Abstract:

By adding geological hazard potential as a parameter, the statistical model of sudden geological hazards early warning has solved the problem caused by rainfall criterion method which could not express the geological environmental condition clearly. The calculation of potential, the design of variable parameters of rainfall and the optimization of early warning equation were also improved in this model. In order to increase the accuracy, an additional step for relatively independent geological environmental factor judging based on chi-square test was supplemented in this model. The initial value of the certainty level of the geological environmental factor and the set of initial value of the certainty level change in the weight calculation were also ruled.  And valid cumulative rainfall was adopted in this model instead of cumulative rainfall, which makes this modle more reasonable. An early warning equation based on a joint probability analysis was proposed in this study, which avoids inconsistency with the basic geological cognition and the physics law, and a lack of the equation coefficients meaning. A 24 h forecast of July 2nd, 2009 was taken as a case study. The results showed that the geological hazard points within the forecast area ratio increased from 44% to 62%, although the forecast area decreased from 101 008 km² to 27 553 km². It indicates that the improved early warning model will be more rigorous theoretically, increase its spatial accuracy and decrease its invalid report rate.

Key words: geological hazards, early warning, certainty factors, potential, multiple linear regression, chi-square test, joint probability