JP3721407B2 - Sediment disaster prediction system, method, and program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a wide-area sediment disaster prediction system, method, and program.
[0002]
[Prior art]
Conventionally, weather forecasts such as weather conditions, weather conditions, rainfall, and lightning have been put into practical use based on weather information acquired by meteorological satellites or observation devices of weather stations. In recent years, meteorological observation equipment has become more sophisticated and sophisticated, and it has become possible to observe the amount of water vapor, wind speed, and wind direction at higher levels using a wind profiler, GPS water vapor amount observation device, etc. Due to budgetary limitations, traditional weather information remains important. The ground weather map and the high-rise weather map in which this conventional weather data is described can be accessed via the Internet, etc. from the meteorological agency of a public institution or a private weather related company. Research on numerical prediction that incorporates high-density observations with such devices, non-hydrostatic equilibrium models, and small-scale convective cloud processes is underway. However, there is no disaster prediction system in which such meteorological data, a weather prediction system, and its technique are directly applied to wide-area (ie, mesoscale) sediment-related disasters and mountainous disasters.
[0003]
As a conventional technique, for example, as a method for predicting a landslide disaster based on observed weather data such as rainfall, “a method for setting a critical rainfall curve for an individual slope for landslide warning and evacuation support” (Non-Patent Document 1) This is to predict the occurrence of sediment-related disasters based on the high-density rainfall data obtained by the rainfall observation system and its prediction information. A wide-area sediment disaster cannot be predicted.
[0004]
As another conventional technique, there is a “development example of a sediment disaster occurrence prediction system using rainfall prediction information” (see Non-Patent Document 2), which is also a high density obtained by the rain observation system. It is intended to predict the occurrence of sediment disasters based on rainfall data and its prediction information, and cannot predict wide-area sediment disasters directly from wide-area meteorological data.
[Non-Patent Document 1]
"Method of setting the critical rainfall line for landslide occurrence for landslide warning and evacuation support" (Summary of the 2002 Sabo Society Research Presentation, 2002, p106-107, Kazumasa Kuramoto et al.)
[Non-Patent Document 2]
"Development example of sediment disaster occurrence prediction system using rainfall prediction information" (2001 Sabo Society Research Presentation Summary, 2001, p198-199, Mitsuhide Tomomura et al.)
[0005]
[Problems to be solved by the invention]
An object of the present invention is to provide a system, a method, and a program for solving the above-described problems and predicting the occurrence of earth and sand disasters and mountainous disasters directly from wide-area weather data.
[0006]
[Means for Solving the Problems]
The wide-area sediment disaster prediction system according to the present invention is
High-level meteorological data including at least one altitude humidity and wind speed / direction in the target area, and front / track score based on whether there is stagnation of the front in the target area and whether there is low pressure or typhoon approach to the target area, An input means for inputting
Storage means for storing the input data;
Based on the input / stored high-rise weather data and the front / track score, analysis using a neural network (NNW) previously trained with past data in a predetermined system or storage device, or based on past data Judgment to determine whether or not mountainous sediment disasters / slope disasters (collapse, landslides, debris flow disasters, etc.) occur from about half a day to about one day later, using the multiple discriminant formula obtained by multiple discriminant analysis (MDA) Means,
Display means for displaying the determination result of the determination means;
It is characterized by including.
According to the present invention, from the small number of data factors extracted from the high-rise weather map that can be obtained free of charge through the Internet, the occurrence of landslide disasters caused by fronts such as the rainy season and autumn rain, or typhoons, is further simplified. Can be predicted with high accuracy. By effectively utilizing such predictions, it is possible to minimize damage by taking measures against disasters. In addition, even if the relationship between rainfall patterns and disasters changes due to local development activities or global warming, the system can be easily adjusted because there are few meteorological factors to use.
[0007]
Moreover, the wide-area sediment disaster prediction system according to the present invention is
The determination means includes ranking means for ranking the degree of risk of occurrence of sediment disaster using a predetermined at least one threshold.
It is characterized by that.
According to the present invention, for example, if two threshold values are used, the degree of risk of occurrence of earth and sand disaster can be ranked in a plurality of stages of “high risk”, “medium risk”, and “low risk”. it can. Therefore, since this ranking can be presented to the user, the user can easily and intuitively grasp the risk of occurrence of a landslide disaster.
[0008]
The present invention can be implemented not only in the form of the system (apparatus) described above but also in the form of a method, program, or recording medium storing the program substantially corresponding to these systems.
For example, the wide area sediment disaster prediction method according to the present invention is as follows.
High-level meteorological data including at least one altitude wet number, wind speed and direction in the target area, and front / track score based on whether there is stagnation of the front in the target area and whether there is low pressure or typhoon approach to the target area An input step of inputting and storing in the storage means;
Based on the input / stored high-rise weather data and the front / track score, analysis by the neural network (NNW) trained by the past data stored in the storage means or multiple discriminant analysis by the past data ( Using the discriminant formula obtained by MDA), calculate means (CPU, etc.) to determine whether or not there will be a landslide / slope disaster (collapse, landslide, debris flow disaster, etc.) A determination step using and determining;
A display step of displaying the determination result of the determination step on a display means;
It is characterized by including.
The method according to the invention also comprises:
The determining step includes a ranking step of ranking the degree of risk of occurrence of sediment disaster using a predetermined at least one threshold value.
It is characterized by that.
[0009]
For example, a program for causing a computer to execute a wide-area sediment disaster prediction method according to the present invention,
The method comprises
High-level meteorological data including at least one altitude wet number, wind speed and direction in the target area, and front / track score based on whether there is stagnation of the front in the target area and whether there is low pressure or typhoon approach to the target area An input step of inputting and storing in the storage means;
Based on the input / stored high-rise weather data and the front / track score, analysis by the neural network (NNW) trained by the past data stored in the storage means or multiple discriminant analysis by the past data ( Using the discriminant formula obtained by MDA), calculate means (CPU, etc.) to determine whether or not there will be a landslide / slope disaster (collapse, landslide, debris flow disaster, etc.) A determination step using and determining;
A display step of displaying the determination result of the determination step on a display means;
It is characterized by including.
The program according to the present invention is
The determining step includes a ranking step of ranking the degree of risk of occurrence of sediment disaster using a predetermined at least one threshold value.
It is characterized by that.
[0010]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a block diagram showing basic components constituting a sediment disaster prediction system according to the present invention. As shown in the figure, a sediment disaster prediction system 100 according to the present invention includes an input unit 110, a determination unit 120, a ranking unit 130, a display unit 140, and a storage unit 150. Meteorological data of the prediction target area required for input is received via the network from a weather data server that publishes the data on the network such as the Internet, or the weather data input from the user or the administrator of this system Are received directly or via the Internet. The meteorological data received in this way is input by the input means 110 and temporarily stored in the storage means 150. In response to this input, the determination means 120 uses a neural network analysis based on past weather data and disaster data, or a multiple discriminant analysis based on past weather data and disaster data (approximately one prefecture or so). It is determined whether or not a landslide disaster or a mountainous disaster will occur in about 1 day after about half a day in the prediction target area.
[0011]
FIG. 2 is a flowchart for explaining an example of the processing flow of the sediment disaster prediction method according to the present invention in which southern Kyushu is the target area. As shown in the figure, in step S10, it is determined whether the input person is in the rainy season, typhoon season, or other time. If the time is other than that, the process ends because it is difficult to predict. In the case of the rainy season, in step S20a, the wind speed and humidity at the 850 hpa altitude of Kagoshima, the position of the rainy season front and the presence or absence of a meso cyclone, etc. are input as weather factors. On the other hand, in the case of a typhoon, in step S20b, the wind speed and the number of humidity at 850 hpa altitude in Kagoshima between half a day and one day before the closest typhoon are input as weather factors. In order to improve the accuracy of prediction, when the wind direction is “northward” or “eastward”, the wind speed is set to 0 instead of the actual wind speed. In the rainy season, in step S30a, in the case of a typhoon, in step S30b, a neural network calculation or a calculation based on a multiple discriminant is performed to determine whether or not a landslide disaster occurs. Similarly, in step S40a or S40b, the risk of landslide disaster is displayed (output) in three stages (high risk, medium, and low) based on the determination in the previous step, and the process is terminated. The determination of the occurrence of a disaster in steps S30a and S30b is based on the learning result of the NNW three-layer fully coupled BP (back-propagation method) net or the multiple discriminant obtained by multiple discriminant analysis, whichever has the better predictive value. It is preferable to use it.
From the viewpoint of prediction accuracy, it is desirable to select at least three of the available weather factors as the weather factors to be used, and this selection is adjusted to select the optimal one for each prediction target area. It is desirable to do.
[0012]
Next, several examples of the multiple discriminant obtained by the multiple discriminant analysis based on past data will be given. For example, the discriminant for the rainy season is as follows.
Xc = a · W ₈₅₀ + b · Td ₈₅₀ + c · FmLs + d (1)
Here, Xc is a discriminant function, and a, b, c, and d are coefficients for each observation point regarding each meteorological factor, and are preferably set to be optimal for each predicted region. For example, in the case of South Kyushu, a = 0.115, b = −0.538, C = 3.13, and D = −0.875 are set. W ₈₅₀ is wind speed at 850hpa altitude (m / s, northward wind is 0), FmLs is front score, and Td ₈₅₀ is humidity at 850hpa altitude (° C, difference between dew point temperature and temperature) It is. Furthermore, Xc = 0.0 is a threshold value, and it is predicted that a sediment disaster will occur when Xc> 0.0.
[0013]
For example, the discriminant for a typhoon is as follows.
Xc = a · W ₈₅₀ + b · Td ₈₅₀ + c · FmLs + d (2)
Here, Xc is a discriminant function, and a, b, c, and d are coefficients for each observation point regarding each meteorological factor, and are preferably set to be optimal for each predicted region. For example, in the case of South Kyushu, a = 0.197, b = -0.181, C = 0.00, and D = -3.81 are set. W ₈₅₀ is the wind speed at 850hpa altitude (m / s, northward wind is 0), FmLs is the course score, and Td ₈₅₀ is the humidity at 850hpa altitude (° C, difference between dew point temperature and temperature) It is. Furthermore, Xc = 0.0 is a threshold value, and it is predicted that a sediment disaster will occur when Xc> 0.0.
[0014]
The front and course scores FmLs used as input data in the present invention are basically set based on empirical rules of meteorological research and forecasting work common to past disaster weather, but sensitivity analysis of score values (values were changed) It is desirable to set an optimal value by analyzing the increase or decrease of the predictive predictive value at the time. For example, in the northern Kyushu to Chugoku region, we tried to analyze by changing the score FmLs = 0.5 when either the front or low pressure only exists in the area to 0.2, 0.3, 0.4, 0.7, There were few cases of disasters when either the front or the low pressure existed in the area, and only a reasonable result was obtained when the score was close to 0.1. However, FmLs = 0.5 was used because it is empirically known that disasters may occur even if only such fronts or lows exist in the area.
[0015]
FIG. 3 is a diagram showing an example of a neural network used in the sediment disaster prediction system according to the present invention. As shown in the figure, the neural network consists of three layers: an input layer, a hidden layer (learning and judgment area), and an output layer. The input layer has the same number of input factors as the input layer, and the hidden layer has the same number as the input layer. The output layer has one unit. In this example, feedback learning using learning data (teacher signal consisting of weather data at the time of past disaster occurrence / non-occurrence) is a fully-coupled type in which units of different layers can be associated with each other and exchange information. “BP (Back Propagation) Net” is used.
[0016]
Next, using the earth and sand disaster prediction system according to the present invention, the Kagoshima region of the South Kyushu type was selected as the prediction target area, and an experiment was performed to predict the occurrence of earth and sand disasters during the rainy season by the multiple discriminant analysis. The meteorological factors used in this system include front and track score FmLs based on whether the rainy season front is stagnant on Kyushu and whether the Meso cyclone is close to the East China Sea or Kyushu. Is 1.0, 0.5 for the front line only, and 0.0 for the case without both. This should be adjusted according to the target area.), Water supply and weather conditions for the lower and middle layers of Kagoshima In addition to using high-rise weather data (850 hpa, 700 hpa high humidity T-Td and wind speed W) as representative factors, data from Fukuoka, Nase, and Jeju Island, Korea were also used as factors. As a result, Fukuoka, Nase, and Jeju are inappropriate as a cause of disaster conditions because there are many cases where the lower layer jet is not seen and the wet number T-Td is large when a sediment disaster occurs in Kagoshima in the target area. I did not use it. In addition, we obtained the most accurate results when using the wet number T-Td, wind speed W and front / track score at each altitude of 700hpa and 850hpa, but the wind of 700hpa is weak as a problem of "over adaptation" It is against the mechanism of precipitation, such as the occurrence of a disaster, so "700hpa, 850hpa altitude wet number T-Td and front / course score FmLs" or "850hpa altitude wet number T-Td, wind speed W , The front discriminant / course score FmLs ”was used, and the multiple discriminant analysis was performed by the system according to the present invention. Comparing each result, the hit rate Rh of sediment-related disasters is 0.875 for the former and 0.857 for the latter, and the Mahalanobis square distances are about 3.34 and 3.77, respectively (the higher this value is, the better the accuracy). The threat score Ts, which is a critical success index focusing only on, was 0.75 and 0.76, respectively. In the case of disasters, it is important that the accuracy of Ts is higher than that of Rh, and because it is consistent with the meteorological precipitation mechanism, the latter meteorological factor, that is, “wet-number T- It is preferable to use “Td, wind speed W, front / course score FmLs”. A similar prediction experiment using multiple discriminant analysis was conducted for the rainy season in the Chugoku region, where the weather factors were almost the same as those in Southern Kyushu, and Rh = 0.938 and Ts = 0.800 were obtained. Thus, the prediction accuracy was slightly higher in the Sanin region than in Southern Kyushu.
[0017]
Next, a similar prediction experiment using the earth and sand disaster prediction system according to the present invention was performed using a neural network (three-layer fully coupled BP network shown in FIG. 3). The tolerance was set to 0.01 according to previous studies, and the learning data was 16 data from 1993 to 1995 in order to consider the effects of global warming and to make the same number of disaster occurrence / non-occurrence data. . The forecast targets were 16 data from 1969 to 1986, mainly from 1993 to 2000. The prediction results are Rh = 0.842, Ts = 0.750, miss rate = 0.182, idle rate = 0.125. By the way, according to the data of 1998 Japan Meteorological Society, Ts of short-term precipitation prediction of the Japan Meteorological Agency is about 0.3 to 0.6, and the prediction according to the present invention is much more accurate. In addition, as a result of predicting landslide disasters during the rainy season in the Chugoku region by using this system using a neural network, Rh = 0.875-0.926 and Ts = 0.268-0.400, which was lower than that of Kagoshima. This is probably because the weather conditions are slightly more complicated in Chugoku than in Kagoshima and the difference in learning data.
[0018]
In the present specification, the principle of the present invention has been described in various embodiments. However, the present invention is not limited to the above-described embodiments, and many variations and modifications can be made. It should be understood that these are also included in the present invention.
For example, in the embodiment, the disaster prediction is performed with the southern Kyushu and the San-in / Chugoku regions as target regions. However, the principle of the present invention can be applied to other regions and the occurrence of the disaster can be predicted. Further, according to the prediction result obtained by implementing the prediction system according to the present invention, it has been found that the hit ratio in the range from half a day to one day later is very good. However, it should be noted that although the prediction accuracy is slightly reduced, the present invention can also predict the occurrence of a disaster before half a day or after 24 hours.
[Brief description of the drawings]
FIG. 1 is a block diagram showing basic components constituting a sediment disaster prediction system according to the present invention.
FIG. 2 is a flowchart for explaining an example of the processing flow of the sediment disaster prediction method according to the present invention in which southern Kyushu is the target area.
FIG. 3 is a diagram showing an example of a neural network used in a sediment disaster prediction system according to the present invention.
[Explanation of symbols]
110 Input means 120 Determination means 130 Ranking means 140 Display means 150 Storage means

Claims (6)

A wide-area sediment disaster prediction system,
Input means for inputting high-rise meteorological data including at least one altitude wet number and wind speed, and front / track score based on whether there is stagnation of the front and whether or not low pressure or typhoon is approaching the target area;
Based on the high-rise weather data and the front and course scores that have been input, landslide disasters occurred using multiple discriminant formulas obtained by analysis using a neural network that was previously learned from past data or multiple discriminant analysis using past data Determining means for determining whether or not to do;
Sediment disaster prediction system characterized by including In the earth and sand disaster prediction system according to claim 1,
The determination means includes ranking means for ranking the degree of risk of occurrence of sediment disaster using a predetermined at least one threshold.
A system characterized by that. A wide-area sediment disaster prediction method,
High-level meteorological data including at least one altitude wet number, wind speed and direction in the target area, and front / track score based on whether there is stagnation of the front in the target area and whether there is low pressure or typhoon approach to the target area An input step to input,
Based on the input high-rise weather data and the front / course score, the multiple discriminant obtained by the analysis by the neural network previously learned by the past data stored in the storage means or the multiple discriminant analysis by the past data A determination step for determining whether or not a sediment disaster occurs using a computing means,
A display step of displaying the determination result of the determination step on a display means;
A method comprising the steps of: In the sediment disaster prediction method according to claim 3,
The determination step includes a ranking step of ranking a degree of risk of occurrence of sediment disaster using a predetermined at least one threshold value.
A method characterized by that. A program for causing a computer to execute a wide-area sediment disaster prediction method,
The method comprises
High-level meteorological data including at least one altitude wet number, wind speed and direction in the target area, and front / track score based on whether there is stagnation of the front in the target area and whether there is low pressure or typhoon approach to the target area An input step to input,
Based on the input high-rise weather data and the front / course score, the multiple discriminant obtained by the analysis by the neural network previously learned by the past data stored in the storage means or the multiple discriminant analysis by the past data A determination step for determining whether or not a landslide disaster occurs,
A display step of displaying the determination result of the determination step on a display means;
The program characterized by including. The program according to claim 5,
The determining step includes a ranking step of ranking the degree of risk of occurrence of sediment disaster using a predetermined at least one threshold value.
A program characterized by that.

Images