Starting from 1950, the Hong Kong government of China began to collect complete landslide disaster data. 90% of landslides in China and Hongkong are shallow landslides (less than 3m) with a scale of less than 50m3. China's geotechnical engineering office receives an average of 200 ~ 300 landslide reports every year, all of which are slope instability. After 1977, the Geotechnical Engineering Office conducted a detailed classification, stability investigation and evaluation of 54 000 slopes in Hong Kong (Wong, 200 1). At present, China Mainland and Hongkong have accumulated at least 100 monographic research reports and public papers on rainfall landslides.

Hongkong, China is the earliest area in the world to study the relationship between rainfall and landslide and to carry out meteorological forecast of rainfall and landslide. China and Hongkong use radar images to interpret small-scale geological structures to identify potential landslide areas. At present, 86 automatic rain gauges managed by the Geotechnical Engineering Office and 24 automatic rain gauges operated by the China Hong Kong Observatory transmit rainfall data to the Geotechnical Engineering Office every five minutes through advanced data acquisition and transmission systems.

The geological disasters in China and Hongkong are mainly shallow landslides. 1984, the Geotechnical Engineering Office (GEO) in Hongkong, China realized the automatic landslide early warning system (LWS), which is the first system used for landslide prediction in the world, and it is now running every day (Premchitt, 1997). Its technical support work includes: ① Rainfall collection: 5-minute rainfall forecast by meteorological department, real-time rainfall by automatic rain gauges (86 from Civil Engineering Department and 24 from China Hong Kong Observatory), and automatic transmission of rainfall data every 5 minutes. In addition, there are radar and satellite images to monitor the movement and change of rain clouds; (2) The early warning model adopts the critical rainfall standard of Brand et al. (1984): 24-hour rainfall is greater than 175mm, or hourly rainfall is greater than 70 mm ... which was revised several times later; ③ Real-time early warning system; (4) release to the public through television and radio. On average, China and Hong Kong issue early warning of flash floods and landslides three times a year.

In addition, in addition to the early warning model, the disaster situation of the day was also considered. Even if the rainfall is lower than the warning value, the landslide warning will take effect immediately when 1d occurs 15 or more (brand, 1995).

Research status of 1.3.2

Some scholars emphasize the importance of early rainfall and suggest that the total amount of a rainfall process should be used as the criterion of a large number of landslides. However, the Research Data of Landslides in China Mainland and Hongkong (brand, 1984) put forward the hourly rainfall as the critical value of catastrophic landslides, and for the convenience of early warning, put forward the daily rainfall critical value (the critical value of rainfall causing geological disasters).

Brand and others studied the landslide disaster and rainfall data in China and Hong Kong in recent 20 years, and thought that most of the landslides in China and Hong Kong were caused by short-term heavy rainfall, and the occurrence time of these landslides was synchronized with the maximum hourly rainfall time (Figure 1. 12).

Fig. 1. 12 Relationship between landslide occurrence time and maximum hourly rainfall in China.

◆ Lumb( 1975). By this method, the relationship between landslides and rainfall in China and Hongkong is put forward for the first time, and it is considered that the influencing factors are 24-hour daily rainfall and previous accumulated rainfall 15d. 24-hour rainfall (100) and 15d rainfall (200,400). It is considered that in China and Hong Kong, when the 24-hour rainfall is more than 100mm and the accumulated rainfall in the previous15 days is more than 350 mm, the residual soil landslide will occur. Through the analysis of the data in China and Hong Kong 1982, when the maximum hourly rainfall exceeds 40mm, a large landslide will occur (Figure/kloc-0

Figure 1. 13 1982 Maximum hourly rainfall, days of rainfall and days of more than three landslides.

Because it is difficult to predict landslides in advance due to short-term heavy rainfall, and the accumulated rainfall can be estimated several hours before reaching the critical value, it is of great significance to predict landslides by using 24-hour rainfall. Through the analysis of the relationship between the 24-hour rainfall of 1963 ~ 1982 and landslides, it is found that landslides will occur if the 24-hour rainfall in China and Hongkong exceeds100 mm.

The study on the relationship between rainfall and landslides in Hong Kong, China began on June 1972 with the Saomaoping and Poshan landslides, which killed 7 1 person and 69 people respectively. After analyzing the landslide and rainfall data of 1950 ~ 1973, Lumb( 1975) first proposed the relationship between landslide and rainfall in China (Figure 1. 14).

Figure 1. 14 Relationship between landslide and rainfall in China (according to Lumb, 1975).

Lumb pointed out that the number of landslides is related to the 24-hour daily rainfall and the previous accumulated rainfall 15d. When the 24-hour daily rainfall is 100mm and the previous rainfall is 15d, at least 50 landslides will be triggered, leading to catastrophic landslide events; When the 24-hour daily rainfall is 100mm and the previous rainfall is 200mm, 10 ~ 50 landslides will be triggered, causing serious landslide disasters.

Brand et al. (1984) analyzed in detail the relationship between the number of landslides in 1963 ~ 1983 and the cumulative rainfall in 1d ~ 30d, and concluded that the daily average number of landslides and the number of landslide casualties in China and Hongkong had nothing to do with the previous rainfall, but were closely related to the hourly rainfall (Figure/kloc

Fig. 1. 15 relationship between landslide and rainfall in China (according to Brand et al., 1984).

The hourly rainfall of 75 mm is the critical rainfall of catastrophic landslide. At the same time, the 24-hour daily rainfall can also be used as an early warning index of rainfall landslide. When the 24-hour daily rainfall is less than 100mm, the possibility of landslide is very small. When the 24-hour daily rainfall exceeds 200mm, a serious landslide disaster is bound to occur. According to the relationship between rainfall infiltration and pore water pressure, Brand thinks that the high permeability of weathering layer of igneous rocks in China and Hongkong determines the close relationship between landslide and hourly rainfall intensity.

Brand also found that the catastrophic landslides in China and Hongkong occur once every five years and once every two years due to the influence of rainfall cycle, 1a3 times.

According to the research results of Brand and others, the China Hong Kong government started the landslide early warning system in 1984 (Figure 1. 16). The hourly rainfall of 75 mm and the 24-hour daily rainfall 175 mm are determined as the critical rainfall for landslide warning. The forecast results of China and Hongkong show that when the hourly rainfall is more than 75mm, there are 35 landslides on average, with the highest landslide occurring at 55 1 and the lowest landslide occurring at 5.

Since the early warning system was started, landslide warning has been issued three times a year on average, and the actual warning is 1A 1 ~ 5 times. Landslide warnings are usually issued during the heaviest rainfall every year. In addition, even if the rainfall is lower than the warning value, the landslide warning will take effect immediately when 1d (brand, 1995) has a landslide above 15.

In order to constantly revise and improve the landslide early warning system, following 1984, the Hong Kong government of China has intensified its research on rainfall landslides. In addition to the release of the annual reports of rainfall-type landslide survey and rainfall-type landslide survey, it is particularly important to study the relationship between landslides and rainfall, the distribution and development law of rainfall-type landslides, the hydrogeological model of rainfall infiltration, and establish a more accurate relationship between landslides and rainfall by using mathematical methods such as probability and statistics. With the deepening of the research, the researchers agree that the unsaturated soil properties and residual soil properties of the weathered layer of igneous rocks in China and Hongkong control the formation mechanism of shallow rainfall landslides (below 3m).

Au( 1993) systematically studied the relationship between landslide and rainfall.

A) When the daily rainfall is more than 70mm within 24 hours, a large number of landslides will occur, and landslides will also occur when the daily rainfall is 50mm within 24 hours. The probability and number of landslides are directly proportional to the rainfall intensity.

B) Under the same rainfall intensity, the higher the urbanization level, the greater the landslide density.

Au( 1998) later analyzed that:

A) When the rainstorm exceeds 20% ~ 30% of the annual average rainfall, catastrophic landslides will inevitably occur.

Figure 1. 16 China Hongkong Landslide and Debris Flow Early Warning System

(2) Infinite slope model can be used to estimate the critical depth of rainfall landslide. Ma Long (1999) pointed out that the critical rainfall intensity set by the early warning system is not suitable for large-scale deep landslides (the depth is more than 5m), and these landslides are obviously related to the previous rainfall.

Kay and Chen( 1995) suggested that the hourly rainfall and daily rainfall in China and Hongkong jointly affected the landslide activity. Finlay et al. (1997), after analyzing the rainfall and landslide data of 1984 ~ 1993, think that the landslide occurrence probability is closely related to 3h rolling rainfall, and the previous rainfall also directly affects the landslide occurrence probability. According to the data of collapse and rainfall of 1984 ~ 1996, Chau et al. (1997) pointed out that the daily rainfall of rock collapse in Hong Kong, China is 150 ~ 200mm, which is close to the warning rainfall in Hong Kong, China.

Pun et al. (1999) re-evaluated the effectiveness of the landslide early warning system based on the brand results, and came to the conclusion that:

A) The critical rainfall of a)75 mm/h is not suitable for hillside areas with residents;

B) The probability of landslide is related to the rainfall before 15d, and the degree of correlation varies with the scale of landslide;

C) The 24-hour rolling rainfall automatically recorded and calculated every 5 minutes is more closely related to landslides. According to the evaluation results of Pun et al. (1999), the geotechnical engineering department revised the critical rainfall intensity of the landslide early warning system in 1999.

At present, more in-depth research on the relationship between landslide and rainfall is still going on. For example, after studying the landslide and rainfall data of11984 ~ 1997, Dai and Li (200 1) think that:

A)a) 12h rolling rainfall is the most important to predict the number of small landslides, and 24h rolling rainfall has obvious influence on large landslides;

B) The cumulative percentage of the landslide with the scale less than 4m3 is exponentially related to the landslide scale;

C) The reliability of the relationship between landslide and rainfall intensity is closely related to the data accumulation period and data volume. The longer the data accumulation period, the larger the data volume, and the more reliable the results.