Measuring system and flood warning method

Abstract

The invention relates to a measuring system and to a corresponding method for predicting an area response or a possible flood situation at defined target locations (x) in a catchment area (1) of a body of water (2). Therefore, at least the water level, the soil moisture and / or possible precipitation is measured by means of corresponding measuring devices (10, 10', 11, 12) at significant points in the catchment area (1). These measured values are used by means of a decentralised evaluation unit (3) to predict the area response in the form of a flow rate (l / s) and / or a water level (L) for the target locations (x) in the relevant period. To this end, an algorithm that is self-learning to the greatest extent possible and is based on artificial intelligence is implemented in the evaluation unit (3), said algorithm establishing the connection between the incoming measured values and the resulting flow rates (l / s) or water levels (L). This allows an early and high-resolution flood warning to be provided for the corresponding catchment area (1), for example by having the measuring system issue a corresponding warning signal to the responsible control centres when defined limit values are predicted to be exceeded.

Description

[0001] Measuring system and procedure for flood warning

[0002] The invention relates to a measuring system and a corresponding method by means of which a flow rate or a water level can be predicted in a catchment area of ​​a water body as central factors of a regional reaction in order to be able to issue appropriate warnings in the event of flooding in this area.

[0003] With the increasing intensity of extreme weather events, warning of flood events in the catchment area is becoming increasingly important. According to the current state of the art, the mere detection of flood events is based on water level measurements at relevant locations along watercourses, such as:

[0004] At the upper reaches of the watercourse, at the mouth of the watercourse, downstream of important tributaries with a significant increase in the size of the catchment area, at reservoirs, natural and artificial lakes with water level dynamics as well as at their inflows and outflows.

[0005] The water levels can either be read manually from structures using appropriate scales. In more modern designs, the water level can also be measured automatically, for example, using radar-based water level measurement. Radar-based measuring devices that can be used to determine water levels or fill levels are known, for example, from the publication DE 102020100867 A1.

[0006] A possible spatially resolved analysis of extreme precipitation events is described in publication DE102020119488 A1. The measurement system is based on the local measurement of precipitation amounts at multiple measuring points and subsequent decentralized evaluation using a self-learning, AI-based algorithm ("artificial intelligence"). The advantage of this approach is that a very fine resolution can be achieved.

[0007] Official flood warnings are primarily based on the provision of meteorological data, which forecast precipitation as the key decision criterion. On this basis, a reliable forecast of the area's response can be made for a future period or point in time, ranging from several hours to several days. The primary physical target variables for a regional response are the temporal course of the water level and the discharge volume at defined target locations in the area.

[0008] The minimum possible grid with which an area reaction forecast can be spatially resolved based on purely meteorological data is usually several hundred meters. Precipitation amounts can fluctuate significantly locally, especially during extreme weather events. Furthermore, any resulting flood situation depends on the possibly unknown water body situation in this forecast area, as precipitation is carried away via the water bodies or their catchment areas. The area-specific result of a precipitation event is referred to as the area reaction, with the water level and flow rate at the respective target location in the grid representing the key physical parameters of the area reaction. If the forecast water level or flow rate exceeds a limit value defined individually for the respective target location, this is defined as a flood situation.In this context, not only first-order water bodies, but also second- and third-order water bodies and their catchment areas are particularly at risk, since precipitation is mainly discharged along these catchment areas.

[0009] The invention is based on the object of being able to provide a reliable flood warning for at least one defined destination in the corresponding catchment area of ​​a water body.

[0010] The invention solves this problem by providing a measuring system for predicting a water level and / or flow rate for at least one defined target location in a watercourse's catchment area. The term "prediction" is defined in the context of the invention as the calculation of the expected value for a defined point in time, time interval, or period in the future. The measuring system according to the invention comprises the following components:

[0011] One or more measuring devices which are arranged at the destination and / or within the catchment area of ​​the body of water in order to determine as a measured value o a water level o a soil moisture, as described for example in the publication WO 2019096766 A1, o and / or precipitation, for example as an accumulated amount of precipitation or as precipitation intensity, an evaluation unit which is connected to the at least one measuring device and is designed to o forecast the flow rate and / or the water level at the destination for a defined time interval or over a defined period of time on the basis of the at least one measured value, and o based on a defined algorithm which describes a dependence of the flow rate and / or the water level at the destination of the body of water on the at least one measured value.

[0012] The measuring system according to the invention can provide flood warnings with sufficient lead time in cases of doubt. Furthermore, the warning can be issued with very fine spatial resolution. This also makes it possible to identify any locally limited warning situations that would not be recognized as such using conventional forecasts based purely on meteorological forecast data. In particular, if the water levels / flow rates and, if applicable, flow directions at the individual target locations are forecast by the measuring system over a defined period of time, a correspondingly corrected or updated forecast may also be obtained for specific points in time.

[0013] The measuring system according to the invention can be designed particularly efficiently if a machine learning algorithm is implemented in the evaluation unit, which is designed in particular as an artificial neural network or according to a deep learning method. In this case, the algorithm can be trained or optimized based on continuously measured values, and on a water level and / or flow rate measured at least once at the target location.

[0014] In order that the measuring system according to the invention requires as little infrastructure as possible and can be installed with the least possible effort, it is advantageous to design the measuring device or devices to be energy self-sufficient, for example by equipping them with suitable buffer batteries. In particular, to ensure that the battery capacity is maintained for as long as possible, the measuring rate of the measuring device or its transmission rate to the evaluation unit can be regulated proportionally or in stages depending on the measured value or the forecast flow rate or the forecast water level. This ensures that, in the event of a possible impending flood event, the respective measuring device measures or transmits the measured values ​​sufficiently often, especially when defined limit values ​​are exceeded. If the measured value of the respective measuring device or devicesIf the predicted area response (water level / flow rate) does not exceed the corresponding limit values, the measurement or transmission rate can be slowed down accordingly, e.g. by the evaluation unit.

[0015] In general, the evaluation unit within the scope of the measuring system according to the invention can be implemented using any suitable digital circuits, such as FPGAs, microcontrollers, or storage media, in conjunction with appropriate programs. Preferably, a decentralized server can function as the evaluation unit, with the at least one measuring device being connected to the server via a wireless interface, in particular a GSM interface. This ensures fast and secure transmission of measured values.

[0016] Optionally, meteorological forecast data for the catchment area of ​​the water body can also be included to optimize the water level or flow rate to be forecast at the respective target locations. For this purpose, the evaluation unit must be designed to receive this data, which can be done, for example, via the GSM interface. In this case, the algorithm must be designed to incorporate the meteorological forecast data into the calculation of the forecast flow rate or water level at the target location.

[0017] Corresponding to the measuring system according to the invention, the object underlying the invention is achieved by a corresponding measuring method. The method serves to predict a flow rate and / or a water level at at least one defined target location in a catchment area of ​​a body of water using the measuring system according to one of the previously described embodiments. The method comprises the following method steps:

[0018] Recording of at least o a water level o a soil moisture, and / or o any precipitation at the destination and / or within the catchment area of ​​the water body as the at least one measured value, and

[0019] Calculating the flow rate predicted for a defined time interval and / or the water level predicted for the time interval at the at least one target location at least o based on the at least one measured value, and o based on the algorithm.

[0020] The flood warning system according to the invention can be improved by determining the water level and / or flow rate not just for one target location, but for two or more different target locations within the catchment area. The "virtual" arrangement of the target locations by the algorithm can be grid-like in this case. Especially with a server-based design of the evaluation unit, the correspondingly increased computational effort when implementing multiple target locations does not have a negative impact.

[0021] The invention is explained in more detail with reference to the following figure. It shows:

[0022] Fig. 1 : A schematic representation of a catchment area of ​​a river section.

[0023] To provide a basic understanding of the flood forecast according to the invention, Fig. 1 shows a schematic section of a flowing water body 2 together with its catchment area 1. The term “catchment area” refers to the area in which hydrological factors influence the level or the water flow rate of the flowing water body 2 in this area 1. Accordingly, factors include, for example, water body inflows or the corresponding topography or the water absorption capacity of the soil in this area 1. In contrast to the illustration shown, the term “water body” does not only refer to flowing water bodies such as streams or rivers, but also, for example, temporarily dried-up stream courses and standing water bodies such as lakes.

[0024] For flood monitoring of the catchment area 1 under observation, the catchment area can be divided into a grid with a grid spacing of, for example, 10x10m and 100x100m, as schematically shown in Fig. 1. Each of the grid points represents a potential target location x, which can be used to describe a response within the area. This means that for each of these target locations x, a potential water level L or a potential flow rate l / s, and possibly a discharge direction, are forecast depending on various input variables.

[0025] In the representation shown in Fig. 1, for reasons of clarity, only one of the grid points is defined as the destination x. Contrary to this simplified representation, it is probable within the scope of the invention that at least some of the grid-arranged destination x are located outside the flowing water 2. In these cases or at such destination x, the water level L and the flow rate l / s in the absence of a flood situation are logically 0 L and 0 L / sec, respectively. If, due to a specific precipitation event, a water level L and / or a flow rate l / s is forecast for one of the destination x that is above a limit value L defined individually for this destination m ax, I / Smax, this is equivalent to a flood warning situation.

[0026] In order to forecast a corresponding regional reaction or a possible flood situation in response to a precipitation event in the catchment area 1, at least the following measured values ​​are recorded according to the invention at suitable measuring locations: o the current water level o the current soil moisture, and / or o precipitation, such as the amount of precipitation that has fallen to date or a precipitation intensity. In the schematic representation in Fig. 1, a measuring system according to the invention for recording the corresponding measured values ​​within the catchment area 1 therefore comprises the following measuring device types: a level measuring device 10 at an inlet of the flowing water body 2, a level measuring device 10' on the main course of the flowing water body 2 in the flow direction upstream of the target location x, a moisture measuring device 11 for determining the soil moisture, and a precipitation measuring device 12, which can record at least current precipitation as a binary event (yes / no).

[0027] It goes without saying that the area response, i.e., the grid density of the target locations x and the water levels L or flow rates l / s determined there, can be predicted more accurately and reliably the more measuring devices 10, 10', 11, 12 or measuring device types are used within the catchment area 1 by the measuring system. To ensure that the measuring devices 10, 10', 11, 12 can operate autonomously or wirelessly, it is advantageous to equip them each with at least a corresponding buffer battery and / or additional solar cells. This also simplifies any repositioning of the measuring devices 10, 10', 11, 12 within the catchment area 1.

[0028] The measured values ​​recorded by the measuring devices 10, 10', 11, 12 are fed as input variables to a corresponding algorithm of an evaluation unit 3 of the measuring system. In this case, the evaluation unit 3 is preferably implemented in a decentralized manner according to the embodiment shown in Fig. 1 and is not a component of one of the measuring devices 10, 10', 11, 12. This prevents the flood warning from failing even in the event of a failure of one of the measuring devices 10, 10', 11, 12. In the embodiment shown in Fig. 1, the evaluation unit 3 is implemented in the form of a decentralized server, which can be addressed, for example, via a GSM interface. Correspondingly, the measuring devices 10, 10', 11, 12 of the measuring system according to the invention also have a corresponding wireless interface in order to be able to communicate with the server-based evaluation unit 3.In this context, the transmission rate of the measuring devices 10, 10', 11, 12 to the evaluation unit 3 can be regulated by the latter depending on the currently measured value or depending on the forecast area reaction: If the current measured value and / or the forecast flow rate l / s or the forecast water level L is / are increased, for example, compared to a previously defined limit value, the transmission rate or even the measuring rate itself can also be increased. An increase can be made in stages or quasi-linearly. The advantage of this is that the measuring rate and / or the transmission rate of the individual measuring devices 10, 10', 11, 12 can be set very low by default if no warning situation is expected. This means that the operational readiness of the autonomously operating measuring devices 10, 10', 11, 12 can be significantly extended.

[0029] The algorithm implemented in the evaluation unit 3 describes the relationship between o the measured values ​​obtained from the measuring devices 10, 10', 11 , 12, and o the flow rates l / s to be predicted and / or the water levels L at the target locations x in the catchment area 1 .

[0030] It goes without saying that, depending on the design of the algorithm, not every target location x needs to be arranged on the grid. Rather, the target location x or the target locations x can also be individually distributed depending on the specific catchment area 1. Conversely, within the scope of the invention, depending on the situation in the specific catchment area 1, it is also conceivable that at least individual measuring devices (types) 10, 10', 11, 12 are arranged on the grid or at dedicated target locations x. In order for the measuring system according to the invention to be adaptable to a wide variety of bodies of water 2 or a wide variety of catchment areas 1, the algorithm must take into account the topology of the individual catchment area 1, its soil composition, and the extent of the flowing water 2 or its inlets and outlets.One way to account for this is to design the algorithm in evaluation unit 3 as a machine learning algorithm (also known as "machine learning"). For this purpose, a corresponding artificial neural network can be developed in evaluation unit 3, for example, in the form of "deep learning." The target variables to be optimized are, in particular, the accuracy and reliability of the forecast of the area response, i.e., the forecast accuracy and reliability of the water levels L and the flow rates l / s at the individual target locations x, as well as, if necessary, an optimal grid or position of the corresponding target locations x.

[0031] For such optimization, the measured values ​​determined by the measuring devices 10, 10', 11, 12 must be transmitted to the evaluation unit 3 or the algorithm, if possible continuously over at least one defined initial learning phase. On the other hand, the corresponding water levels L or flow rates l / s must be specified to the evaluation unit 3 or the machine learning algorithm, at least during this learning phase. In this case, it is not crucial within the scope of the invention per se whether the values ​​L, l / s specified to the algorithm originate from measurements carried out specifically for this purpose at corresponding target locations x, or whether suitable literature or estimated values ​​are specified. Furthermore, it is not crucial whether the machine learning algorithm is designed to learn in a "supervised" or "unsupervised" manner.The initial learning phase does not necessarily have to be time-limited, so that in this case the algorithm is continuously optimized or adapted throughout the entire operation of the measuring system.

[0032] In contrast to the embodiment shown in Fig. 1, the measuring system according to the invention can be used not only on a flowing water body 2, but also on any other water bodies whose catchment area is at risk of flooding. Overall, the flood forecast according to the invention using the measuring system offers the advantage that, on the one hand, a very timely forecast of flow rates and water levels can be provided. This means that a possible flood warning can be issued early enough in case of doubt. On the other hand, the warning can be issued with a very high spatial resolution. This also makes it possible to identify any locally very limited warning situations that would not be recognized as such using conventional forecasts based purely on meteorological forecast data.Within the scope of the invention, it is conceivable to also include such meteorological forecast data in the inventive forecast of water levels / flow rates if they are transmitted accordingly to the evaluation unit 3, or if the implemented algorithm incorporates these forecast data accordingly. In the case of a self-learning algorithm, this can also serve to improve the adaptation of the flood warning to the specific catchment area (1).

[0033] List of reference symbols

[0034] 1 catchment area

[0035] 2 Water bodies 3 Evaluation unit

[0036] 10 , 10' level meter

[0037] 11 Soil moisture meter

[0038] 12 Rain gauge

[0039] L Water level l / s Flow rate x Destination

Claims

Patent claims 1. Measuring system for predicting a water level (L) and / or a flow rate (l / s) at at least one defined target location (x) in a catchment area (2) of a body of water (3), comprising the following components: At least one measuring device (10, 10', 11, 12) that is located at the target location (x) and / or within the catchment area (1) of the water body (2) in order to determine at least a water level, soil moisture, and / or precipitation amount as a measured value; an evaluation unit (3) connected to the at least one measuring device (10, 10', 11, 12) that is designed to predict the flow rate (l / s) and / or the water level (L) at the target location (x) for at least a defined time interval based on the at least one measured value and a defined algorithm that describes a dependence of the flow rate (l / s) and / or the water level (L) at the target location (x) of the water body (2) on the at least one measured value.

2. Measuring device according to claim 1, wherein in the evaluation unit (3) a machine learning algorithm is configured to train the algorithm o on the basis of continuously measured values ​​from the measuring device (10, 10', 11 , 12), and o on the basis of a water level (L) measured at least once at the target location (x) and / or a flow rate (l / s) measured there once.

3. Measurement system according to claim 2, wherein the machine learning algorithm is designed as an artificial neural network, in particular in the form of a deep learning method.

4. Measuring system according to one of the preceding claims, wherein the evaluation unit (3) is designed to predict the flow rate (l / s) and / or the water level (L) of the body of water (2) over a defined period of time.

5. Measuring system according to one of the preceding claims, wherein the at least one measuring device (10, 10', 11 , 12) is designed to set a measuring rate or a transmission rate to the evaluation unit (3), in particular proportionally or stepwise, depending on the measured value or the predicted flow rate (l / s) or the predicted water level (L).

6. Measuring system according to one of the preceding claims, wherein the evaluation unit (3) is designed as a decentralized server, wherein the at least one measuring device (10, 10', 11 , 12) is connected to the server (3) via a wireless interface, in particular a GSM interface.

7. Measuring system according to claim 6, wherein the at least one measuring device (10, 10', 11 , 12) is designed to be energy self-sufficient, in particular battery-operated.

8. Measuring system according to at least one of the preceding claims, wherein the evaluation unit (3) is designed to receive meteorological forecast data for the catchment area (2) of the water body (2), wherein the algorithm is designed to incorporate the meteorological forecast data into the calculation of the predicted flow rate (l / s) or the predicted water level (L) at the target location (x).

9. Method for predicting a flow rate (l / s) and / or a water level (L) at at least one defined target location (x) in a catchment area (1) of a body of water (2) using the measuring system according to one of the preceding claims, comprising the following method steps: Recording of at least one water level, soil moisture, and / or precipitation amount at the target location (x) and / or within the catchment area (1) of the water body (2) as the at least one measured value, and Calculation of the predicted flow rate (l / s) and / or water level (L) for a defined time interval at at least one target location (x), at least o based on at least one measured value, and o based on the algorithm.

10. Method according to claim 9, wherein the water level (L) and / or the flow rate (l / s) is determined at at least four different target locations (x) in the catchment area (1), which are in particular arranged in a defined grid relative to each other.

Images