KR101563244B1 - System and Method for physically based forecasting of Changma precipitation - Google Patents

Abstract

The present invention relates to an objective method of constructing a prediction model for seasonally predicting rainy season precipitation by applying a statistical method using the sea surface temperature physically affecting rainy season precipitation and selecting an area having a correlation coefficient with a rainy season index An association factor database unit for defining and databaseing the association factors using the region associated with the rainy season precipitation selected by the regional selection unit, a correlation factor defined above, and a cross-correlation test using the forward stepwise regression method, A prediction model constructing unit for constructing a multiple regression prediction model using the preceding factors determined in the preceding factor determiner, a physical process analyzing unit for analyzing a physical process of each preceding factor on rainfall precipitation, ; ≪ / RTI >

Description

Technical Field [0001] The present invention relates to a system and method for predicting physical statistics of rainy season precipitation,

In particular, the present invention focuses on the physical phenomena that affect the interannual variability of rainy season precipitation and uses the seasonal sea surface temperature, which is a related boundary forcing, to predict the rainy season precipitation intensity To a system and method for predicting physical statistics.

In the East Asian region including the Korean peninsula, a lot of rain comes along with the formation of electric wires between mid-June and late July. In this way, the period of rain in early summer is referred to as the rainy season in Korea, and it has a great influence on human activities such as agriculture, commerce, and leisure activities.

Since the rainy season has a great impact on real life, accurate forecasting of rainy season precipitation is becoming more important.

However, the prediction of the rainy season is still being attempted by using the dynamical model of the global or regional scale, but the prediction accuracy is low.

In fact, simulation of East Asian summer precipitation by ensemble of the best available numerical models has resulted in a remarkable decrease in prediction performance due to the correlation coefficient between 0.1 and 0.25 (Wang et al. 2009).

On the other hand, Wu et al. (2009) proposed a method of predicting physical statistics for the East Asian monsoon intensity using the northern spring vibration, development and declining El Niño.

The constructed prediction method has a correlation coefficient of 0.79 with observations and showed that the physical statistical prediction method is an alternative to the numerical model.

The prediction of the high prediction rate using the above-mentioned physical statistical prediction method can be applied to the rainy season precipitation.

It is possible to build predictive statistical model by effectively analyzing the strong interannual variability of rainy season precipitation, and it is possible to clarify causal relationship between predecessor factor and rainy season precipitation by physical process analysis.

Therefore, it is necessary to construct a statistical model for seasonal prediction, including studies on spring physical phenomena that affect interannual variability of rainy season precipitation.

Korean Patent No. 10-1015249

The present invention focuses on the physical phenomenon that affects the interannual variability of rainy season precipitation and the associated boundary forcings, The purpose of this paper is to provide a method for constructing a physical statistical prediction model of rainy season precipitation using spring sea level temperature forcing.

It is an object of the present invention to provide an objective system and method for seasonally predicting rainy season precipitation using the sea surface temperature, which is a boundary forcing related to physical phenomena affecting interannual variability of rainy season precipitation.

The present invention provides a method for defining the boundary forcing associated with the rainy season, and a system for identifying the choice of preceding factors that perform effective prediction using the forward stepwise regression method and the physical role of the preceding factors And a method thereof.

The present invention relates to a physical statistical prediction system of rainy season precipitation that can stabilize and reliably predict the preceding factor period by normalizing 20 ~ 60 days average by using average and standard deviation And a method thereof.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

In order to achieve the above object, according to the present invention, a system for predicting the physical storm of rainy season precipitation includes a region selection unit for selecting an area having a correlation coefficient with a rainy season index, An association parameter database for defining a parameter and defining a database, a pre-factor determiner for finding a pre-factor combination having a high predictive value by cross-testing the association factor defined above with a forward step-by-step regression method, A prediction model constructing unit for constructing a multiple regression prediction model, and a physical process analysis unit for analyzing a physical process of each preceding factor on rainfall precipitation.

Here, the area selecting unit selects an area based on a correlation coefficient of sea surface temperature which is a boundary forcing with the rainy season index, and the selection period is limited to March to May, which is a spring period.

In the association parameter database, the association factors are defined as the association area averages. The average of the correlation factors is averaged for 20 to 60 days in order to reduce the stability of the prediction model and the error of the data, and normalization is performed using the mean and standard deviation And is converted into a database.

The rainy season index used in the regional selection system is defined as the mean value of the rainy season index to be predicted from the Korean Peninsula during the period from June 15 to July 31, .

In order to analyze the physical process of each predecessor to rainfall precipitation, the Pacific Ocean sea surface temperature line of 40 ° ~ 45 ° North is strengthened atmospheric pressure and feedback of the amount of wind and wind The North Pacific development physics process, which sustains the fronts until summer and affects the rainy season precipitation, the North Atlantic physics process where the effects of the North Atlantic oscillations affect the rainy season precipitation through wave propagation in the upper atmosphere, and the northwestern Pacific region by El Niño and La Niña Precipitation phenomenon is maintained through cyclonic cycling and positive feedback and analyzes the central pacific physics process which affects rainy precipitation.

In order to analyze the physical processes of each predecessor on rainy season precipitation, the physical process analysis department analyzes the physics of the three predecessors of the North Pacific Development Index (A), North Atlantic 1 Index (B) and Central Pacific El Niño Index (C) The process is commonly characterized by the formation of anomalies in the southeastern part of Japan and the process of influencing the rainy season precipitation by increasing the amount of water entering south of the peninsula.

According to another aspect of the present invention, there is provided a physical statistical prediction method for rainy season precipitation, comprising: a first step of calculating a predicted value by applying a cross test to each correlation factor to determine a preceding factor for predicting physical statistics of rainy season precipitation; A second step of selecting a preceding factor having a high correlation coefficient between the predicted value and the observed value, a third step of verifying whether the correlation coefficient between the selected factor and the rainy season index has a value larger than the 10% significance level threshold, And verifying that the coefficient of variance of the factor with the factor in the verification step has a value less than 2. The fourth step of verifying whether the correlation factor between the determined preceding factor and the factor in the verification step has a value smaller than the 10% And a fifth step of determining a preceding factor.

The area and duration of the determinant preceding factors are determined by performing a spatial average (160 ° -210 ° E, 20 ° -35 ° N) of sea surface temperature anomalies in the North Pacific Development Index (A) The North Atlantic 1 index (B) represents the first spatial mean (40 ° to 15 ° W, 55 ° to 60 °) of the sea surface temperature anomalies, N), the time average (4.6 ~ 4.25) was performed after subtracting the second spatial mean (80 ° ~ 40 ° W, 30 ° ~ 45 ° N) and the central Pacific El Niño Index (C) And a time average (4.1 to 4.20) is performed after the spatial average (160 ° to 140 ° W, 15 ° S to 10 ° N) of the sea surface temperature anomaly is performed.

The predicted factors were the North Pacific Development Index (A), the North Atlantic 1 Index (B) and the Central Pacific El Niño Index (C), and the multiple regression model conceived as a determinant predictor was rainy precipitation = 0.458A - 0.413B - 0.396C + 0.0136.

The system and method for predicting the physical storm of rainy season precipitation according to the present invention have the following effects.

First, it can be used as an objective indicator to anticipate the social and economic effects of the Korean peninsula by forecasting the rainy season precipitation using the spring data.

Second, from the viewpoint of the large scale, understanding the physical mechanism related to the fluctuation of the rainy season can improve the understanding of the rainy season from the meteorological point of view.

Third, it can be used as a basis for the development of the physical statistical model using the sea surface temperature prediction result of the mechanical model.

Fourth, the averaging time of 20 ~ 60 days is applied to reduce the error of the data, and each correlation factor is normalized by using mean and standard deviation, so that reliable and reliable prediction is possible.

FIG. 1A is a block diagram of a method for predicting physical statistics of rainy season precipitation in accordance with the present invention
1B is a flowchart showing a forward stepwise regression method for constructing a physical statistical prediction method of rainy season precipitation;
Figure 2 shows the distribution of regression coefficients of the rainy season index and spring anomalies
FIG. 3 is a block diagram showing an area used to determine a determined preceding factor
4 is a time-series graph showing the results of prediction using the rainy season index and the statistical prediction model
5 to 7 are diagrams showing the regression coefficients of the rainy season rainfall anomaly, the spring sea surface temperature anomaly, and the rainy season sea surface temperature anomaly,

Hereinafter, a preferred embodiment of a system and method for predicting physical statistics of rainy season precipitation according to the present invention will be described in detail.

The characteristics and advantages of the system and method for predicting the physical statistics of rainy season precipitation according to the present invention will be apparent from the detailed description of each embodiment below.

FIG. 1A is a configuration diagram of a physical statistical predicting method of rainy season precipitation according to the present invention, and FIG. 1B is a configuration diagram of a forward stepwise regression method for constructing a predictive model.

The present invention focuses on physical phenomena that affect interannual variability of rainy season precipitation, and relates to an objective method for predicting rainy season precipitation using the seasonal spring forcing temperature associated therewith.

To do this, we propose a method to define the boundary forcing related to the rainy season, and to select the preceding factors that perform effective prediction using the forward stepwise regression method and to identify the physical role of the preceding factors And methods.

As shown in FIG. 1A, the system for predicting rainy season precipitation according to the present invention includes an association region selection unit 10 having a correlation coefficient with a rainy season precipitation and a correlation factor A predicting model constructing unit 40 for determining a regression coefficient and a regression equation of a determined preceding factor, a predicting model determining unit 30 for determining a predicting factor having a high predicting rate by using a forward stepwise regression method, And a physical process analysis unit 50 for analyzing the physical process of the determined preceding factors on the rainy season.

Here, the preceding factor determiner 30 includes a first step (S101) of calculating a predicted value by applying a cross test to each of the correlation factors to determine a preceding factor having a high predictive value using a forward stepwise regression method, The second step (S102) of obtaining a correlation coefficient with the rainy season index and selecting a factor with a high correlation coefficient and the third step (S102) of verifying whether the correlation coefficient between the selected factor and the rainy season index has a value greater than the 10% A fourth step (S104) of verifying whether the variance inflation factor between the determined preceding factors and the factor in the verification step has a value smaller than 2, (S105) for verifying whether each factor has a correlation coefficient smaller than the 10% significance level threshold, and three associative factors satisfying the criteria of the above 3-5 (S103 ~ S105) 6 < RTI ID = 0.0 > Step S106 is performed.

The physical statistical prediction method of the physical statistical prediction system of rainy season precipitation according to the present invention having such a configuration is as follows.

First, the rainy season precipitation that is the subject of prediction of the present invention is indexed.

The rainy season index is calculated by averaging the rainy season from June 15 to July 31 of the 60 precipitation points distributed in the Korean peninsula. The precipitation of the date when the typhoon affected the Korean peninsula was the mean value .

Excluding the precipitation days on the Korean peninsula, typhoons can consider the interannual variability of rainy season precipitation due to large-scale atmospheric movements, thus making it possible to construct stable prediction models.

On the other hand, since the features of cyclic fields and sub-tropical jets of the East Asian and Pacific Northwest Pacific monsoon including the rainy season vary rapidly from 1994 (Kwon et al. 2005), the period considered to constitute the predictive model of the present invention is 1994 Then,

FIG. 2 is a regression coefficient distribution diagram of the rainy season index and the spring anomalies of the sea surface temperature.

Negative values are maintained during the spring period in the Central Pacific and weak values are distributed in the northwest Pacific.

The positive values of the North Pacific advance little by little during the spring period. In the North Atlantic, patterns associated with the North Atlantic oscillations are seen by April.

On the other hand, although the overall values are weak in the Indian Ocean, the related region is selected based on the previous research that sea surface temperature in India may affect the East Asian monsoon.

Table 1 shows the database of association factors obtained through the difference between the area averages or the area averages based on the spatio-temporal characteristics of each selected association area.

In Table 1, each association factor is averaged over a period of 20 to 60 days.

For example, in the case of the El Niño 3 index, an association factor is defined by averaging the region averages over 20 days, and an association factor is defined by increasing the average period by 5 days and averaging over 25 days.

In this way, a total of 12 correlation factors are created within 20 to 60 days by increasing the average period by 5 days, and then normalized using the mean and standard deviation.

The preceding factors determined in the predecessor determining unit 30 in Table 1 are denoted by A, B, and C in the order of magnitude contributing to prediction in the prediction model.

In the embodiment of the present invention, the forward step-by-step regression method is applied in order to determine the preceding factors to be used in the prediction model in the constructed association parameter database.

Here, the number of preceding factors to be determined in order to prevent prediction errors due to over-fitting in the prediction model is limited to three.

For the stepwise regression method, the first step is to apply a cross-validation test for each of the relevant factors from 1994 to 2011 to calculate the predicted value.

In this case, the unit of calculating the predicted value is 4 years, which is within a range of 20 to 30% of the 1994 to 2011 period, thereby preventing the over-fit in the cross-check process (Blockeel and Struyf 2002).

Next, the correlation coefficient between the predicted value and the rainy season index is obtained and the factor having the highest correlation coefficient is selected.

The selected factors must meet the following three criteria.

1) It should have a statistically significant relationship with the rainy season precipitation, so check whether the correlation coefficient with the rainy season index is above the 10% significance level threshold.

2) Confirm that the coefficient of dispersion expansion is less than 2 in order to check whether the relationship between the determined preceding factors and the factor of the verification step is independent.

3) It is verified whether each correlation coefficient between the determinants and the correlation factors of the verification step has a value below the 10% significance level threshold, and whether the relationship between each factor is statistically independent.

If the above three criteria are not satisfied, the third step is not determined as a preceding factor and the second highest level of correlation coefficient is applied.

When the above three criteria are satisfied, the factor of the verification process has a high linear relationship with the rainy season precipitation and the relation with the preceding factors is independent. Therefore, the selected factor is determined as the preceding factor.

If the number of determinants is less than three, go back to the first step, apply the cross-validation, including the determinant, and proceed to the next step, repeating this sequence until the three preceding factors are determined.

The preceding factors determined in the examples of the present invention are the North Pacific Development Index (160 ° -210 ° E, 20 ° -5 ° N, [4.11-4.30] - [3.22-4.10]), North Atlantic 1 Index ° W, 55 ° - 60 ° N] - [80 ° - 40 ° W, 30 ° - 45 ° N], 4.6 to 4.25), the Central Pacific El Niño Index (160 ° - 40 ° W, N, 4.1 ~ 4.20), and the area used in the construction of the leading factor is shown in FIG.
Here, the North Pacific Development Index is calculated as the mean (160 ° to 210 ° E, 20 ° to 35 ° N) of the sea surface temperature anomalies and then the second period (3.22 to 4.10) in the first period (4.11 to 4.30) The North Atlantic 1 index measures the second spatial mean (80 ° - 40 ° W, 30 °) in the first spatial mean (40 ° to 15 ° W, 55 ° to 60 ° N) of sea surface temperature anomalies The central Pacific El Niño index is the average of the spatial mean anomalies of the sea surface (160 ° to 140 ° W, 15 ° S to 45 ° N), and the time average (4.6 to 4.25) 10 ° N), followed by a time average of 4.1 to 4.20.

The statistical model based on the determinants is rainfall precipitation = 0.458A - 0.413B - 0.396C + 0.0136.

Figure 4 shows the result of cross-validation using the preceding factors from 1994 to 2009, and the result of 2010 ~ 2012 is the prediction using the constructed statistical model.

The dotted line is a line dividing the number of 1994-2012 by 1/3 based on the normalized size, which corresponds to a value of 0.43 in the normal distribution. Using this line, it can be judged whether the predictive model can effectively predict not only the trend of rainy season precipitation but also the intensity.

When it is located in the above area based on the dotted line, it means a normal year if it is located between the strong rainy season and the weak rain when it is located below.

Table 2 is a contingency table comparing the results of observations and predictions by dividing into strong, normal, and weak solutions based on the dotted line in FIG.

The forecasting models fit well in 16 diagonal directions from the upper left to the lower right, which is an excellent predictor of 84.2% of the total period.

Table 3 shows the prediction performance of the prediction model.

The correlation coefficient between observed and predicted values for the period 1994 to 2012 is 0.83 and the mean square root deviation is 0.55.

In addition, the Gerrity-skill score obtained from Table 2 is 0.80 (it has a value of 1 when both predictions and observations coincide), indicating that the prediction is well performed.

The physical process analysis unit (50) clarifies the cause and effect of the preceding factors of the statistical model physically having rainy season precipitation.

FIG. 5 shows the North Pacific Index and the 850-hPa winds, showing the regression coefficient of the summer precipitation anomalies (Fig. 5a), spring sea surface temperature anomalies (Fig. 5b) and summer sea surface temperature anomalies The regression coefficient is the vector.

6 is a graph showing the North Atlantic 1 index and the 850-hPa (Fig. 6a), and the North Atlantic 1 index and the summer precipitation anomaly (Fig. 6a), the spring sea surface temperature anomaly (Fig. 6b) and the summer sea surface temperature anomaly The regression coefficient of the wind is expressed as a vector.

FIG. 7 shows the regression coefficients of the central Pacific El Niño and summer precipitation anomalies (FIG. 7A), spring sea level temperature anomalies (FIG. 7B) and summer sea level temperature anomalies (FIG. 7C) -hPa The regression coefficient of the wind is expressed as a vector.

Specifically, Figure 5 shows the North Pacific (160 ° -210 ° E, 20 ° - 35 ° N, 4.11 ~ 4.30) and the rainy season precipitation anomaly (Figure 5a) (A line) of the sea surface temperature anomalies in spring (Fig. 5b) and the sea surface temperature anomalies (Fig. 5c) in the rainy season and the distribution of the regression coefficients (vectors) of the 850-hPa wind field anomalies.

The Pacific Ocean's sheer sea-level temperature Anomalies gradually move northward, forming a sea-surface temperature line with a sudden temperature change from north to south around 40 degrees north.

The sea surface temperature wire enhances atmospheric pressure and strengthens the westerly anomalies from the atmospheric to the atmospheric upper layers. The strengthening of the westerly anomalies again strengthens the temperature hardness around the sea surface temperature wire to create a feedback relationship between the atmosphere and the ocean.

Through this process, the westerly anomalies (Fig. 5c) around 40 degrees north of the North Pacific maintained until summer, forming an anomaly in the south-eastern part of Japan by creating an environment in which the North Pacific high pressure is extended to the north, The amount of water vapor is increased to form a positive precipitation anomaly (Fig. 5A).

Figure 6 shows the North Atlantic 1 index and the rainy season precipitation anomalies (Figure 6a), spring sea level anomalies (Figure 6b), and rainy season sea level temperature anomalies (Figure 6c ) And the regression coefficient distribution (vector) of 850-hPa wind field anomalies.

The North Atlantic vibration pattern shown in FIG. 6b affects the upper atmosphere circulation field, and this influence propagates in the form of waves to the East Asian region and is maintained until summer (FIG. 6C).

At this time, 60 degrees north latitude and 30 degrees north latitude spread. This propagated wave forms a strong anomaly in the south of Japan and inflows a large amount of water vapor to the south of the peninsula to increase rainfall precipitation (Fig. 6a).

Figure 7 shows the effect of the tropical Pacific sea level temperature pattern on the rainy season in order to investigate the effects of the central Pacific El Niño and the rainy season precipitation anomalies (Figure 7a), spring sea level anomalies (Figure 7b) 7c) and the regression coefficient distribution (vector) of 850-hPa wind field anomalies.

At this time, the coefficient of the Central Pacific El Niño Index is negative, and the Tropical El Niño Index used in Figure 7 is multiplied by -1 to take into account the direct effect of the Central Pacific El Niño Index on forecasting.

The spring La Niña pattern shown in Figure 7b is characterized by a negative sea surface temperature anomalies and a dystrophic anomaly in the tropical Pacific region.

Dongfeng Anomal strengthens the cyclone circulation in the northwest Pacific region and increases precipitation anomalies. The increase in precipitation anomalies creates a positive feedback relationship that re-strengthens the cyclone circulation in the northwest Pacific region through latent heat release and maintains a cyclonic circulation until summer (Fig. 7c).

The impact of the low-pressure anomalies in the northwest Pacific is propagated northeastward to form the hypertonic anomalies in the southeastern part of Japan via Pacific-Japan teleconnection and increase the amount of water entering south of the peninsula (Fig. 7a ).

The system and method for predicting the rainy season precipitation in accordance with the present invention focuses on physical phenomena affecting interannual variability of rainy season precipitation and predicts precipitation using the seasonal spring forcing temperature, It predicts precipitation.

As described above, it will be understood that the present invention is implemented in a modified form without departing from the essential characteristics of the present invention.

It is therefore to be understood that the specified embodiments are to be considered in an illustrative rather than a restrictive sense and that the scope of the invention is indicated by the appended claims rather than by the foregoing description and that all such differences falling within the scope of equivalents are intended to be embraced therein It should be interpreted.

10. Regional governance 20. Associative factors Database author
30. Precedence factor determination unit 40. Prediction model construction unit
50. Physical Process Analysis Department

Claims (9)

Regional prefectural governments selecting regions with correlation coefficient with rainy season index;
An association factor database unit for defining and databaseing association factors using an area associated with rainfall precipitation selected by the regional selection unit;
A preceding factor determiner for finding a combination of preceding factors having a high predictive value by cross-testing the association factor defined above with a forward stepwise regression method;
A prediction model constructing unit for constructing a multiple regression prediction model using the preceding factors determined by the preceding factor determiner;
And a physical process analysis unit for analyzing the physical process of each preceding factor on rainy season precipitation. The apparatus according to claim 1,
The region is selected based on the correlation coefficients of sea surface temperature, which is the boundary forcing, with the rainy season index. The selected period is limited to March to May, which is the spring season. 2. The apparatus according to claim 1,
The correlation factors are defined as the average of the correlated regions, and are used as averages for 20 to 60 days to reduce the stability of the prediction model and the error of the data. Normalization is performed using the mean and the standard deviation, Physical Statistical Prediction System for Precipitation. [2] The method of claim 1,
The definition of rainy season index, which is the subject of the forecast, is defined as the mean value of observations on the Korean peninsula during the period from June 15 to July 31, and the value of the date when the typhoon influenced is excluded from the mean. system. [2] The method of claim 1, wherein the physical process analyzing unit analyzes physical processes of each preceding factor on rainy season precipitation,
The North Pacific developmental physics course, which strengthens the atmospheric pressure of the Pacific Ocean at latitudes 40 ° to 45 ° N and strengthens the atmospheric pressure and maintains the line until summer due to the feedback of the amount of wind and wind,
The North Atlantic physics process, which affects rainy season precipitation through wave propagation in the upper atmosphere,
A physical statistical prediction system for rainy season precipitation characterized by precipitation phenomena in the Northwest Pacific region due to El Niño and La Niña, which is maintained through cyclonic cycling and positive feedback and analyzes the central Pacific physical processes affecting rainfall precipitation. 6. The method according to claim 5, wherein the physical process analyzing unit analyzes physical processes of each preceding factor on rainy season precipitation,
The physical processes of the three predecessors of the North Pacific Development Index (A), the North Atlantic 1 Index (B), and the Central Pacific El Niño Index (C) are common to the southwest of Japan with the formation of hyperbaric anomalies, And analyzing the processes affecting the rainy season precipitation. In order to determine the leading factor for predicting physical statistics of rainy season precipitation,
A first step of calculating a predicted value by applying an intersection test to each association factor;
A second step of selecting a preceding factor having a high correlation coefficient between the predicted value and the observed value calculated in the first step;
A third step of verifying whether the correlation coefficient between the factor selected in the second step and the rainy season index has a value greater than the 10% significance level threshold;
A fourth step of verifying whether the coefficient of dispersion expansion of the determined preceding factors and the factor in the verification step of the third step has a value smaller than 2;
And a fifth step of verifying whether the correlation coefficients of the determined preceding factors and the factors in the verification step of the fourth step have a value smaller than the 10% significance level threshold,
Wherein the first to fifth steps are performed by a preceding factor determiner for finding a combination of preceding factors having a high predictive value by cross-testing a defined correlation factor of a physical statistical prediction system of rainy season precipitation with a forward stepwise regression method. A Method of Physical Statistical Prediction of Precipitation. 8. The method of claim 7,
The North Pacific Development Index (A) is the average of the sea surface temperature anomalies (160 ° to 210 ° E, 20 ° to 35 ° N), followed by the first period (4.11 to 4.30) ) Is subtracted,
The North Atlantic 1 index (B) measures the second spatial mean (80 ° to 40 ° W, 30 ° to 45 ° N) in the first spatial mean of the sea surface temperature anomalies (40 ° to 15 ° W, 55 ° to 60 ° N) ) Is subtracted, and the time average (4.6 ~ 4.25) is performed.
The Central Pacific El Niño Index (C) is characterized by the time average (4.1 to 4.20) after performing the spatial mean (160 ° to 140 ° W, 15 ° S to 10 ° N) of sea surface temperature anomalies A Method of Physical Statistical Prediction of Rainfall Precipitation. The method according to claim 7, wherein the determined preceding factors are the North Pacific Development Index (A), the North Atlantic 1 Index (B), the Central Pacific El Niño Index (C)
The multiple regression model conceived as the determinant preceding factor is characterized by rainfall precipitation = 0.458A - 0.413B - 0.396C + 0.0136.

Images