Water level prediction model construction method, water level prediction model construction program, predicted water level data output method, predicted water level data output program, test data interpolation method, test data interpolation program, water level prediction hit ratio calculation method and water level prediction hit ratio calculation program

Abstract

To easily construct a water level prediction model by using a water level prediction model at a water level prediction point in a non-tidal area even in a tidal area similarly to the non-tidal area.SOLUTION: The present disclosure pays attention to that for a water level in a tidal area, an effect of a low tide and a high tide in a river-mouth region is nearly independent from an effect of a water level and precipitation in an upstream area. Then, tidal level data T is subtracted from tidal area water level data E, and the tidal area water level data E is converted into water level data obtained by removing an effect of a tide from the data E. Further, assumed water level data, upstream area water level data U and precipitation amount data R are used to construct a water level prediction model M obtained by removing the effect of the tide similar to a non-tidal area.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present disclosure relates to a technique for learning and predicting water levels at water level prediction points in tidal areas. [Background technology]

[0002] Patent Document 1 and other publications disclose technology that uses a machine learning model such as a neural network to learn and predict water levels at water level prediction points in a river area.

[0003] First, as training data required for learning the water level at a certain time at a water level prediction point, the following are extracted: prediction point water level data indicating at least one of the water levels at the water level prediction point at that time and in the past from that time; upstream water level data indicating at least one of the water levels at the water level prediction point at that time and in the past from that time from that time at a point upstream of the water level prediction point; and rainfall data indicating at least one of the rainfall at the water level prediction point and at least one of the points upstream of the water level prediction point at that time, in the past from that time, and in the future from that time. Then, a water level prediction model for predicting the water level at the water level prediction point is constructed using the prediction point water level data, the upstream water level data, and the rainfall data.

[0004] Next, the water level at the water level prediction point is predicted using the water level prediction model constructed using the training data. That is, as test data required for predicting the water level at a certain time at the water level prediction point, prediction point water level data indicating at least one of the water levels at the water level prediction point at the certain time and in the past from the certain time, upstream water level data indicating at least one of the water levels at the water level prediction point at the certain time and in the past from the certain time at a point upstream from the water level prediction point, and rainfall data indicating at least one of the rainfall at the water level prediction point and at least one of the points upstream from the water level prediction point at the certain time, in the past from the certain time, and in the future from the certain time are input to the water level prediction model. Then, the predicted water level data at the water level prediction point at the certain time is output from the water level prediction model. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2019-095240 A Summary of the Invention [Problem to be solved by the invention]

[0006] The issues with water level prediction processing of the conventional technology are shown in Figure 1. The upper part of Figure 1 shows the time variations in water level and rainfall in non-tidal areas that are not affected by low and high tides. It can be seen that the water level in non-tidal areas rises as the amount of rainfall in the non-tidal areas increases. The lower part of Figure 1 shows the time variations in water level and rainfall in tidal areas that are affected by low and high tides. It can be seen that the water level in the tidal areas repeatedly rises and falls even though the amount of rainfall in the tidal areas is not increasing.

[0007] Therefore, unlike non-tidal areas, in tidal areas, the accuracy of water level prediction cannot be improved even if the water level prediction model at the water level prediction point in non-tidal areas of Patent Document 1 is used. However, in tidal areas, like non-tidal areas, if it is acceptable to use the water level prediction model at the water level prediction point in non-tidal areas of Patent Document 1, the water level prediction model can be easily constructed.

[0008] Therefore, in order to solve the above problems, the present disclosure aims to easily construct a water level prediction model and improve the accuracy of water level prediction by using a water level prediction model at water level prediction points in non-tidal areas, just as in non-tidal areas. [Means for solving the problem]

[0009] To solve the above problem, we focused on the fact that the effects of low and high tides in the river mouth and the effects of water levels and rainfall in the upstream area on water levels in tidal areas are almost independent. We then subtracted tide level data from tidal area water level data to convert the tidal area water level data into deemed water level data with the effects of tides removed. Furthermore, we constructed a water level prediction model with the effects of tides removed in the same way as for non-tidal areas, using the deemed water level data, upstream area water level data, and rainfall data.

[0010] Specifically, the present disclosure provides a tidal area data extraction step of extracting, as training data necessary for learning the water level at a certain time at a water level prediction point in a tidal area, (1) tidal area water level data indicating at least any of the water levels at the water level prediction point at that time and in the past from that time, and (2) tide level data indicating at least any of the observed tide levels or predicted tide levels at a tide observation point or tide level prediction point downstream of the water level prediction point at that time and in the past from that time, and (3) upstream area water level data indicating at least any of the water levels at a point upstream of the water level prediction point at that time and in the past from that time, and (4) tidal area data indicating at least any of the water levels at the water level prediction point and in the past from that time, as the training data. This water level prediction model construction method comprises an upstream data extraction step of extracting at least one of the following: rainfall data indicating rainfall at at least one of a given time, at a given time, in the past, and in the future from a given time; an assumed water level data conversion step of subtracting the tidal area water level data from the tidal area water level data and converting the tidal area water level data into assumed water level data from which the influence of tides has been removed; and a water level prediction model construction step of constructing a water level prediction model for predicting the water level from which the influence of tides has been removed at the water level prediction point using the assumed water level data from which the influence of tides has been removed and at least one of the upstream water level data and the rainfall data.

[0011] With this configuration, in tidal areas, as in non-tidal areas, a water level prediction model can be easily constructed by using a water level prediction model at a water level prediction point in the non-tidal area.

[0012] The present disclosure also relates to a water level prediction model construction method, characterized in that the deemed water level data conversion step subtracts (the tide level data + data on the elevation of the observation datum level at the tide observation point) from the tidal area water level data, or subtracts (the tide level data + data on the elevation of the tide table datum level at the tide prediction point) from the tidal area water level data, thereby converting the tidal area water level data into the deemed water level data from which the influence of tides has been removed.

[0013] With this configuration, the influence of tides can be almost completely removed from the deemed water level data, taking into account the difference between the elevation of the datum level of the water level observation station and the elevation of the datum level of the tide observation station.

[0014] The present disclosure also relates to a water level prediction model construction program for causing a computer to execute each processing step of the water level prediction model construction method described above.

[0015] According to this configuration, it is possible to provide a program having the above-described effects.

[0016] To solve the above problem, we focused on the fact that the effects of low and high tides in the river mouth and the effects of water levels and rainfall in the upstream area on water levels in tidal areas are almost independent. Then, we input the deemed water level data, upstream water level data, and rainfall data into a water level prediction model, and output deemed predicted water level data from the water level prediction model that has removed the effects of tides. Furthermore, we add tide level data to the deemed predicted water level data, and convert the deemed predicted water level data into predicted water level data that takes into account the effects of tides.

[0017] Specifically, the present disclosure is a predicted water level data output method for predicting a water level taking into account the influence of tides at the water level prediction point using the water level prediction model constructed by the water level prediction model construction method described above, comprising: an deemed-forecasted water level data output step of inputting at least one of the deemed water level data, the upstream water level data, and the rainfall data into the water level prediction model as test data required for predicting the water level at a certain time at the water level prediction point, and outputting from the water level prediction model deemed-forecasted water level data from which the influence of tides at the water level prediction point has been removed; a predicted tide level data extraction step of extracting predicted tide level data indicating the predicted tide level at the certain time at the tide prediction point downstream from the water level prediction point, as the test data; and a predicted water level data conversion step of adding the predicted tide level data to the deemed-forecasted water level data and converting the deemed-forecasted water level data into predicted water level data taking into account the influence of tides.

[0018] With this configuration, the accuracy of water level prediction can be improved in tidal areas, just as in non-tidal areas, even if a water level prediction model at a water level prediction point in a non-tidal area is used.

[0019] The present disclosure also relates to a predicted water level data output method, characterized in that the predicted water level data conversion step adds (the predicted tide level data + data on the elevation of the tide table reference level at the tide level prediction point) to the deemed predicted water level data, thereby converting the deemed predicted water level data into predicted water level data that takes into account the influence of tides.

[0020] With this configuration, the influence of tides can be almost completely taken into account in the predicted water level data by taking into account the difference between the elevation of the datum level of the water level observation station and the elevation of the datum level of the tide observation station.

[0021] The present disclosure also relates to a predicted water level data output program for causing a computer to execute each processing step of the predicted water level data output method described above.

[0022] According to this configuration, it is possible to provide a program having the above-described effects.

[0023] In order to solve further issues, as test data, when tidal water level data at the water level prediction point was missing, the tidal water level data for the missing time at the water level prediction point was interpolated using predicted water level data for the missing time at the water level prediction point, which was converted to reflect rainfall and actual water level conditions at a predicted time any time before the missing time, and then interpolated using predicted water level data for the missing time at the water level prediction point, which was any time after the predicted time.

[0024] Specifically, the present disclosure is a test data interpolation method which, when using the predicted water level data output method described above, when the tidal area water level data at the water level prediction point is missing, interpolates the tidal area water level data at the missing time at the water level prediction point, and which includes a test data interpolation step in which the predicted water level data conversion step converts the tidal area water level data to a predicted time an arbitrary time before the missing time, and interpolates with the predicted water level data at the missing time after the predicted time at the water level prediction point.

[0025] According to this configuration, even when water level test data at the water level prediction point is missing, the water level prediction accuracy at the water level prediction point can be improved by interpolating the water level test data at the water level prediction point while reflecting the rainfall and actual water level conditions.

[0026] The present disclosure also relates to a test data interpolation method, characterized in that the deemed-forecast water level data output step inputs at least one of the upstream water level data and the rainfall data as the test data into the water level prediction model for also predicting the water level at the upstream point, and also outputs the deemed-forecast water level data for the relevant time at the upstream point (which is not affected by tides at the upstream point), and the test data interpolation step, when the upstream water level data at the upstream point is missing, interpolates the upstream water level data for the missing time at the upstream point using the deemed-forecast water level data for the missing time after the predicted time at the upstream point (which is not affected by tides at the upstream point) output by the deemed-forecast water level data output step at the predicted time before the missing time.

[0027] According to this configuration, even when water level test data is missing from a point upstream of the water level prediction point, the accuracy of water level prediction at the water level prediction point can be improved by interpolating the water level test data from the point upstream of the water level prediction point while reflecting the rainfall and actual water level conditions.

[0028] The present disclosure also relates to a test data interpolation method, characterized in that, when rainfall data is missing at at least one of the water level prediction point and the upstream point, the test data interpolation step interpolates the rainfall data for the missing time at that point using rainfall forecast data for the missing time at that point.

[0029] According to this configuration, even when rainfall test data is missing at the water level prediction point and an upstream point, the accuracy of water level prediction at the water level prediction point can be improved by reflecting the rainfall forecast data and interpolating the rainfall test data at the water level prediction point and an upstream point.

[0030] The present disclosure also relates to a test data interpolation method, characterized in that, when the tide data at the tide observation point is missing, the test data interpolation step interpolates the tide data at the missing time at the tide observation point using astronomical tide data (tidal forecast data) at the missing time at the tide observation point.

[0031] According to this configuration, even when tide test data at a tide observation point is missing, the accuracy of water level prediction at the water level prediction point can be improved by reflecting astronomical tide data (tidal forecast data) and interpolating the tide test data at the tide observation point.

[0032] The present disclosure also relates to a test data interpolation method, characterized in that the arbitrary time set in the test data interpolation step is a time equal to the time interval at which the predicted water level data conversion step converts into the predicted water level data.

[0033] With this configuration, the predicted water level data for the missing time a short time after (the water level prediction time interval) the predicted time at the water level prediction point can be interpolated using predicted water level data that has been converted to reflect the rainfall and actual water level conditions at the predicted time a short time before (the water level prediction time interval) the missing time.

[0034] The present disclosure also provides a test data interpolation program for causing a computer to execute each processing step of the test data interpolation method described above.

[0035] According to this configuration, it is possible to provide a program having the above-described effects.

[0036] In actual operation, tidal basin water level data, tide level data, predicted tide level data, upstream water level data, rainfall data showing "past observed" rainfall, and rainfall data showing "forecast" rainfall are extracted as test data. However, since the rainfall data showing "forecast" rainfall is less accurate than the rainfall data showing "future observed" rainfall, the "forecast" water level data may deviate significantly from the "observed" water level data.

[0037] In order to solve the above problem, in the verification stage, tidal basin water level data, tide level data, upstream water level data, rainfall data showing "past observation" rainfall, and rainfall data showing "future observation" rainfall are extracted as verification data, a probability distribution of water level prediction errors is calculated, and an error band of predicted water level data relative to observed water level data is set. Then, as verification data, rainfall data showing "future observation" rainfall is replaced with rainfall data showing "forecast" rainfall, and the proportion of predicted water level data that falls within the error band around the observed water level data is calculated as the water level prediction accuracy rate.

[0038] In the prediction stage, as test data, in actual operation, tidal basin water level data, tide level data, predicted tide level data, upstream water level data, rainfall data showing "past observation" rainfall, and rainfall data showing "forecast" rainfall are extracted, and the water level prediction accuracy rate is output together with the predicted water level data. Here, the water level prediction accuracy rate is output for each issuance standard and / or for each N-hour prediction.

[0039] Specifically, the present disclosure provides a verification data extraction step of extracting the tidal basin water level data, the tide level data, the upstream basin water level data, the rainfall data indicating past observed rainfall, and the rainfall data indicating future forecast rainfall from a verification target time as verification data for verifying the water level prediction model constructed by the water level prediction model construction method described above, subtracting the tide level data from the tidal basin water level data, and converting the tidal basin water level data into the deemed water level data from which the influence of tides has been removed; and and a water level prediction accuracy calculation step of calculating the proportion of the predicted water level data for the multiple times at the water level prediction point that falls within an error band around the observed water level data for the multiple times at the water level prediction point, among the predicted water level data for the multiple times at the water level prediction point, as the prediction accuracy of the water level at the water level prediction point.

[0040] According to this configuration, when predicting water levels in tidal areas using machine learning, even when rainfall data indicating "forecasted" rainfall is extracted as test data, the accuracy rate of water level predictions can be calculated in the verification stage, making it possible to visualize the reliability of the predicted water level data.

[0041] The present disclosure also relates to a water level prediction accuracy calculation method, characterized in that the verification data extraction step also extracts, as the verification data, the rainfall data indicating future observed rainfall beyond the verification time, and inputs, as the verification data, the deemed water level data for verifying the water levels at the multiple times at the water level prediction point, the upstream water level data, the rainfall data indicating past observed rainfall, and the rainfall data indicating future observed rainfall into the water level prediction model, calculating a probability distribution of the prediction error of the water level at the water level prediction point, and setting an error range of the predicted water level data relative to the observed water level data in the water level prediction accuracy calculation step, between the verification data extraction step and the predicted water level data output step.

[0042] According to this configuration, when predicting water levels in tidal areas using machine learning, in the verification stage, the probability distribution of water level prediction errors can be calculated, an error range of predicted water level data relative to observed water level data can be set, and the accuracy rate of water level predictions can be calculated.

[0043] The present disclosure also relates to a water level prediction accuracy calculation method, characterized in that the water level prediction accuracy calculation step calculates the prediction accuracy of the water level at the water level prediction point for each issuance criterion for water level alert at the water level prediction point and / or for each period length from the prediction output time in the predicted water level data output step to each prediction target time.

[0044] According to this configuration, when predicting water levels in tidal areas using machine learning, even when rainfall data indicating "forecasted" rainfall is extracted as test data, the accuracy rate of the water level prediction can be calculated in the verification stage to assist in determining whether each issuance standard will be reached.

[0045] The present disclosure also relates to a water level prediction accuracy calculation method, characterized in that the water level prediction accuracy calculation step outputs a prediction accuracy rate of the water level at the water level prediction point together with the predicted water level data for the predicted target time at the water level prediction point based on the tidal area water level data, the tide level data, predicted tide level data indicating predicted tide levels, the upstream area water level data, the rainfall data indicating past observed rainfall, and the rainfall data indicating forecast rainfall, when predicting the water level at the water level prediction point at the predicted target time.

[0046] According to this configuration, when predicting water levels in tidal areas using machine learning, even when rainfall data indicating "forecasted" rainfall is extracted as test data, the accuracy rate of the water level prediction can be output at the prediction stage, making it possible to visualize the reliability of the predicted water level data.

[0047] The present disclosure also provides a water level prediction accuracy calculation program for causing a computer to sequentially execute each of the processing steps included in the water level prediction accuracy calculation method described above.

[0048] According to this configuration, it is possible to provide a program having the above-described effects.

[0049] The above disclosed inventions can be combined to the greatest extent possible. Effect of the Invention

[0050] In this way, the present disclosure makes it possible to easily construct a water level prediction model and improve the accuracy of water level prediction in tidal areas as well as non-tidal areas by using a water level prediction model at water level prediction points in non-tidal areas. [Brief description of the drawings]

[0051] [Figure 1] FIG. 1 is a diagram showing a problem with the water level prediction process of the prior art. [Diagram 2] FIG. 1 is a diagram illustrating a configuration of a water level prediction device according to the present disclosure. [Diagram 3]FIG. 2 is a diagram showing the procedure of the water level prediction process of the present disclosure. [Figure 4] FIG. 1 illustrates the process of the training phase of the water level prediction model of the present disclosure. [Diagram 5] FIG. 13 is a diagram showing a specific example of the conversion process to deemed water level data disclosed herein. [Figure 6] FIG. 1 illustrates the process of the testing phase of the water level prediction model of the present disclosure. [Figure 7] A figure showing a specific example of the conversion process to predicted water level data of the present disclosure. [Figure 8] 1 is a diagram showing an example of a water level prediction process according to the prior art and the present disclosure; [Figure 9] 1 is a diagram showing an example of a water level prediction process according to the prior art and the present disclosure; [Figure 10] FIG. 13 is a diagram showing a specific example of the water level test data interpolation process of the comparative example. [Figure 11] FIG. 1 is a diagram illustrating a configuration of a water level prediction device according to the present disclosure. [Figure 12] FIG. 1 is a diagram showing a procedure of a test data interpolation process according to the present disclosure. [Figure 13] FIG. 13 illustrates a specific example of the water level test data interpolation process of the present disclosure. [Figure 14] FIG. 13 is a diagram showing a specific example of the rainfall test data interpolation process of the present disclosure. [Figure 15] FIG. 13 is a diagram showing a specific example of the tide level test data interpolation process of the present disclosure. [Figure 16] FIG. 1 is a diagram illustrating a configuration of a water level prediction device according to the present disclosure. [Figure 17] FIG. 11 is a diagram showing the procedure of the water level prediction accuracy calculation process of the present disclosure. [Figure 18] FIG. 13 is a diagram showing a specific example of the water level prediction error range setting process of the present disclosure. [Figure 19] FIG. 13 is a diagram showing a specific example of the water level prediction accuracy rate calculation process of the present disclosure. [Figure 20] FIG. 13 is a diagram showing an example of the water level prediction accuracy rate output process of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0052] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present disclosure will be described with reference to the accompanying drawings. The embodiments described below are examples of the present disclosure, and the present disclosure is not limited to the following embodiments.

[0053] (Overview of the water level prediction process of the present disclosure) The configuration of the water level prediction device of the present disclosure is shown in Fig. 2. The procedure of the water level prediction process of the present disclosure is shown in Fig. 3. The following description of the present disclosure discloses a technology for learning and predicting water levels at water level prediction points in tidal areas using a machine learning model such as a neural network.

[0054] In order to execute the water level prediction model construction method (steps S1 to S4), the water level prediction device W is equipped with a training data extraction unit 1, an assumed water level data conversion unit 2 and a water level prediction model construction unit 3, and can be realized by constructing a water level prediction model M and installing the water level prediction model construction program (steps S1 to S4) shown in Figure 3 on a computer.

[0055] The water level prediction device W is equipped with a test data extraction unit 4 and a predicted water level data conversion unit 5 in order to execute the predicted water level data output method (steps S5 to S8), and can be realized by applying the constructed water level prediction model M and installing the predicted water level data output program (steps S5 to S8) shown in Figure 3 on a computer.

[0056] In this disclosure, we have focused on the fact that the effects of low and high tides in the estuary and the effects of water levels and rainfall in the upstream area on the water level in the tidal area are almost independent. In other words, although it is believed that there is some effect of typhoons or low pressure in the estuary on low and high tides in the estuary, it is believed that there is almost no effect of rainfall, which is much smaller than the seawater in the upstream area.

[0057] Therefore, in the water level prediction model construction method (steps S1 to S4), the tidal area water level data E is converted into deemed water level data with the influence of tides removed by subtracting the tide level data T from the tidal area water level data E. Furthermore, the deemed water level data, upstream area water level data U, and rainfall data R are used to construct a water level prediction model M with the influence of tides removed in the same way as for non-tidal areas.

[0058] On the other hand, in the predicted water level data output method (steps S5 to S8), the deemed water level data, upstream water level data U, and rainfall data R are input to a water level prediction model M, and deemed predicted water level data with the influence of tides removed is output from the water level prediction model M. Furthermore, predicted tide level data A is added to the deemed predicted water level data, and the deemed predicted water level data is converted into predicted water level data P that takes into account the influence of tides.

[0059] In the following, we will first explain the training phase processing (steps S1 to S4) of the water level prediction model M of the present disclosure, then explain the testing phase processing (steps S5 to S8) of the water level prediction model M of the present disclosure, and finally explain examples of water level prediction processing of the conventional technology and the present disclosure.

[0060] (Training stage process of the water level prediction model of the present disclosure) The training stage process (steps S1 to S4) of the water level prediction model M of the present disclosure is shown in Fig. 4. A specific example of the conversion process to deemed water level data of the present disclosure is shown in Fig. 5.

[0061] The training data extraction unit 1 extracts, as training data required for learning the water level at a certain time at a water level prediction point in a tidal area, tidal area water level data E indicating the water level at the water level prediction point at that time and / or in the past from that time, and tide level data T indicating the observed tide level or predicted tide level at a tide observation point or tide level prediction point downstream from the water level prediction point at that time and / or in the past from that time (step S1).

[0062] The training data extraction unit 1 extracts, as training data necessary for learning the water level at a certain time at a water level prediction point in a tidal basin, upstream water level data U indicating at least one of the water levels at the water level prediction point at that time and in the past from that time at a point upstream of the water level prediction point, and rainfall data R indicating at least one of the rainfall at the water level prediction point at that time and in the past from that time and in the future from that time at at least one of the water level prediction point and a point upstream of the water level prediction point (step S2). However, the training data extraction unit 1 may extract only one of the upstream water level data U and the rainfall data R, or may extract both.

[0063] Here, it is desirable for the training data extraction unit 1 to extract, as training data, data on rainfall and water level for a period with high correlation compared to the water level during flooding at the water level prediction point in the tidal area, and it is desirable to discard data on rainfall and water level for a period with low correlation.

[0064] For example, the training data extraction unit 1 may select, as training data, (1) a water level peak discrimination threshold for rejecting periods of no water level fluctuation and then select an extraction period for water level data, (2) a water level peak extraction width for rejecting periods of no water level fluctuation and then select an extraction period for water level data, (3) a specified water level period for extracting periods of rising and falling water levels and then select an extraction period for water level data, (4) a specified rainfall period taking into account the time it takes rainfall to reach the water level prediction point and then select an extraction period for rainfall data, or (5) a specified accumulation period taking into account the time it takes rainfall to flow out of the soil or from a reservoir, etc. into a river and then select an extraction period for rainfall data (see Patent Document 1).

[0065] The deemed water level data conversion unit 2 subtracts the tide level data T from the tidal area water level data E to convert the tidal area water level data E into deemed water level data from which the influence of tides has been removed (step S3).

[0066] The water level prediction model construction unit 3 uses the assumed water level data with the effects of tides removed, the upstream water level data U and the rainfall data R to construct a water level prediction model M for predicting water levels with the effects of tides removed at water level prediction points in the tidal area (step S4).

[0067] In the rightmost column of Figure 4, an elevation of 0m is the elevation of the mean sea level of Tokyo Bay, etc. In the leftmost column of Figure 4 and its adjacent columns, the elevation of the observation datum of the water level observation station is 0m. In the adjacent column to the rightmost column of Figure 4, the elevation of the observation datum of the tide gauge station or the tide table datum is not necessarily 0m. Therefore, the elevation of the datum of the water level observation station and the elevation of the datum of the tide gauge station are made equal in the software. Note that while the predicted tide level only takes into account the positions of the sun, earth, and moon, and the topography of the seabed, the observed tide level is highly accurate as it also takes into account typhoons or low pressure systems in the river mouth area.

[0068] From the adjacent column in the rightmost column of Figure 4 to the adjacent column in the leftmost column, the deemed water level data conversion unit 2 subtracts (tide level data T + data on the elevation of the observation datum level at the tide observation point = tide level data after elevation correction) from the tidal area water level data E, or subtracts (tide level data T + data on the elevation of the tide table datum level at the tide prediction point = tide level data after elevation correction) from the tidal area water level data E, converting the tidal area water level data E into deemed water level data with the influence of tides removed (step S3).

[0069] In the leftmost column and adjacent columns of Figure 4, the water level prediction model construction unit 3 constructs a water level prediction model M for predicting water levels with the effects of tides removed at water level prediction points in the tidal area so that when the deemed water level data converted from the tidal area water level data E in step S3, the upstream area water level data U, and the rainfall data R are input to the water level prediction model M, deemed water level data with the effects of tides removed is output from the water level prediction model M (step S4).

[0070] Figure 5 shows the changes over time in tidal basin water level data E, tide level data T, deemed water level data, and rainfall data R. It can be seen that as the rainfall data R increases, the tidal basin water level data E also rises, and that even if the rainfall data R does not increase, the tidal basin water level data E repeatedly rises and falls. It can also be seen that the tide level data T repeatedly rises and falls almost independently of the rainfall data R. Furthermore, it can be seen that the deemed water level data has been almost completely free of the influence of tides, and only retains the influence of rainfall.

[0071] Here, the water level prediction model M can also be constructed without making the elevation of the reference surface of the water level observation station and the elevation of the reference surface of the tide gauge station equal on the software (because in the processing in the test stage of the water level prediction model M of the present disclosure, the deemed predicted water level data is reverse-converted into predicted water level data P). However, if the elevation of the reference surface of the water level observation station and the elevation of the reference surface of the tide gauge station are made equal on the software, the water level prediction model M can be constructed with high accuracy.

[0072] In this way, in tidal areas, just like in non-tidal areas, the water level prediction model M can be easily constructed by using the water level prediction model M at the water level prediction point in the non-tidal area. Then, taking into account the difference between the elevation of the datum of the water level observation station and the elevation of the datum of the tide observation station, the influence of the tides can be almost completely removed from the deemed water level data.

[0073] (Processing of the test phase of the water level prediction model of the present disclosure) The processing (steps S5 to S8) in the test stage of the water level prediction model M of the present disclosure is shown in Fig. 6. A specific example of the conversion processing to predicted water level data of the present disclosure is shown in Fig. 7.

[0074] The test data extraction unit 4 extracts, as test data required for predicting the water level at a certain time at the water level prediction point in the tidal basin, the tidal basin water level data E indicating at least any of the water levels at the water level prediction point at the certain time and in the past from the certain time, the tide level data T indicating at least any of the observed tide level or predicted tide level at the tide observation point or the tide level prediction point downstream from the water level prediction point at the certain time and in the past from the certain time, the upstream water level data U indicating at least any of the water level at the upstream point from the water level prediction point at the certain time and in the past from the certain time, and the rainfall data R indicating at least any of the rainfall at the water level prediction point or at least any of the water level prediction point and the water level prediction point upstream from the water level prediction point at the certain time, in the past from the certain time, and in the future from the certain time (step S5). However, the test data extraction unit 4 may extract only one of the upstream water level data U and the rainfall data R, or may extract both of them.

[0075] The test data extraction unit 4 extracts predicted tide level data A indicating the predicted tide level at a certain time at a tide level prediction point downstream from the water level prediction point, as test data required for predicting the water level at that time at the water level prediction point in the tidal area (step S7, the observed tide level at that time has not been observed).

[0076] Before predicting the water level taking into account the influence of tides at the water level prediction point in the tidal area, the water level prediction model M inputs the deemed water level data, upstream water level data U and rainfall data R (step S5), and outputs deemed predicted water level data from which the influence of tides at the water level prediction point at that time has been removed (step S6).

[0077] The predicted water level data conversion unit 5 adds the predicted tide level data A to the deemed predicted water level data, and converts the deemed predicted water level data into predicted water level data P that takes into account the influence of tides (step S8, the observed tide level at that time has not been observed).

[0078] In the rightmost column of Figure 6, an elevation of 0m is the elevation of the mean sea level of Tokyo Bay, etc. In the leftmost column of Figure 6 and its adjacent columns, the elevation of the observation datum level of the water level observation station is 0m. In the adjacent column to the rightmost column of Figure 6, the elevation of the tide table datum level of the tide gauge station is not necessarily 0m. Therefore, the elevation of the datum level of the water level observation station and the elevation of the datum level of the tide gauge station are made equal in the software. Note that the predicted tide level at a given time can be predicted, but the observed tide level at that time has not been observed.

[0079] In the leftmost column and adjacent columns of Figure 6, before predicting the water level taking into account the influence of tides at the water level prediction point in the tidal area, the water level prediction model M inputs deemed water level data, upstream water level data U and rainfall data R (step S5), and outputs deemed predicted water level data from which the influence of tides at the water level prediction point at that time has been removed (step S6).

[0080] From the adjacent column in the rightmost column of Figure 6 to the adjacent column in the leftmost column, the predicted water level data conversion unit 5 adds (predicted tide level data A + data on the elevation of the tide table reference level at the tide level prediction point = predicted tide level data after elevation correction) to the deemed predicted water level data, and converts the deemed predicted water level data into predicted water level data P that takes into account the effects of tides (step S8, observed tide level at that time has not been observed).

[0081] Figure 7 shows the time changes of the deemed predicted water level data, predicted tide level data A, predicted water level data P, and rainfall data R. It can be seen that although the deemed predicted water level data rises as the rainfall data R increases, the deemed predicted water level data does not repeatedly rise and fall unless the rainfall data R increases. It can also be seen that the predicted tide level data A repeatedly rises and falls almost independently of the rainfall data R. Furthermore, it can be seen that the predicted water level data P takes into account almost complete influences of rainfall and tides.

[0082] Here, the water level prediction model M can also be constructed without making the elevation of the reference surface of the water level observation station and the elevation of the reference surface of the tide gauge station equal on the software (because in the processing in the test stage of the water level prediction model M of the present disclosure, the deemed predicted water level data is reverse-converted into predicted water level data P). However, if the elevation of the reference surface of the water level observation station and the elevation of the reference surface of the tide gauge station are made equal on the software, the water level prediction model M can be constructed with high accuracy.

[0083] In this way, in tidal areas, just like in non-tidal areas, the accuracy of water level prediction can be improved even when using the water level prediction model M at the water level prediction point in the non-tidal area. Furthermore, by taking into account the difference between the elevation of the datum level of the water level observation station and the elevation of the datum level of the tide observation station, the influence of tides can be taken into account almost completely in the predicted water level data P.

[0084] (Examples of water level prediction processing in the prior art and the present disclosure) 8 and 9 show examples of water level prediction processing in the conventional technology and the present disclosure. In the conventional technology, a water level prediction model M is constructed using prediction point water level data that maintains the influence of tides, upstream water level data U, and rainfall data R. In the present disclosure, a water level prediction model M is constructed using assumed water level data that has been removed from the influence of tides, upstream water level data U, and rainfall data R.

[0085] Figure 8 shows the changes over time of actual water level data, predicted tide level data A (prediction result after 6 hours), predicted water level data P (prediction result after 6 hours), and rainfall data R when it is not raining. It can be seen that in the conventional technology, the predicted water level data P is significantly different from the actual water level data. On the other hand, in the present disclosure, it can be seen that the predicted water level data P is almost identical to the actual water level data.

[0086] Figure 9 shows the changes over time in actual water level data, predicted tide level data A (prediction result after 6 hours), predicted water level data P (prediction result after 6 hours), and rainfall data R during rainfall. It can be seen that in the conventional technology, the predicted water level data P differs greatly from the actual water level data, and, unlike the actual water level data, does not reach the flood danger level. On the other hand, in the present disclosure, the predicted water level data P is almost identical to the actual water level data, and reaches the flood danger level several hours earlier than the actual water level data, making it possible to issue evacuation guidance instructions.

[0087] (Procedure for water level test data interpolation processing in comparative example) However, it is possible that the tidal water level data E at the water level prediction point may be missing as test data. This is because a code error may be detected in the input data from the telemeter of the water level observation station, preventing normal reception. In that case, the test data cannot be input to the water level prediction model M, and the water level at the water level prediction point cannot be predicted. Therefore, it is considered to interpolate the tidal water level data E at the water level prediction point.

[0088] A specific example of the water level test data interpolation process of the comparative example is shown in Figure 10. When tidal area water level data E at the water level prediction point is missing, in order to interpolate the tidal area water level data E at the missing time at the water level prediction point, the tidal area water level data E at the time before and / or after the missing time at the water level prediction point that was transmitted by the telemeter of the water level observation station at a time before and / or after the missing time is referenced.

[0089] In the first row of Fig. 10, to predict the water level at 12:30, 30 minutes from the current time, the water level data of 3.0 m at 11:00, one hour before the current time, the water level data of 4.0 m at 11:30, 30 minutes before the current time, and the water level data of 5.0 m at the current time of 12:00 are extracted as test data. In the second to fourth rows of Fig. 10, some water level data is missing as test data.

[0090] In the second row of Figure 10, the water level data of 3.0m at 11:00 and the water level data of 4.0m at 11:30 are missing, and the water level data of 5.0m is observed at 12:00. Therefore, the water level data of 5.0m at 12:00 is time shifted to 3.0m at 11:00 and 4.0m at 11:30, resulting in a large error in the interpolation of the water level data of 5.0m.

[0091] In the third row of Figure 10, the water level data of 4.0m at 11:30 and the water level data of 5.0m at 12:00 are missing, and the water level data of 3.0m is observed at 11:00. Therefore, the water level data of 3.0m at 11:00 is time shifted to 4.0m at 11:30 and 5.0m at 12:00, resulting in a large error in the interpolation of the water level data of 3.0m.

[0092] In the fourth row of Figure 10, the water level data of 4.0m at 11:30 is missing, and the water level data of 3.0m at 11:00 and the water level data of 5.0m at 12:00 are observed. Therefore, the water level data of 4.0m at 11:30 is linearly interpolated between the water level data of 3.0m at 11:00 and the water level data of 5.0m at 12:00, and the water level data of 4.0m is almost correctly interpolated.

[0093] In the fourth row of Fig. 10, the tidal basin water level data E for both the time before and the time after the missing data is referenced, so the rainfall conditions are reflected. In the second and third rows of Fig. 10, the tidal basin water level data E for either the time before or the time after the missing data is referenced, so the rainfall conditions are not reflected.

[0094] (Water Level Test Data Interpolation Procedure of the Present Disclosure) The configuration of the water level prediction device of the present disclosure (particularly, the test data interpolation process) is also shown in Fig. 11. The procedure of the test data interpolation process of the present disclosure is shown in Fig. 12. A specific example of the water level test data interpolation process of the present disclosure is shown in Fig. 13. The water level prediction device W also includes a test data interpolation unit 6, and the test data interpolation program shown in Fig. 12 can also be installed and realized in a computer.

[0095] When performing the water level test data interpolation process (particularly, step S14) of the present disclosure, the water level prediction model M, which also predicts the water level at upstream points, inputs at least one of the upstream water level data U and the rainfall data R as test data, and also outputs assumed predicted water level data for the corresponding time at the upstream point (which is not affected by tides at upstream points).

[0096] When the tidal area water level data E at the water level prediction point is missing (step S11), the test data interpolation unit 6 interpolates the tidal area water level data E at the missing time at the water level prediction point (step S12).

[0097] Here, the test data interpolation unit 6 interpolates using the predicted water level data P for the missing time at the water level prediction point, which has been converted by the predicted water level data conversion unit 5 to a predicted time any time before the missing time (step S12).

[0098] The arbitrary time set by the test data interpolation unit 6 is equal to the time interval at which the predicted water level data conversion unit 5 converts the data into predicted water level data P. In other words, the rainfall forecast for a short time later is highly accurate, and the arbitrary time set by the test data interpolation unit 6 is short.

[0099] When the upstream water level data U at the upstream point is missing (step S13), the test data interpolation unit 6 interpolates the upstream water level data U at the time of the missing measurement at the upstream point (step S14).

[0100] Here, the test data interpolation unit 6 interpolates using the assumed predicted water level data (not affected by tides at upstream points) for a missing time any time after the predicted time at the upstream point, which is output by the water level prediction model M at a predicted time any time before the missing time (step S14).

[0101] In addition, when upstream water level data U at the upstream point is missing, even if the tidal area water level data E at the water level prediction point is not missing, the predicted water level data P at the water level prediction point cannot be output. Therefore, the upstream water level data U at the time of the missing data at the upstream point is interpolated using the assumed predicted water level data at the upstream point (which is not affected by tides at upstream points).

[0102] Moreover, the arbitrary time set by the test data interpolation unit 6 is equal to the time interval at which the water level prediction model M outputs the deemed predicted water level data. In other words, the rainfall forecast for a short time later is highly accurate, and the arbitrary time set by the test data interpolation unit 6 is short.

[0103] The test data extraction unit 4 inputs test data (including tidal area water level data E and upstream area water level data U at the missing time, which are interpolated by the test data interpolation unit 6; rainfall data R and tide level data T at the missing time are described later) to the water level prediction model M (step S19). The predicted water level data conversion unit 5 outputs predicted water level data P at the water level prediction point at that time from the water level prediction model M, and the water level prediction model M outputs assumed predicted water level data at the upstream point at that time (not affected by tides at upstream points) (step S20).

[0104] In the first row of Fig. 13, to predict the water level at 12:30, 30 minutes from the current time, the water level data of 3.0 m at 11:00, one hour before the current time, the water level data of 4.0 m at 11:30, 30 minutes before the current time, and the water level data of 5.0 m at the current time of 12:00 are extracted as test data. In the third to fifth rows of Fig. 13, some water level data is missing as test data.

[0105] In the second stage of Fig. 13, as the results of predicting the water level 30 minutes after the current time at each time, predicted water level data of 2.9 m at 11:00, one hour before the current time, predicted water level data of 3.9 m at 11:30, 30 minutes before the current time, and predicted water level data of 4.8 m at 12:00, the current time, are output. In the second stage of Fig. 13, the arbitrary time set by the test data interpolation unit 6 is 30 minutes.

[0106] In the third row of Figure 13, the water level data of 3.0m at 11:00 and the water level data of 4.0m at 11:30 are missing, and the water level data of 5.0m is observed at 12:00. Therefore, the water level data of 3.0m at 11:00 and the water level data of 4.0m at 11:30 are interpolated with the predicted water level data of 2.9m at 11:00 and the predicted water level data of 3.9m at 11:30, respectively, and the water level data of 2.9m and 3.9m are almost correctly interpolated.

[0107] In the fourth row of Figure 13, the water level data of 4.0m at 11:30 and the water level data of 5.0m at 12:00 are missing, and the water level data of 3.0m is observed at 11:00. Therefore, the water level data of 4.0m at 11:30 and the water level data of 5.0m at 12:00 are interpolated with the predicted water level data of 3.9m at 11:30 and the predicted water level data of 4.8m at 12:00, respectively, and the water level data of 3.9m and 4.8m are almost correctly interpolated.

[0108] In the fifth row of Figure 13, the water level data of 4.0 m at 11:30 is missing, and the water level data of 3.0 m at 11:00 and the water level data of 5.0 m at 12:00 are observed. Therefore, the water level data of 4.0 m at 11:30 is interpolated with the predicted water level data of 3.9 m at 11:30, and the water level data of 3.9 m is almost correctly interpolated.

[0109] In this way, even when water level test data at a water level prediction point is missing, the accuracy of water level prediction at the water level prediction point can be improved by interpolating the water level test data at the water level prediction point while reflecting the rainfall and actual water level conditions.

[0110] Furthermore, even when water level test data is missing from a point upstream of the water level prediction point, the accuracy of the water level prediction at the water level prediction point can be improved by interpolating the water level test data from the point upstream of the water level prediction point while reflecting the rainfall and actual water level conditions.

[0111] Furthermore, the predicted water level data for the missing time a short time after (the water level prediction time interval) the predicted time at the water level prediction point can be interpolated using predicted water level data that has been converted to reflect rainfall and actual water level conditions at the predicted time a short time before (the water level prediction time interval) the missing time.

[0112] (Procedure for the Rainfall Test Data Interpolation Process of the Present Disclosure) A specific example of the rainfall test data interpolation process of the present disclosure is shown in Fig. 14. When rainfall data R is missing at at least one of the water level prediction point and the upstream point (step S15), the test data interpolation unit 6 interpolates the rainfall data R at the missing time at at least one of the water level prediction point and the upstream point (step S16).

[0113] Here, the test data interpolation unit 6 interpolates with the rainfall forecast data RD for the missing measurement time at least at either the water level prediction point or the upstream point (step S16).

[0114] In addition, when rainfall data R at an upstream point is missing, even if the rainfall data R at the water level prediction point is not missing, the predicted water level data P at the water level prediction point cannot be output, so the rainfall data R at the time of the missing data at the upstream point is interpolated using the rainfall forecast data RD at the upstream point.

[0115] The test data extraction unit 4 inputs test data (including tidal area water level data E, upstream area water level data U, and rainfall data R at the missing time, which are interpolated by the test data interpolation unit 6; the tide level data T at the missing time will be described later) to the water level prediction model M (step S19). The predicted water level data conversion unit 5 outputs predicted water level data P at the water level prediction point at that time from the water level prediction model M, and the water level prediction model M outputs assumed predicted water level data at the upstream point at that time (not affected by tides at upstream points) (step S20).

[0116] In the first row of Fig. 14, to predict the water level at 12:30, 30 minutes from the current time, the following rainfall data are extracted as test data: 30 mm at 11:00, one hour before the current time, 40 mm at 11:30, 30 minutes before the current time, and 50 mm at 12:00, the current time. In the third to fifth rows of Fig. 14, some rainfall data is missing as test data.

[0117] In the second row of Figure 14, as rainfall forecast data RD (short-term precipitation forecast), rainfall forecast data of 29 mm at 11:00, one hour before the current time, rainfall forecast data of 39 mm at 11:30, 30 minutes before the current time, and rainfall forecast data of 48 mm at 12:00, the current time, are input.

[0118] In the third row of Figure 14, the rainfall data of 30 mm at 11:00 and 40 mm at 11:30 are missing, and the rainfall data of 50 mm is observed at 12:00. Therefore, the rainfall data of 30 mm at 11:00 and 40 mm at 11:30 are interpolated with the rainfall forecast data of 29 mm at 11:00 and 39 mm at 11:30, respectively, and the rainfall data of 29 mm and 39 mm are almost correctly interpolated.

[0119] In the fourth row of Figure 14, the rainfall data of 40 mm at 11:30 and the rainfall data of 50 mm at 12:00 are missing, and the rainfall data observed at 11:00 is 30 mm. Therefore, the rainfall data of 40 mm at 11:30 and the rainfall data of 50 mm at 12:00 are interpolated with the rainfall forecast data of 39 mm at 11:30 and the rainfall forecast data of 48 mm at 12:00, respectively, and the rainfall data of 39 mm and 48 mm are almost correctly interpolated.

[0120] In the fifth row of Figure 14, the rainfall data of 40 mm at 11:30 is missing, and the rainfall data of 30 mm at 11:00 and the rainfall data of 50 mm at 12:00 are observed. Therefore, the rainfall data of 40 mm at 11:30 is interpolated with the rainfall forecast data of 39 mm at 11:30, and the rainfall data of 39 mm is almost correctly interpolated.

[0121] In this way, even when rainfall test data is missing at the water level prediction point and upstream point, the accuracy of water level prediction at the water level prediction point can be improved by reflecting the rainfall forecast data and interpolating the rainfall test data at the water level prediction point and upstream point.

[0122] (Procedure for the tide level test data interpolation process disclosed herein) A specific example of the tide level test data interpolation process of the present disclosure is shown in Fig. 15. When tide level data T at a tide observation point is missing (step S17), the test data interpolation unit 6 interpolates the tide level data T at the missing time at the tide observation point (step S18).

[0123] Here, the test data interpolation unit 6 interpolates using astronomical tide data AD (tidal forecast data) for the missing observation time at the tide observation point (step S18).

[0124] The test data extraction unit 4 inputs the test data (including the tidal area water level data E, upstream area water level data U, rainfall data R, and tide level data T at the missing time, which are interpolated by the test data interpolation unit 6) to the water level prediction model M (step S19). The predicted water level data conversion unit 5 outputs the predicted water level data P at the water level prediction point at the relevant time from the water level prediction model M, and the water level prediction model M outputs the deemed predicted water level data at the upstream point at the relevant time (not affected by tides at upstream points) (step S20).

[0125] In the first row of Fig. 15, to predict the water level at 12:30, 30 minutes from the current time, the following tide level data are extracted as test data: 11:00, one hour before the current time (3.0 m), 4.0m, 11:30, 30 minutes before the current time, and 5.0m, the current time, 12:00. In the third to fifth rows of Fig. 15, some tide level data is missing as test data.

[0126] In the second row of Figure 15, the following astronomical tide data AD (tidal forecast data) are input: tide level prediction data of 2.9m at 11:00, one hour before the current time, tide level prediction data of 3.9m at 11:30, 30 minutes before the current time, and tide level prediction data of 4.8m at 12:00, the current time.

[0127] In the third row of Figure 15, the tide level data of 3.0m at 11:00 and 4.0m at 11:30 are missing, and the tide level data of 5.0m is observed at 12:00. Therefore, by interpolating the tide level data of 3.0m at 11:00 and 4.0m at 11:30 with the predicted tide level data of 2.9m at 11:00 and 3.9m at 11:30, the tide level data of 2.9m and 3.9m at 11:30 are almost correctly interpolated.

[0128] In the fourth row of Figure 15, the tide level data at 11:30, 4.0m, and the tide level data at 12:00, 5.0m, are missing, and the tide level data observed at 11:00 is 3.0m. Therefore, by interpolating the tide level data at 11:30, 4.0m, and the tide level data at 12:00, 5.0m, with the predicted tide level data at 11:30, 3.9m, and the predicted tide level data at 12:00, 4.8m, respectively, the tide level data of 3.9m and 4.8m are almost correctly interpolated.

[0129] In the fifth row of Figure 15, the tide level data for 11:30 of 4.0 m is missing, and the tide level data for 11:00 of 3.0 m and the tide level data for 12:00 of 5.0 m are observed. Therefore, the tide level data for 11:30 of 4.0 m is interpolated with the predicted tide level data for 11:30 of 3.9 m, and the tide level data of 3.9 m is almost correctly interpolated.

[0130] In this way, even when tide test data at a tide observation point is missing, the accuracy of water level prediction at the water level prediction point can be improved by reflecting astronomical tide data (tidal forecast data) and interpolating the tide test data at the tide observation point.

[0131] (Configuration of the water level prediction device of the present disclosure) The configuration of the water level prediction device disclosed herein is shown in Fig. 16. The procedure of the water level prediction accuracy calculation process disclosed herein is shown in Fig. 17. In addition to Figs. 2 and 11, the water level prediction device W includes a verification data extraction unit 7, a water level prediction error range setting unit 8, and a water level prediction accuracy calculation unit 9. The verification data extraction unit 7, the water level prediction error range setting unit 8, and the water level prediction accuracy calculation unit 9 can be realized by installing the water level prediction accuracy calculation program shown in Fig. 17 in a computer.

[0132] The training data extraction unit 1 (not shown in FIG. 16) extracts the tidal basin water level data E, tide level data T, upstream water level data U, rainfall data R indicating past observed rainfall, and rainfall data R indicating future observed rainfall as training data required for learning the water level at a certain time at the water level prediction point. Here, the training data extraction unit 1 may extract the tidal basin water level data E, tide level data T, upstream water level data U, rainfall data R indicating past observed rainfall, and rainfall data R indicating future observed rainfall when the water level fluctuates, or may extract the tidal basin water level data E, tide level data T, upstream water level data U, rainfall data R indicating past observed rainfall, and rainfall data R indicating future observed rainfall when the water level does not fluctuate, etc., immediately before and after the water level fluctuates. The water level prediction model construction unit 3 constructs the water level prediction model M using the training data.

[0133] The test data extraction unit 4 extracts tidal basin water level data E, tide level data T, predicted tide level data A, upstream water level data U, rainfall data R indicating past observed rainfall, and rainfall data R indicating future forecast rainfall as test data TD required for predicting the water level at a certain time at the water level prediction point. Here, in actual operation, the test data extraction unit 4 does not extract rainfall data R indicating future observed rainfall, but extracts rainfall data R indicating forecast rainfall. The predicted water level data conversion unit 5 uses a water level prediction model M to output predicted water level data P.

[0134] However, the rainfall data R indicating the "forecast" rainfall is less accurate than the rainfall data R indicating the "future observation" rainfall, so the "predicted" water level data P may deviate significantly from the "observed" water level data. Therefore, in this disclosure, the reliability of the "predicted" water level data P is visualized.

[0135] In the verification stage, tidal basin water level data E, tide level data T, upstream water level data U, rainfall data R indicating "past observed" rainfall, and rainfall data R indicating "future observed" rainfall are extracted as verification data VD, the water level prediction error probability distribution D is calculated, and the error band of predicted water level data P relative to the observed water level data is set. Then, as verification data VD, rainfall data R indicating "future observed" rainfall is replaced with rainfall data R indicating "forecast" rainfall, and the proportion of predicted water level data P that falls within the error band around the observed water level data is calculated as the water level prediction accuracy rate H.

[0136] In the prediction stage, in actual operation, tidal area water level data E, tide level data T, predicted tide level data A, upstream water level data U, rainfall data R indicating "past observed" rainfall, and rainfall data R indicating "forecast" rainfall are extracted as test data TD, and the water level prediction accuracy rate H is output along with the predicted water level data P. Here, the water level prediction accuracy rate H is output for each issuance standard and / or for each prediction N hours later.

[0137] Here, the training data is known data for water level learning, and the water level prediction accuracy rate H is nearly 100%, so it cannot be applied to calculate the water level prediction accuracy rate H. The test data TD is unknown data for water level prediction, but is specialized for a specific rainfall condition, so it cannot be applied to calculate the water level prediction accuracy rate H. On the other hand, the verification data VD is unknown data for verifying the water level prediction model M, and is not specialized for a specific rainfall condition, so it can be applied to calculate the water level prediction accuracy rate H.

[0138] (Specific example of water level prediction error range setting process of the present disclosure) A specific example of the water level prediction error range setting process of the present disclosure is shown in Fig. 18. The verification data extraction unit 7 extracts the tidal basin water level data E, the tide level data T, the upstream water level data U, and the rainfall data R indicating the past observed rainfall, and the rainfall data R indicating the future observed rainfall from the verification target time, as the verification data VD for verifying the water level prediction model M constructed in the water level prediction model construction process of the present disclosure (step S21). Then, the deemed water level data conversion unit 2 (not shown in Fig. 16) subtracts the tide level data T from the tidal basin water level data E to convert the tidal basin water level data E into deemed water level data (step S21). Here, the verification data extraction unit 7 does not extract the rainfall data R indicating the forecast rainfall in the water level prediction error range setting process.

[0139] The verification data extraction unit 7 inputs the deemed water level data for verifying the water level at the water level prediction point at multiple times, the upstream water level data U, the rainfall data R indicating past observed rainfall, and the rainfall data R indicating future observed rainfall as verification data VD to the water level prediction model M (step S22). Then, the water level prediction model M outputs the deemed predicted water level data at the water level prediction point at the multiple times, and the predicted water level data conversion unit 5 adds the tide level data T to the deemed predicted water level data and converts the deemed predicted water level data into predicted water level data P (step S22). Furthermore, the water level prediction error range setting unit 8 calculates the prediction error probability distribution D of the water level at the water level prediction point, and sets the error range of the predicted water level data P relative to the observed water level data in the water level prediction hit rate calculation step S25 (step S23). Here, the water level prediction error range setting unit 8 calculates a prediction error probability distribution D for each issuance criterion for water level alert at the water level prediction point and / or for each period length from the predicted output time in the predicted water level data output step S22 to each prediction target time, and sets the error range of the predicted water level data P (step S23).

[0140] In the "Prediction error N hours later" column in Fig. 18, the prediction error probability distribution D of the water level at the water level prediction point is calculated for the observed water level of 0 m or more and less than the announcement standard water level A, the observed water level of A or more and less than the announcement standard water level B, and the observed water level of B or more. In the "Error range N hours later" column in Fig. 18, the e of the prediction error probability distribution D is calculated for the observed water level of 0 m or more and less than the announcement standard water level A, the observed water level of A or more and less than the announcement standard water level B, and the observed water level of B or more. lоw-min That's all lоw-max The following and e mid-min That's all mid-max The following and e high-min That's all high-max The error range of the predicted water level data P relative to the observed water level data is set so that is within the error range. Here, the minimum and maximum values ​​of the prediction error probability distribution D are set to 0% and 100%, and 0.1%≦e high-min ≦e mid-min ≦e lоw-min ≦20% and 80%≦e lоw-max ≦e mid-max ≦e high-max ≦99.9% is satisfied. In other words, the error band of the predicted water level data P is set wider for higher water levels, narrower for lower water levels, and intermediate for medium water levels. The upper and lower ranges of the error band of the predicted water level data P may be equal or different depending on the distribution shape of the prediction error probability distribution D.

[0141] In addition, e high-min , e mid-min , e lоw-min , e lоw-max , e mid-max , e high-max is tentatively set as above, but is finally set as follows. First, the assumed water level data, upstream water level data, and rainfall data showing the analyzed rainfall (interpolated from radar with a rain gauge) are input as test data to the water level prediction model M. Next, e is set so that 80% or more of the predicted water level data for each of the high water level, low water level, and medium water level falls within the error band of the predicted water level data. high-min , e mid-min , e lоw-min , e lоw-max , e mid-max, e high-max is set.

[0142] In this way, when predicting water levels in tidal areas using machine learning, in the verification stage, the water level prediction error probability distribution D can be calculated, the error range of the predicted water level data P relative to the observed water level data can be set, and preparations can be made to calculate the water level prediction accuracy rate H.

[0143] Furthermore, when using machine learning to predict water levels in tidal areas, even when rainfall data R indicating "forecasted" rainfall is extracted as test data TD, by preparing to calculate the water level prediction accuracy rate H in the verification stage, it is possible to prepare to support the judgment of whether each issuance standard will be reached.

[0144] (Specific example of water level prediction accuracy calculation process of the present disclosure) A specific example of the water level prediction accuracy rate calculation process of the present disclosure is shown in Fig. 19. The verification data extraction unit 7 extracts the tidal basin water level data E, the tide level data T, the upstream water level data U, and the rainfall data R indicating past observed rainfall, and the rainfall data R indicating future forecast rainfall from the verification target time, as verification data VD for verifying the water level prediction model M constructed in the water level prediction model construction process of the present disclosure (step S21). Then, the deemed water level data conversion unit 2 (not shown in Fig. 16) subtracts the tide level data T from the tidal basin water level data E to convert the tidal basin water level data E into deemed water level data (step S21). Here, the verification data extraction unit 7 does not extract the rainfall data R indicating future observed rainfall in the water level prediction accuracy rate calculation process.

[0145] The verification data extraction unit 7 inputs the deemed water level data for verifying the water level at the water level prediction point at multiple times, the upstream water level data U, the rainfall data R indicating the past observed rainfall, and the rainfall data R indicating the forecast rainfall as verification data VD to the water level prediction model M (step S24). Then, the water level prediction model M outputs the deemed predicted water level data at the water level prediction point at the multiple times, and the predicted water level data conversion unit 5 adds the tide level data T to the deemed predicted water level data and converts the deemed predicted water level data into predicted water level data P (step S24). Furthermore, the water level prediction accuracy calculation unit 9 calculates the proportion of the predicted water level data P at the multiple times at the water level prediction point that falls within the error band around the observed water level data at the water level prediction point at the multiple times as the water level prediction accuracy rate H at the water level prediction point (step S25). Here, the water level prediction accuracy rate calculation unit 9 calculates the water level prediction accuracy rate H for each issuance criterion for water level alert at the water level prediction point and / or for each period length from the predicted output time in the predicted water level data output step S24 to each prediction target time (step S25).

[0146] In the "Predicted data after each N hours" column in FIG. 19, predicted water level data P for the multiple times at the water level prediction point is output for observed water levels of 0 m or more and less than the alert standard water level A, observed water levels of 0 m or more and less than the alert standard water level A and observed water levels of 0 m or more and less than the alert standard water level B, and observed water levels of 0 m or more and less than the alert standard water level A, there are 80 pieces of data that fall within the error range, 10 pieces of data that fall below the error range, and 10 pieces of data that exceed the error range. For observed water levels of 0 m or more and less than the alert standard water level A, there are 30 pieces of data that fall within the error range, 15 pieces of data that fall below the error range, and 15 pieces of data that exceed the error range. For observed water levels of 0 m or more and less than the alert standard water level B, there are 10 pieces of data that fall within the error range, 10 pieces of data that fall below the error range, and 10 pieces of data that exceed the error range.

[0147] In the "Prediction accuracy rate after each N hours" column in Figure 19, the water level prediction accuracy rate H at the water level prediction point is calculated for observed water levels 0m or more and less than the announcement standard water level A, observed water levels A or more and less than the announcement standard water level B, and observed water levels B or more. For observed water levels 0m or more and less than the announcement standard water level A, the water level prediction accuracy rate is 80 / (80+10+10)×100=80%. For observed water levels A or more and less than the announcement standard water level B, the water level prediction accuracy rate is 30 / (30+15+15)×100=50%. For observed water levels B or more, the water level prediction accuracy rate is 10 / (10+10+10)×100=33%.

[0148] Even if test data TD is accumulated, the water level prediction model M is not updated, so there is no need to update the water level prediction accuracy rate H. When training data is accumulated, the water level prediction model M may be updated, and the water level prediction accuracy rate H may also be updated.

[0149] In this way, when using machine learning to predict water levels in tidal areas, even when rainfall data R indicating "forecasted" rainfall is extracted as test data TD, the reliability of the predicted water level data P can be visualized by calculating the water level prediction accuracy rate H in the verification stage.

[0150] Furthermore, when using machine learning to predict water levels in tidal areas, even when rainfall data R indicating "forecasted" rainfall is extracted as test data TD, the water level prediction accuracy rate H can be calculated in the verification stage to assist in determining whether each issuance standard will be reached.

[0151] (Example of the water level prediction accuracy output process of the present disclosure) An example of the water level prediction accuracy rate output process of the present disclosure is shown in Fig. 20. When predicting the water level at the water level prediction point at a prediction target time based on the tidal area water level data E, tide level data T, predicted tide level data A, upstream water level data U, rainfall data R indicating past observed rainfall, and rainfall data R indicating forecast rainfall, the water level prediction accuracy rate calculation unit 9 outputs the water level prediction accuracy rate H at the water level prediction point together with the predicted water level data P at the prediction target time at the water level prediction point (step S26). Here, the water level prediction accuracy rate calculation unit 9 outputs the water level prediction accuracy rate H for each issuance criterion for water level alert at the water level prediction point and / or for each period length from the prediction output time to each prediction target time (step S26).

[0152] In Figure 20, the standard water level for issuing the warning is L 0 , L 1 , L 2 , L 3 , L 4 , L 5 In the verification stage, the standard water level for issuance L 0 ~L 1 , L 1 ~L 2 , L 2 ~L 3 , L 3 ~L 4 , L 4 ~L 5 The accuracy of water level prediction for the observed water levels is H 01 %, H 12 %, H 23 %, H 34 %, H 45 In the prediction stage, the accuracy rate of the water level prediction for the first day at around 12:00, the second day at around 14:00, the third day at around 8:00, and the fourth day at around 12:00 (six-hour prediction) is calculated as H 01 %(hit), H 23 % (miss), H 45 %(hit), H 01 The output will be % (hit).

[0153] In this way, when using machine learning to predict water levels in tidal areas, even when rainfall data R indicating "forecasted" rainfall is extracted as test data TD, the reliability of the predicted water level data P can be visualized by outputting the water level prediction accuracy rate H at the prediction stage.

[0154] Furthermore, when using machine learning to predict water levels in tidal areas, even when rainfall data R indicating "forecasted" rainfall is extracted as test data TD, the water level prediction accuracy rate H can be output at the prediction stage to assist in determining whether each issuance standard will be reached. [Industrial Applicability]

[0155] According to the water level prediction model construction method, water level prediction model construction program, predicted water level data output method, and predicted water level data output program disclosed herein, in tidal areas, as in non-tidal areas, a water level prediction model can be easily constructed by using a water level prediction model at a water level prediction point in the non-tidal area, and the water level prediction accuracy can be improved. [Explanation of symbols]

[0156] W: Water level prediction device M: Water level prediction model 1: Training data extraction part 2: Deemed water level data conversion section 3: Water level prediction model construction section 4: Test data extraction part 5: Predicted water level data conversion section 6: Test data interpolation part 7: Verification data extraction section 8: Water level prediction error range setting section 9: Water level prediction accuracy calculation section E: Tidal water level data T: Tide data U: Upstream water level data R: Rainfall data A: Predicted tide data VD: Verification data TD: Test data P: Predicted water level data D: Forecast error probability distribution H: Water level prediction accuracy rate RD: Rainfall forecast data AD: Astronomical tide data

Claims

1. a tidal area data extraction step for extracting, as training data necessary for learning the water level at a certain time at a water level prediction point in a tidal area, (1) tidal area water level data indicating at least one of the water levels at the water level prediction point at that time and in the past from that time, and (2) tide level data indicating at least one of the observed tide level or predicted tide level at a tide observation point or tide level prediction point downstream from the water level prediction point at that time and in the past from that time; an upstream data extraction step of extracting, as the training data, at least one of: (3) upstream water level data indicating at least one of the water levels at the water level prediction point at the time and in the past from the time; and (4) rainfall data indicating at least one of the rainfall at the water level prediction point and at least one of the water level prediction point and a point upstream from the water level prediction point at the time, in the past from the time, and in the future from the time; a deemed water level data conversion step of subtracting the tide level data from the tidal area water level data to convert the tidal area water level data into deemed water level data from which the influence of tides has been removed; a water level prediction model construction step of constructing a water level prediction model for predicting a water level at the water level prediction point from which the influence of tides has been removed, using the deemed water level data from which the influence of tides has been removed and at least one of the upstream water level data and the rainfall data; A water level prediction model construction method comprising:

2. The deemed water level data conversion step subtracts (the tide level data + data on the elevation of the observation datum at the tide observation point) from the tidal water level data, or subtracts (the tide level data + data on the elevation of the tide table datum at the tide level prediction point) from the tidal water level data, thereby converting the tidal water level data into the deemed water level data from which the influence of tides has been removed. The method for constructing a water level prediction model according to claim 1 .

3. A water level prediction model construction program for causing a computer to execute each processing step of the water level prediction model construction method according to claim 1 or 2.

4. When predicting a water level at the water level prediction point taking into account the influence of tides using the water level prediction model constructed by the water level prediction model construction method according to claim 1, an assumed-water-level-data output step of inputting at least one of the assumed-water-level data, the upstream water-level data, and the rainfall data into the water-level-prediction model as test data required for predicting the water level at the water-level-prediction point at a certain time, and outputting assumed-water-level data from the water-level-prediction model, the assumed-water-level data having been removed from the influence of the tides at the water-level-prediction point at that time; a predicted tide level data extraction step of extracting, as the test data, predicted tide level data indicating a predicted tide level at the tide level prediction point downstream from the water level prediction point at the relevant time; a predicted water level data conversion step of adding the predicted tide level data to the deemed predicted water level data to convert the deemed predicted water level data into predicted water level data taking into account the influence of tides; A method for outputting predicted water level data, comprising:

5. The predicted water level data conversion step adds (the predicted tide level data + data on the elevation of the tide table datum at the tide level prediction point) to the deemed predicted water level data, and converts the deemed predicted water level data into predicted water level data that takes into account the influence of tides.

5. The method for outputting predicted water level data according to claim 4.

6. A predicted water level data output program for causing a computer to execute each processing step of the predicted water level data output method according to claim 4 or 5.

7. a test data interpolation step of interpolating the tidal area water level data at the missing time at the water level prediction point when the tidal area water level data at the water level prediction point is missing, the test data interpolation step being performed using the predicted water level data at the missing time after the predicted time at the water level prediction point, the predicted water level data converted to a predicted time an arbitrary time before the missing time in the predicted water level data conversion step; 11. A method for test data interpolation comprising:

8. The deemed-forecast water level data output step inputs at least one of the upstream water level data and the rainfall data as the test data into the water level prediction model for predicting the water level at the upstream point, and also outputs the deemed-forecast water level data at the upstream point at the relevant time (which is not affected by tides at the upstream point); In the test data interpolation step, when the upstream water level data at the upstream point is missing, the upstream water level data at the missing time at the upstream point is interpolated using the deemed predicted water level data at the missing time after the arbitrary time from the predicted time at the upstream point (which is not affected by tides at the upstream point) outputted at the predicted time before the arbitrary time from the missing time by the deemed predicted water level data output step.

8. The method of claim 7, further comprising:

9. The test data interpolation step, when the rainfall data at at least one of the water level prediction point and the upstream point is missing, interpolates the rainfall data at the missing time at the point using rainfall forecast data at the missing time at the point.

9. The method of test data interpolation according to claim 7 or 8, characterized in that:

10. The test data interpolation step, when the tide level data at the tide level observation point is missing, interpolates the tide level data at the missing time at the tide level observation point using astronomical tide data (tidal forecast data) at the missing time at the tide level observation point.

9. The method of test data interpolation according to claim 7 or 8, characterized in that:

11. The arbitrary time set in the test data interpolation step is equal to the time interval for converting the predicted water level data in the predicted water level data conversion step.

9. The method of test data interpolation according to claim 7 or 8, characterized in that:

12. 9. A test data interpolation program for causing a computer to execute each processing step of the test data interpolation method according to claim 7.

13. a verification data extraction step of extracting the tidal basin water level data, the tide level data, the upstream basin water level data, the rainfall data indicating past observed rainfall, and the rainfall data indicating future forecast rainfall from a verification target time as verification data for verifying the water level prediction model constructed by the water level prediction model construction method according to claim 1, subtracting the tide level data from the tidal basin water level data, and converting the tidal basin water level data into the deemed water level data from which the influence of tides has been removed; a predicted water level data output step of inputting the deemed water level data for verifying the water levels at the water level prediction point at multiple times, the upstream water level data, the rainfall data indicating past observed rainfall, and the rainfall data indicating forecast rainfall as the verification data into the water level prediction model, outputting deemed predicted water level data from which the influence of tides at the multiple times at the water level prediction point has been removed, adding the tide level data to the deemed predicted water level data, and converting the deemed predicted water level data into predicted water level data taking into account the influence of tides; a water level prediction accuracy calculation step of calculating a ratio of the predicted water level data at the water level prediction point at the multiple times that falls within an error band around the observed water level data at the water level prediction point at the multiple times as a prediction accuracy rate of the water level at the water level prediction point; A water level prediction accuracy calculation method comprising the steps of:

14. The verification data extraction step also extracts, as the verification data, the rainfall data indicating a future observed rainfall from the verification target time, a water level prediction error range setting step of inputting the deemed water level data for verifying the water levels at the water level prediction point at the multiple times, the upstream water level data, the rainfall data indicating past observed rainfall, and the rainfall data indicating future observed rainfall as the verification data into the water level prediction model, calculating a probability distribution of the prediction error of the water level at the water level prediction point, and setting an error range of the predicted water level data relative to the observed water level data in the water level prediction hit rate calculation step; between the verification data extraction step and the predicted water level data output step. The water level prediction accuracy rate calculation method according to claim 13 .

15. The water level prediction accuracy calculation step calculates a prediction accuracy rate of the water level at the water level prediction point for each issuance criterion for water level alert at the water level prediction point and / or for each period length from the prediction output time in the predicted water level data output step to each prediction target time. The water level prediction accuracy rate calculation method according to claim 13 or 14,

16. The water level prediction accuracy calculation step outputs a prediction accuracy rate of the water level at the water level prediction point together with the predicted water level data at the prediction target time at the water level prediction point based on the tidal basin water level data, the tide level data, the predicted tide level data indicating a predicted tide level, the upstream water level data, the rainfall data indicating a past observed rainfall, and the rainfall data indicating a forecast rainfall, when predicting the water level at the water level prediction point at a prediction target time. The water level prediction accuracy rate calculation method according to claim 13 or 14,

17. A water level prediction accuracy rate calculation program for causing a computer to execute each processing step of the water level prediction accuracy rate calculation method according to claim 13 or 14 in sequence.

Images