shallow landslide depends not only on the rainfall amount but also on intensity. Two quadratic equations of the percentage of incremental landslide areas in terms of the rainfall erosivity factor that is composed of rainfall amount and intensity were developed. The number and size of landslides caused Typhoon Morakot in CYL and TWR were different, but they followed a similar power-law frequency–area distribution. Extreme triggers play the most important roles in the evolution of landslides in these watersheds.

Appendix 1: Rainfall erosivity factor

Traditional rainfall erosivity factor R is evaluated using long period of rainfall data with 15-min or smaller resolution (Wischmeir and Smith 1965, 1978). However, in many cases, the hourly rainfall data are more available and convenient, compared with rainfall data with 15-min or smaller resolution. The hourly rainfall data are used to evaluate the rainfall erosivity factor as well as other rainfall parameters in this study. The rainfall erosivity factor evaluated by using hourly rainfall data is denoted as R _hr for distinguishing R that is evaluated from the rainfall data with 15-min or smaller resolution. Therefore, for a rainfall event, R _hr is the product of rainfall kinetic energy and maximum hourly rainfall intensity:

$$R_{hr} = E \times I_{hr}$$

(9)

where E is the kinetic energy of a rainfall event (MJ/ha) and I _hr is the maximum hourly rainfall intensity (mm/h). In practice, the kinetic energy is usually expressed in a discrete form as:

$$E = \sum\limits_{j}^{n} {e_{j} P_{j} }$$

(10)

where n is the number of segments in an event-based hyetograph of rainfall, e _j is the rainfall energy per unit rainfall for the jth segment, and P _j is the rainfall amount for the jth segment. The rainfall energy per unit rainfall e _j depends on the size and terminal velocity of the raindrops, both of which are related to rainfall intensity. Foster et al. (1981) proposed a function for rainfall intensity as:

$$e_{j} = \left\{ \begin{array}{ll} 0.119 + 0.0873\log I_{j} & {\text{ if }}I_{j} \le 76{\text{ mm/h}} \hfill \\ 0.283 & {\text{ if }}I_{j} > 76{\text{ mm/h }} \hfill \\ \end{array} \right.$$

(11)

where I _j is the rainfall intensity (mm/h) calculated using the following formula:

$$I_{j} = \frac{{P_{j} }}{{\Delta t_{j} }}$$

(12)

where Δt _j is the time interval of a segment of the rainfall event hyetograph (mm). Δt _j  = 1 h for hourly rainfall data in this study. The total rainfall erosivity factor R _hr for a given period having M rainfall events is evaluated by the linear summation approach, i.e.,

$$R_{hr} = \sum\limits_{j = 1}^{M} {(R_{hr} )_{j} }$$

(13)

According to Yang et al. (2010), the annual rainfall erosivity factor R _hr evaluated from hourly rainfall data is about 2/3 of the annual rainfall erosivity factor R evaluated from rainfall data of 10 min resolution.