A systematic review on rainfall thresholds for landslides occurrence

Abstract

The study of rainfall thresholds is vital in understanding the factors that trigger landslides, being one of the criteria applied to landslide early warning systems that aim to mitigate their consequences. These thresholds enable the prediction of landslide occurrences as a function of rainfall measurements. This work presents an overview of the parameters involved in defining rainfall thresholds based on scientific articles published between 2008 and 2021 that discuss the subject through statistical or physical methods. These articles provided data such as publication information, threshold types, details on the data used in the works, methodology, and application of the threshold in early warning systems. There was a significant increase in research papers on this theme during this period, possibly due to the strategies advocated by the Sendai Framework. However, some regions of the world severely affected by landslides are barely mentioned in these studies. The results indicate specific trends, such as those found in the methods used to define rainfall thresholds and the parameters relating to the database when a statistical approach was used. Certain deficiencies were found, such as those concerning geological-geotechnical conditions for categorizing thresholds, the time scales of rainfall data, rain gauge density, and the criteria to define the accumulated rainfall period to be considered.

Keywords: Accumulated rainfall; Landslides; Rainfall thresholds; Statistical thresholds.

Fig. 1

Flowchart presenting the development of the research.

Fig. 2

Word cluster of works from the Scopus database that contained the set of words “landslide”, “rainfall” and “threshold” in the title, keywords, or abstract from 2008 to 2021. The size of the word letter indicates the frequency the word is quoted, whereas the colors represent the type of word cluster group. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

Distribution of articles and landslide disasters per year between 2009 and 2021.

Fig. 4

Distribution of articles by country of origin of the first author.

Fig. 5

Distribution of articles by continent of the first author (2008–2021).

Fig. 6

distribution of articles according to the journals most used.

Fig. 7

Distribution of articles by territorial coverage of thresholds (2008–2021).

Fig. 8

Distribution of articles by country in the study area. Group 1 contains the countries in which the studied area is located in only one article: Hong Kong [112], Ireland [113], New Zealand [29], Saudi Arabia [114], Serbia [115], Sri Lanka [116], Rwanda [117], East African Rift, Russia [118], Kyrgyzstan and Tajikistan [119], Turkey [74], Iran [120], Papua New Guinea [121], Thailand [122], Jamaica [123], El Salvador [12], and Czech Republic [124]; Group 2 contains the countries in which the studied area is located in two articles: France [125], Honduras [126], Portugal [127], Mexico [128], Philippines [129], Greece [130,131], and Central America [132]; Group 3 contains the countries in which the studied area is located in three articles: Indonesia [133,134], Bangladesh [135,136], Spain [137,138] and Switzerland [139,140], and Group 4 the countries in which the studied area is located in four articles: Vietnam [141,142] and Bhutan [143,144].

Fig. 9

Distribution of articles in the study area by continent (2008–2021).

Fig. 10

Quantity of articles in which the study area belongs in each country territory (2008–2021).

Fig. 11

Distribution of articles by the intention of the threshold proposed by the work to be applied to an early warning system (2008–2021).

Fig. 12

Distribution of articles by methods used to define rainfall thresholds (2008–2021).

Fig. 13

Distribution of articles by additional factors for the formation of statistical rainfall thresholds.

Fig. 14

Distribution of articles according to rainfall parameters used in the analysis by the statistical method (E-event rain, D-duration of the rain, I-intensity of rain, IE-intensity and event rain, AR-accumulated rainfall, ED-event and duration, ID-intensity and duration).

Fig. 15

Distribution of articles by statistical methods used to define rainfall thresholds.

Fig. 16

Distribution of articles by ranges of observation periods.

Fig. 17

Distribution of articles by range of quantity of landslide events.

Fig. 18

Distribution of articles by frequency of rainfall measurement.

Fig. 19

Distribution of articles by representativeness of rain gauges per area (A = number of rain gauges/100 km²).

Fig. 20

Distribution of articles according to procedures used to determine landslide occurrence time.

Fig. 21

Distribution of articles by criteria for determining landslide time.

Fig. 22

Distribution of articles by definition of threshold according to landslide classification.

Fig. 23

Distribution of articles by categorization of threshold according to physical characteristics of the locality.

Fig. 24

Distribution of articles by the number of periods adopted in the definition of rainfall thresholds (2008–2021).

Fig. 25

Types of periods of rainfall in the accumulated rainfall threshold of (a) one period and (b) two periods.

Fig. 26

Distribution of articles by range of accumulated rainfall period, in thresholds based on only one period.

Fig. 27

Distribution of articles by range of accumulated rainfall period, in thresholds based on two periods: (a) first period and (b) second period.