Patent Application
20240077646
The field of the invention is directed towards automated systems providing parametric flood event impact cover to one or more objects based on forward-looking measurements of occurrences and occurrence rates of measurable physical impacts of catastrophic events, in particular measurably impacting flood events. The measurements related to measurands depending on the physical, location-dependent event strength, location (as geographic area-based, cell-based, event strength line based, or geographic or topographic coordinate based (as latitude and longitude)), and measured time window, in particular to measurable impacts associated with the occurrence of flood events. Further, the invention is directed to automated parametric mitigation or transfer of the measured forward-looking impact to a specific object by an automated risk-transfer or risk-absorption system, where the impact e.g. is measured in units of expected damage rate, percentage or other quantifying and/or measuring units associated with the measured forward-looking impact the specific object. This invention further relates to automated methods and systems for automated location-dependent recognition of flood occurrence probabilities (denoted as flood hazards), where flood states are automatically measured or captured, and location-dependent forward-looking probability values are automatically determined, measured or generated based on the direct measuring link to the physical environment and/or measuring-based stochastically modeling. Finally, the invention relates to digital, modular platforms for automated mitigation of impacted physical damages to physical objects on a certain geographic location and future time window.
Among the most impacting, damaging, and destructive natural or geophysical disaster of the world, floods are most frequent and uncertain type. Floods endangers rives, properties, infrastructures and damage a lot of livelihoods within a short period of time.
Not all flood events have the some impact, wherein the impact may vary in strength as well as in type. In urban contexts, for example, flooding can e.g. pose a significant hazard to moving vehicles and causes traffic disruption by placing water flow in the transportation network, resulting in vehicles being swept away, injuries, and the loss of life of passengers. The remote detection of urban flooding over a large area will allow cities to develop flood maps to reduce risk during weather events. Mapping urban flood events is c challenge for three main reasons: the urban environment is highly complex with waterways and drains at submeter resolutions, the flooding will be shallow and ephemeral, and ponding means that the flooding extent will be discontinuous. Hydrologic models that are the conventional approach in flood forecasting struggle with these factors, making the application of these techniques difficult. Attempts have been made in the prior art to map urban flooding and flood risk with traditional prior art methods. However, high resolution hydrologic modelling structures may be effective at small scales (e.g., a few urban blocks) but the computational resources and highly accurate inputs required to properly model urban flooding at the community scale are not widely available with the current technology. These limiting factors exemplify the need to find technically based methods of mapping or predicting flooding that are less computationally intensive. The advantage of remote sensing is flood detection for large scale flood mopping without the need for highly accurate inputs and computationally intense processes to advance flood risk management.
While extreme flooding, especially that which falls in the 100-year event category, is quite understood, and mapped by a plurality of prior art systems, minor flooding is difficult to map and predict. This less severe flooding, known as nuisance flooding or NF, poses less of a hazard to lives and property, but can still be inconvenient or even dangerous, especially to drivers. Though drier regions such as Southern California may not experience the some extreme, spatially extensive flooding common in other more humid parts. NF remains a problem during the rainy season, especially for aging infrastructure or current systems that are not designed to handle changing climactic patterns. NF is expected to become more of a problem in the future as the climate changes and sea levels rise. Coastal areas such as Southern California are particularly vulnerable to NF. There is a need to develop new reliable techniques to detect catastrophic flooding, also covering urban flooding, which may be also used for nuisance flooding, reducing the risks associated with flooding during heavy rainfall in various context, covering urban and rural environments.
In the technical field of remote sensing, systems have been developed in flood detection using optical methods such as aerial photographs or satellite imagery, such as SAR and LiDAR (Light Detection and Ranging) systems. SAR, or Synthetic Aperture Radar, is an especially promising technique. As an active sensor, the radar can detect the Earth's surface no matter what time of day it is or what cloud conditions prevail. Some prior art systems for the detection of flooding with SAR try to combine SAR imagery measurements from COSMO-SkyMed (Agency Spaziale Italiana, Rome, Italy) and Landsat 8 OLI data (Boa Aerospace and technologies, Boulder, CO. USA) to measure map flooding along rivers. Others rely on measured RADARSAT-2 SAR images and flood stage data based on the return period for the 2011 Richelieu River flood in Canada, and even others rely on using TerroSAR-X in tandem with very high-resolution aerial imagery to measure map floodings. Until now. SAR data was considered insufficient for mapping flooding in more complex topographies and zones as urban zones due to the low resolution and shadow and layover in the complex urban environment.
The technical need for reliable and fast measuring and/or forecasting systems is also reflected by the painfully lacking reliable automated flood impact and impact response or mitigation systems. For many countries, it is hardly possible to do a technically correct flood impact occurrence rating and/or determination based on predictive forward-looking impact measures. A glance at the loss history shows that physical damages and associated losses caused by flood events are equally high or higher than those of other natural catastrophic events as earthquakes, windstorms, or other perils. For many of those other perils various prediction and/or rating and/or early warning systems based on actual measuring parameter values already exist. Large physical part of industrial facilities, industrial power and time are lost by occurring flood events having a physical impact to such objects. Additionally, with the trend of increasing risk-transfer penetration for floods, the insurance and re-insurance industry is affected ever more by flood caused physical damages and losses. To extend the early warning and flood damage rating to detailed and even facultative business, however, the threat of immense data amounts has to be coped with. This is done by completely new simulation approaches simulating allowing to extrapolate actual physical measuring parameters to future, i.e. forward-looking time windows and geographic cells.
Further, in many countries, a large number of industrial facilities and homes have a significant and measurably predictable probability (risk) for being impacted by flood events, and reasonably should be covered by flood mitigation and risk-transfer processes. However, many prior art systems do not capable to reliable hedge against the peril of flood events, inter alia, due to the prevalence of moral hazard and adverse selection phenomena, for example, in entering risk-transfers for objects most affected by the specific peril of flood. In such cases, traditional risk-transfer is not available. Whereas for other damage risks, risk-transfer systems can be based on the use of the low of large numbers to precisely determine a relatively small premium amount to large numbers of objects in order to cover the occurring damages of the small numbers of impacted objects who have suffered a loss due to the event-based impact to their objects. In flood event covers, typically the numbers of impacted objects is larger than the available number of individuals interested in protecting their property/objects from the peril using risk covers, which means that most prior art insurance systems do not provide risk-transfers to occurring flood events since the probability of operating the system in a sound profit range are regarded as being remote. Additionally, while there are risk-transfer systems that are enabled to provide primary flood risk covers for high value homes, the underwriting and provision of such mitigation processes does not account for many flood risks.
The lack of flood reliably automatable risk mitigation, risk-transfer and risk covering systems can be detrimental to the operation of many industrial localities at a certain location or detrimental to local homeowners of the geographic cell concerned. Even, in many cases, the responsible may discover only after an occurring event impacting a damage that their standard damage covers, and risk-transfers do not cover damages caused by flooding. It is to be mentioned that flooding can occur due to or in the wake of various physical natural disasters, for example occurring earthquakes, landslides, tsunamis, volcanic eruptions, or other natural disasters. Furthermore, few risk-transfer systems provide flood damage coverage due to the hazard of flood typically being confined to a few areas. As a result, it is an unacceptable risk due to the inability to spread the risk to a wide enough group of objects in order to technically absorb the potential catastrophic nature of the hazard.
In summary, natural disasters such as floods cause severe damage in various parts of the world. The occurrence of most of such disaster events is difficult, if not impossible, to predict over the long term by prior art measuring systems. Conventional flood mitigation techniques determine, assess and estimate pricing of flood risks using parametric risk-transfer structures so as to mitigate flood risks by gathering information related to base flood elevation data, flood depth by using mitigation devices and survey information. In addition, digital marketplace techniques list online policies that guarantee a pre-agreed payout based on pre-determined parameters in case of a flood hazard. The digital marketplace techniques determine a water-elevation function factoring in high water probability data, leading to saving both risk-transfer providers' and customers' time and cost spent on building, inspection, and damage estimation. Additional mechanisms for assessing flood risks use measured flood levels through detection chambers that assess floods based on a determined flood level measured by air-based pressure measuring devices and/or remote sensing techniques, as e.g. satellite based determination of floods. Further, the measured flood levels are used to automate signaling and triggering of cover for a physical impact, typically measured as loss or damage at the impacted object, for example by a flood damage cover payout.
Other available risk transfer mechanisms capture event data of occurred flood and maps data to a digital mop along with data related to risk transfer of portfolio. The portfolio is mapped by geographic area and value so that an exposure and risk to the portfolio induced by a flood event is determined. Yet another mechanism for assessing flood risks by simulating movement of water in case of floods is to determine potential for water damage to surface of a structure. This mechanism provides automated expert advices relating to whether to apply flood risk-transfer/insurance for risk-exposed structure based on simulation of water movement in relation to the structure.
Parametric Insurance/cover is a type of risk-transfer that is provided to the individual based on one or more pre-agreed measure (the measuring or triggering “porometer” or “index”). The parametric risk cover is levelled by an adjustable degree or threshold value of a predicted risk measure based on, for example, geographic location, exposure threshold etc. Such values can e.g. be generated using prior art hazard mapping applications such as CatNet. The exemplary, proprietary hazard mapping application CatNet is a global, web-based natural hazard analysis and mapping software-based engine, which enables the user to assess and visualize natural hazard exposure for a certain location in the world. Such prior art data-processing engines, as data-link to the application, typically comprise Application Programming Interface (API) providing access to the proprietary risk modeling structures (as e.g. Cot Server API) or any probabilistic flood modeling techniques.
Based on the degree of risk accessed, a parametric expertise may be determined to tailor a product design. This may involve introducing index definition including double trigger functionalities, quoting, and pricing, combining parametric and indemnity risk-transfer. Further, parametric transfer structures may be determined using natural catastrophe (NatCat) pricing tool for parametric damage and damage exposure cover. In addition, based on the inventive parametric structure, a modular computer-supported platform can be provided that provides an end-to-end solution from automated front-end and underwriting to automated natural catastrophe impact covering, e.g. by automated electronic monetary transfer. The platform may be used as front-end for users, for event tracking, for automated activation or electronic triggering of electronic alarm devices e.g. based on forecasted event progressing, measured and/or forecasted event impact and/or physical damage parameters, maintaining proof of loss, generating automatic claims payment, and sustaining policy administration.
The above-disclosed mechanisms/techniques do not discuss about providing a modular system that determines, assesses, and evaluates price risks that arise due to occurrence of floods. In addition, these mechanisms/techniques foil to provide parametric risk transfer structures that facilitate to mitigate the risks associated with occurrence of the floods. There is therefore a need in the art for an improved, automatable system and method for providing a customizable impact cover structure for a flood area risk based on measurable and reliable prediction and assessment of location- and time-window-dependent physical impact damage during the occurrence of flood events.
It is one object of the present invention to provide an automated system and method for reliable forward-looking measurements and ratings of physical impacts of occurring location-dependent catastrophic events, as flood events, to a specific object, and in automated conduct and provision/generation of appropriate covers and/or hedging against the impacted damage to the specific object, it is further an object of this invention to provide a new and better automated system and method for providing a dynamic parametric cover, which does not have the above-mentioned disadvantages of the prior art. In particular, it is an object of the present invention to provide a risk-transfer cover based on an extend of a flooded area. Further it is an object, to generate a pricing measure for a cover based on the chosen limit value (payout cover) and a measured and/or forecasted exposure value (e.g. related to the amount severity and/or extend of flooding to an area). The extent of flooded area is generated by assessing an Area of Interest (AOI) using automated systems such as Synthetic Aperture Radar or Drones and by dividing the area into grid cells. The AOI can e.g. be defined client- or user-specific. Ideally the grid cells cover the entire AOI. Later, with using technology such as SAR, it can e.g. be determined how much of the AOI is flooded (in percent area or number of grid cells).
The spacing between the messed network points, and thus the size of the grid cells can e.g. be predefined and can be different for different risk-transfer structures and user to be covered. The grid size can, thus, be predefined or automatically negotiated between a user and the inventive system, or it can by automatically and/or dynamically adjusted by the system based on a desired resolution within a selected area. The resolution can also be varied by the system e.g. in dependence to the severity of the impact in a certain area. e.g. the severity of flooding. The resolution can also be increased automatically and/or dynamically at the border of the affected area to achieve a more precise quantitative measure for the actual affected or impacted area.
In particular, these aims are achieved by the invention in which by means of an optical-based measuring and/or forecasting system and method for measuring a quantitative flooding extent measure value and/or a flooding impact extent measure value for physical objects located within a topographic and/or geographic area impacted by an occurrence of a flood event, comprising (i) measuring by means of satellite-based and/or airplane-based aerial remote sensing devices of the optical-based measuring and/or forecasting system, optical imaging sensory data and transmitting said optical imaging sensory data via a data transmission link to a central ground station. (ii) capturing a geographic and/or topographic area to be covered by a predefined data structure of a flood map generator, the data structure at least comprising definable area parameters capturing geographic location and/or geographic extent of said geographic and/or topographic area, and generating a flood map by the flood map generator based on the transmitted optical imaging sensory data using the predefined data structure, (iii) generating equally spaced network points over said geographic and/or topographic area providing a meshed network of network points having a definable mesh size and covering the whole geographic and/or topographic area. (iv) aggregating, after an occurrence of a flood event, for an affected area of said geographic and/or topographic area the total number of network points within the affected area, (v) measuring, after the occurrence of the flood event, the affected area of said geographic and/or topographic area based on measuring a flooding at each network point of the meshed network within the affected area based on the flood mop, wherein network points measured as flooded are contributing to the measured affected area while network points measured as not flooded are not contributing to the measured affected area, and (vii) measuring the flooding extent measure value and/or flooding impact extent measure value of the affected area based on the network points measured as flooded to the total number of network points of the affected area.
In an embodiment variant, the method further comprises (i) providing a dynamic parametric flood impact cover for an object physically impacted by the occurrence of the flood event by using an adoptive risk-transfer structure based on the flooding extent measure value and/or the flooding impact extent measure value, (ii) generating the parametric coverage covering a possible loss associated with the occurrence of the flood event and impacting the geographic area measured by the affected area, as per the adjustable risk-transfer structure a threshold measure and is triggered by a threshold-trigger, wherein the threshold-trigger is selected from a percentage of the affected area given by the measured affected area to the geographic area; and (iii) transferring, by an electronic payment transfer module, based on the generated parametric coverage monetary pay-out parameter values by electronic payment transfer to the individual.
The network points can e.g. be measured as flooded when each network point of a specified area is flooded. The network points can e.g. be regularly spaced within the mesh network with a pre-definable spacing. The network points can further e.g. be essentially 0.005×0.005 deg mesh network points. Flood determination per grid cell can e.g. be binary at the centroid. However, it can also be determined dependent on the flood depth, i.e. more refined.
In another embodiment of the invention, the mesh network points are measured as flooded when each network point of the measured affected area is flooded. The network points can be defined as corner points of two dimensional m×n blocks. The m×n blocks are of approx. 8.7210-5 radian. The network points measured as flooded can e.g. be determined using at least a neural network approach, and the affected area is measured using on-air imaging devices.
The grid cells measured as flooded can e.g. be determined using an artificial intelligence-based or machine-learning based engine, such as a neural network data processing structure, and the affected area can e.g. be measured using on-air imaging devices. The variant with the proposed grid-cell-based floodplain mapping is an inventive remote sensing intensive process that can e.g. be implemented all over the defined area. It can e.g. comprise a Digital Elevation Modelling (DEM) structure, with the predefined or adaptive measuring accuracy based on the selected grid size. The DEM structures allow the inventive system to derive the slope and terrain characteristics of the selected area. Is there a river, and if so is there a flat expanse of plain around that river? What might be below sea level that is near the mouth of a delta? To get the highest accuracy possible for these DEMs, the inventive system can e.g. use remote sensing. In particular measuring devices for aerial or space-based photography and/or LIDAR devices and Synthetic Aperture Radar (SAR) devices to produce the required high-accuracy DEMs. Aerial imagery can be used with multiple images to generate a DEM being captured using e.g. a plane or UAS. However, the quality of aerial imagery can e.g. be affected by lighting conditions and clouds. LiDAR, which shoots pulses of light and records their response, is an active system. That means it is less affected by lighting conditions, but can measure through fight haze and clouds. Also SAR is able to measure through clouds at any time day or night, but is one of the most complicated remote sensing systems to be integrated. With the inventive DEMs, which can also be improved by other imagery sources such as satellite, or UAS imagery, features of importance can be monitored and/or detected in relation to how they relate to possible floodplains. Using remote sensing, the present system allows to classify land cover into many different areas of interest, such as low-lying plains, forests, sandy deltas, and urban areas, and then use those classifications to feed into the floodplain modeling structure. With LiDAR, the system can e.g. also automatically identify buildings and detect where they lie within the possible floodplains. i.e. assign floodplain-related measuring parameter values. These multiple inputs create a map thot shows areas where flooding is measured with a higher probability value. The automated mapping can also be used for predictive modelling, cover or flood impact mitigation, and flood preparation or automated alarm signaling for activation of automated alarm systems.
As an embodiment variant, the system is automated to electronically disperse payments or transfer monetary porometer values to achieve the desired automatic cover covering the physical damage or impact caused by the natural event, e.g. the flooding event. The payments are dispersed via an electronic payment transfer module based on the generated parametric coverage monetary pay-out parameter. In particular, it is an object of the present invention to provide an automated method and system for providing dynamic parametric cover to an individual in case of an occurrence of a flood event by using an adaptive risk-transfer structure based on physical flood event measurements.
In another embodiment of the invention, wherein a premium vale is automatically generated based on the selected payout coverage and the forecasted exposure measure, i.e. the forecasted probability measurand for a future time window based on the measured occurrence frequencies and strength. Each risk-transfer can be parameterized as an exchange of risk to cover an actual physical impact to an object for a typically monetary transfer. Each impact by a risk event can be realized by a claim equivalent value, which represents the variable cost factors of a risk-transfer system. It is to be noted that the variable cost factor is difficult to determine compared to variable cost factors in other fields. For example, in the case of a production line, the variable costs, such as row materials, are quite certain, which makes it easier to minimize it. In technical field of risk-transfer, on the other hand, variable cost factors are often measured to be a probabilistic distribution. Therefore, it is technically challenging to minimize it. In the present invention. AI/ML data processing structures are used to process the flood measuring data. In a simplified embodiment variant, also linear regression and/or generalized linear model (GLM) could be used for processing the measuring data and forecast the appropriate prediction parameter value for the future exposure measurands. However, the inventive AI-based underwriting measures risk measurands (i.e. the forecasted exposure probability values) with a much higher precision. In addition, the measuring data of the inventive system are rich compared to prior art system, as it is linked to and comprises a plurality of IoT-measuring devices. Therefore, the inventive system is able to assess the forecasted risk measurands more precisely. Uncertainty of variable cost values diminishes for the inventive system adopting the proposed technologies. Thus, the system makes it easy to determine and optimize automatically and self-sufficiently pricing values for the risk-transfer based on the measured flood parameter values. Further, the geographic area is of unvarying landscape representative of area covered by dry land and wetland.
Further, the occurrence of the flood event can e.g. be detected using loopback signaling. The herein proposed technical loopback mechanism is especially shaped for the flood detection providing an inventive loopback flood detection method and system. A source maintenance end point at the flood area sends a loopback message to the target maintenance point of the central measuring engine, wherein the loopback message comprises a flow identity corresponding to one specific transmission path in a plurality of equal transmission paths to the target maintenance point. After receiving the loopback message, the target maintenance point sends a loopback replay message to the source maintenance end point, wherein the loopback replay message comprises a flow identity corresponding to a reverse common transmission path to the source maintenance end point. After receiving the loopback replay message and detecting to be correct, the source maintenance end point returns an announce of successful loopback detection. According to the inventive loopback detection method, the working mechanism of prior art loopback detection is expanded, and loopback detection can be performed on one specific path in a plurality of equal paths, thus allowing and providing high-speed transmission and detection of occurring flood by flood measurements in real-time or quasi-real-time.
According to the present invention, these objects are achieved particularly through the features of the independent claims. Additional features and advantages will become apparent to those skilled in the art upon consideration of the following detailed description of illustrative embodiments exemplifying the best mode of carrying out the method as presently perceived.
The present disclosure will be described hereafter with reference to the attached drawings, which are given as non-limiting examples only, in which:
By of satellite-based and/or airplane-based aerial remote sensing devices 1023/1024 of an optical-based measuring and/or forecasting system 1, optical imaging sensory data 1021 are measured and said optical imaging sensory data 1021 are transmitted via a data transmission link 111 to a central ground station 10. The occurrence of the flood event 4 can e.g. be detected using loopback signaling 1022. The affected area 21 can e.g. be measured using on-air imaging devices.
To capture flood exposure or flood risk measuring data and to automatically trigger oc-ions or alarm systems, detailed flood measuring parameters values 1021 are measured on depth of flooding and/or probability of flooding and/or other flooding characteristics, transmitted to the central measuring engine 10 and stored as grid measuring datasets (see
In one embodiment variant, for the grid cell resolution, it should e.g. be taken into account when selecting the cell size 10033 for the grids 1002, that the depth and measuring grid cells 1003 have an inherent relationship to the underlying topographic data used during the development of the flood hazard delineations, which can e.g. be depicted on the flood risk rote map. The roster cell size (resolution) 10033 of all roster datasets measured should be based on the density of the ground elevation data used and the appropriate precision that can be supported by the measurements. Normally, all the gird measuring datasets can e.g. use the some grid cell size 10033. However, the cell size 10033 for the grids 1002 can e.g. be no larger than 3×3 m. This will allow for a more accurate depiction and retrieval of the measuring values from that grid dataset. In the present system 1, a two dimensional horizontal modeling structure can e.g. be applied to get the detailed and accurate map of water levels and/or flood patterns and/or potential flood exposed areas. Instead of equal sized grid cells 1003, also structured curvilinear grids 1002 can e.g. be used for the hydraulic forecast processing, where the curvilinear grid cells 1003 provide a precise forecast output signaling by allowing cell stretching along river main channels while orthogonality stays within reasonable bounds. However, the use of curvilinear grid cells can e.g. result in a high resolution in sharp inner bends since grid lines are focused on the bends. A technically not required high resolution can strongly increase computational and processing time. In a preferred embodiment variant, the measuring grid cells 1003 are chosen to be dynamically adopted by the system 1 (see
The flood measuring and trigger system 1 comprises flood detection devices and/or flood sensors 102 for measuring an occurrence of a flood event 4 by measuring floodings at the mesh network points 10081, i.e. within grid cells 1003 of the spatial grid 1002, wherein flood measuring parameters 1021 of the flood detection devices and/or flood sensors 102 are transmitted to the central measuring engine 10 and wherein, based on the transmitted flood measuring parameters 1021, grid cells 1003 measured as flooded are contributing to the area 21 measured as affected while grid cells measured as not flooded are not contributing to the area 22 measured as affected. The geographic area 2 can e.g. comprise surveyed landscape representative of area covered by dry land 24 and wetland 25. The geographic area 2 can e.g. comprise at least parts definable as automatically excluded from the dynamic parametric flood impact cover 1031. The occurrence of the flood event 4 can e.g. be detected using loopback signaling 1022 for the signaling of the flood detection devices and/or flood sensors 102. The herein proposed technical loopback mechanism is especially shaped for the flood detection providing an inventive loopback flood detection method and system. A source maintenance end point of the flood area sends a loopback message to the target maintenance point at the central measuring engine, wherein the loopback message comprises a flow identity corresponding to one specific transmission path in a plurality of equal transmission paths to the target maintenance point. After receiving the loopback message, the target maintenance point sends a loopback replay message to the source maintenance end point, wherein the loopback replay message comprises a flow identity corresponding to a reverse common transmission path to the source maintenance end point. After receiving the loopback replay message and detecting to be correct, the source maintenance end point returns an announce of successful loopback detection. According to the inventive loopback detection method, the working mechanism of prior art loopback detection is expanded, and loopback detection can be performed on one specific path in a plurality of equal paths, thus allowing and providing high-speed transmission and detection of occurring flood by flood measurements in real-time or quasi-real-time.
According to the invention, the above-mentioned objects related to an space-borne sensory satellite and/or airborne sensory and drone-based flood measuring system and method for precise measuring of flood elevations and/or precise forecasting of quantitative flooding elevation measure values and/or flood impact measures on objects within a selected topographic and/or geographic area impacted by an occurrence of a flood event, are achieved particularly by (i) capturing available location data comprising aerial and/or spaceborne optical measuring data and/or surveying measurement data for the selected area, (ii) by measuring, by means of satellite-based and/or drone-based remote sensing devices, digital imagery sensory data and transmitting said imagery sensory data via a data transmission link to a central ground station. (iii) leveraging selectively and directedly the captured location data by the satellite and/or drone sensory measurements to leveraged measuring data and generate location-dependent elevation measurands based on the leveraged measuring data, and (iv) capturing, by a predefined data structure of a flood map generator, a geographic and/or topographic area to be covered, the data structure at least comprising definable area parameters capturing geographic location and/or geographic extent of said geographic and/or topographic area, and generating, by the flood map generator, a flood map with the measured elevation measurements based on the transmitted drone sensory data and the leveraged measuring data using the predefined data structure.
This has, inter alia, the advantage that the inventive system allows for a quasi-real-time flood assessment and satellite-sensory and/or drone-sensory based precise measurement. The measuring system can be realized as a web GIS digital platform allowing users to navigate areas of interest and assess flood risks with the respect to return periods, rainfall data and inundation heights. In addition, the measuring system can e.g. act as an expert system providing precise measurements for loss prevention recommendations. Further, the measuring system is able to provide precise “what if scenario simulations and forecasts”, e.g. by using machine-learning-based forward looking structures. The input parameters of the machine-learning-based forward looking structure can e.g. be varied to detect and identify areas where additional drone-based sensory measurements, e.g. regarding spatial resolution of the measuring data and/or precision of the elevation measurements etc., increase the measuring precision of the system. Such areas for additional drone-based sensory measurements can also be identified by user-specific selection, where in the measuring system refines and augments the measuring precision of the drone-based measurements upon request and selection. By the present measuring system, the user can e.g. be enabled to draw a structure whit a given height and is enabled to specifically simulate dedicated measuring and/or forecasting results and/or specified parts of the measurements, e.g. in respect to location, extend of the area, precision of the elevation measurements etc. In particular, the user is enabled to specifically simulate results to see how much inundation can be protected using definable and adoptable embarkments height.
For example, an embodiment variant can e.g. comprise two stages (e.g. offline and online stages) based on the drone-based elevation measurements. The system is initialized in the first stage by collecting the required airborne and/or spaceborne captured imagery. These imagery represent different flood levels as well as imagery without any flood in the areas of interest. Due to the scarcity of the imagery in unusual situations like flood disasters, the collected imagery can e.g. be preprocessed by a data augmentation module to synthetically generate new images enabling the efficient application of deep learning techniques. Then, the system can e.g. build an object detection modelling structure, for example, using R-CNN by retraining the network on the available imagery. This modelling structure is responsible for classifying and segmenting objects in the airborne and/or spaceborne captured images, including e.g. buildings and cars. Based on the information obtained from each optical imagery, the system can e.g. use an object processing module to decide whether to consider or filter out the objects within the imagery. After that, the system can e.g. build and train a water level estimation model for identifying the water level from the objects detected in each image. During the second phase, the imagery captured by aerial and/or space-based optical sensing device can e.g. be forwarded to the object detection model, which is trained in the offline stage, for identifying and locating objects in the scene. These objects are fed to the water level estimation to measurably estimate the water level in the cell or area based on the optical sensed measuring values.
It is to be noted that the present inventive measuring system 1 can e.g. in an embodiment variant also be used as flood warning system 1. Monitoring is important: While some areas are more exposed to flooding than others, situating measuring devices and flood sensors 102 near waterway or body of water can provide additional critical and precises in-loco measuring data. This additional measuring data can e.g. be used to calibrate the measuring data extracted from the air-borne or space-borne sensory measurements. The affected area 21 can e.g. be measured using gages and telemetry equipment 1025. The present inventive system 1 is based on the regular measuring of local rainfall, stream level, and streamflow data in each grid cell 1003. This can be real-time monitoring with telemetry allowing for the fastest possible response to a flood event. It is clear that a real-time flood warning system 12/121, . . . , 123 can reduce risks involved with flooding. The affected area 21 can e.g. be measured using at least partially air-based and/or space-based optical measuring devices 1023/1024 using digital image recognition processing (see
There are a brood variety of automated stream gages that can transmit stream level data via telemetry. Gages developed according to the NWS ALERT protocol are among the most common. However, it's to be noted that many other gages designed to measure precipitation and water level operate under similar principles, and this guide may be applicable to certain aspects of other systems. The gages, used herein, perform two primary tasks: sensing and communicating. The gages, as used herein, employs sensors to detect changes to measuring parameters. e.g. precipitation volume and/or water level. As an embodiment variant, the used gages may also be equipped with temperature and wind speed sensors. Some gages can also provide site-specific measuring data regarding the health or technical status of the measuring unit. For example, for the present inventive system 1, the gage can e.g. be designed to detect a particular “event”, e.g. 1 millimeter of rainwater entering the gage's lipping bucket through the top of its funnel. When the bucket tips, it pours out any water within, engaging a switch that transmits alert data and resetting the bucket. Any other sensors on the gage will also activate the alert data transmitter alter detecting said specific event. On days without rain, the gages can e.g. transmit a “no rain” report to show that the device 102 is still working. In the embodiment variant with automated flood warning 12/121,122,123 the system 1 can e.g. use radio, cellular, or satellite telemetry to communicate with an alarm host computer or alarm network. The system 1 can also specifically operate alarm signals using radio frequencies and/or satellite and/or cellular telemetry. For the present system 1, it is to be noted that while streams and rivers can be monitored for many mearing qualities and parameters that they shore with lakes, ponds and basins, streams and rivers possess one quality that differentiates them from other freshwater bodies, namely their movement. For the forecast processing by the present system, stream-flow can e.g. be an important measuring parameter that impacts many so other aspects of a river's hydrology and water quality (see
A geographic and/or topographic area 2 to be covered is captured by a predefined data structure 1005 of a flood map generator 100. The data structure 1005 of least comprises definable area parameters 10051 capturing geographic location 100511 and/or geographic extent 100523 of said geographic and/or topographic area 2, and generating a flood map 1007 by the flood map generator 100 based on the transmitted optical imaging sensory data 1021 using the predefined data structure 1005. The geographic and/or topographic area 2 can e.g. comprise unvarying landscape representative of area covered by dry land 24 and wetland 25.
Equally spaced network points 10081 over said geographic and/or topographic area 2 are generated providing a meshed network 1008 of network points 10081 having a definable mesh size 10082 and covering the whole geographic and/or topographic area 2. A gird 1002 of grid cells 1003 over geographic and/or topographic area 2 is defined by the meshed network 1008 each grid cell 1003 having a network point 10081 as a centroid and wherein the geographic and/or topographic area 2 is completely covered by the grid cells 1003 of the grid 1002. The network points 10081 can e.g. be regularly spaced or adjusted to geographic or topographic characteristics within the mesh network 1008 with a pre-definable or adjustable spacing 10083. The network points 10081 can e.g. be essentially 0.005×0.005 deg mesh network points 10081. The network points 10081 can e.g. represent the corner points of two dimensional m×n blocks. The dimensional m×n blocks can e.g. be of approx. 8.72*10-5 radian.
After a measured occurrence of a flood event 4/41, . . . , 43, for an affected area 21 of said geographic and/or topographic area 2 the total number of network points are aggregated within the affected area 21.
After the occurrence of the flood event 4, the affected area 21 of said geographic and/or topographic area 2 is measured based on measuring a flooding at each network point 10081 of the meshed network 1008 within the affected area 21 based on the flood map 1007. Network points 10081 measured as flooded 100811 are contributing to the measured affected area 21 while network points 10081 measured as not flooded 100812 are contributing to the area 22 measured as not affected. As an embodiment variant, network points 10081 are measured as flooded when each network point 10081 of a defined area is flooded, the mesh network points 10081 can measured as flooded 100811 can e.g. be determined using at least a machine-learning approach.
As an embodiment variant, the grid cells 1003 measured as flooded 10031 can e.g. be determined using at least an artificial intelligence as machine-learning approach based data-processing structure and/or a machine-learning-based data-processing structure. It is to be noted that the present inventive system generally can also be operated applying one or more types of numerical modeling structures for predictive flood parameter generation. For example, a hydrological rainfall run-off modeling structure can e.g. be used to forecast distributed river discharges. Also a one-dimensional (1D) drainage modeling structure can e.g. be applied based on the one-dimensional Saint-Venant flow forecasting to predict surcharges or drainages. Further, also a two-dimensional (2D) Saint-Venant flow forecast can e.g. be used for simulating the surface inundation, and obtain a forecasted maximum flood extents, maximum depths, and flow velocity on defined points on the surface. Furthermore, a 1D-2D coupling structure can be applied. All proposed embodiment variant relies on sensory and/or field measurements for capturing the inventive input parameters and their specific technical selection. However in a preferred embodiment variant, the system relies on a data-driven approach for establish a reliable flood forecast structure based on flood measurements. Unlike the numerical structure, the inventive data-driven, physical-based measuring and forecast system requires measuring input/output data only. The inventive data-driven structure has shown its high performance especially for the technical nonlinear flood forecast problems. In particular. ANN-based forecast structures can be applied, however the technical problem of over-fitting or under-fitting the measuring data, and insufficient length of the data sets can lead to erroneous forecast output results. To expand the data-driven forecast structures for short and long term flood forecasts, combined use of neuro-fuzzy structures and/or support vector machine (SVM) and/or support vector regression and/or artificial neural network (ANN) are also possible. More particularly, artificial neural network showed to be a reliable technique for the inventive flood prediction, as for forecasting water levels by applying ANN forecast structures to conventional hydrological modeling in flood-prone catchments. In another embodiment variant, the water level forecast results along a river uses backpropagation and/or conjugate gradient and/or cascade correlation. One possibility is to combine a Levenberg-Marquardt Backpropagation with cross-validation to prevent the under-fitting and overfitting in daily reservoir inflow forecasting.
Finally, in another preferred embodiment variant, the maximum flood inundation in a geographic area is determined by machine-learning based structure applying a backpropagation networks based on multiple inflow measuring data for a grid resolution of m×n. This ANN technique allows for g geographic area to provide high-resolution flood inundation maps from river flooding. For the prediction of maximum flood inundation, only the real-time discharges of the upstream catchments or flood level measurements are needed. The procession consists of two phases: the training phase collects a port of the measuring data from the existing database, tuning the model by changing the weights on input arcs to minimize the bias on the output layer; the recalling phase produces the new outputs for the testing inputs. The rest individuals in the training dataset are used for evaluating the behavior of the network structure. The total bias between the output of ANN and the observed values is defined as the error function. In order to reduce the error function in each iteration, the weights are modified automatically by the system 1. Herein, the learning rate is used for automatedly scaling the gradient in each iteration of the weight update. It is to be noted that the system can be sensitive to select up the correct value, since a large learning rate can miss the optimal point, while a small learning rate can slow the training process. For example, the gradient descent algorithm can be used by the system 1 to generate the update of the weights values. To speed up the convergence of the iteration, resilient backpropagation can e.g. be applied in the present case for treating the update of weight values differently depending on the derivative of the error function. For optimization, larger alternative learning rate can e.g. be set for speeding up the iterations if the error gradient remains in the some direction in neighboring time-steps and smaller alternative learning rate when approaching the optimal weights.
Due to the total number data of measuring grid cells 1003 (resolution of m×n), a single hidden layer can exceed 365 thousand elements. To technically optimize the storage requirement and the ANN structure training time, the geographic area 2 can e.g. be subdivided into 50×50 cell-squared grids 1002, each grid 1002 having its own independent ANN structure (the output layer having 1400 elements). It is to be noted that with the some inventive technical approach, instead of measuring the real-time discharges of the upstream catchments for the prediction of maximum flood inundation, rainfall measurements and/or flood level measurements can be selected as input measuring values for the machine-learning-based structure. To optimize the ANN processing technically further, clustering can be applied to the training measuring dataset. Like this, the size of the training measuring dataset can optimize and reduced while still keeping the main representative events. As such the training time can be reduced and the overfitting effects minimized. Further, to measure the technical performance of the prediction processing of maximum flood inundation by the applied ML structure, the mean squared error (MSE) of each grid can e.g. be used. It is assumed that the flood maps from the events used are the observed values. As each grid 1002 has its own independent training network, the MSE can e.g. be measured using all the grid cells 1003 in each grid 1002.
The flooding extent value 44 and/or flooding impact measure value 45 of the affected area 21 is measured based on the network points 1008 measured as flooded 10081 to the total number of network points of the geographic and/or topographic area 2.
A dynamic parametric flood impact cover 1031 for a physical object 3/31 physically impacted by the occurrence of the flood event 4 is generated by the system 1 by using an adoptive damage-cover structure 1041 based on the measured flooding extent value 44 and/or the flooding impact measure value 45. The parametric coverage 1031 is generated covering a possible loss associated with the occurrence of the flood event 4 impacting the geographic area 2 measured by the affected area 21, as per the adjustable damage-cover structure 1041 a flood threshold measure 1031 is triggered by an electronic threshold-trigger 103. The threshold-trigger is selected from a measured percentage value 23 given by the measured affected area 21 to the geographic area 2 or the measured affected network points 100811 to the total number of network points 100813. By an electronic payment transfer module based on the generated parametric coverage 1031 monetary pay-out parameter values are transferred by electronic payment transfer to an impacted physical object 3/31 and/or a risk-exposed individual associated with an impacted physical object 3/31. A premium value can e.g. be generated based on the payout coverage 1031 associated with a measurement.
In one embodiment, of reference number 2, a trigger is generated by a trigger module and the trigger is activated upon occurrence of the flood event. The trigger may provide information corresponding to such as flood extent, flood depth. The generation of measuring data related to the flood event is explained in detail in conjunction with
As is shown in
In one embodiment, the automated system) first measures an Area of Interest (AOI). The AOI is an extended flooded area of a total area cover of the region. The calculated AOI is then divided into a plurality of network points of a spatial mesh network. The network points defining the desired resolution, may be determined randomly or dynamically, for example, the network points can e.g. be regularly spaced within the mesh network with a pre-definable spacing. The network points can e.g. be essentially 0.005×0.005 deg mesh network points. In a variant, the network points may be defined representing the corner points of two-dimensional m×n blocks such as 500 m×500n (0.005×0.005 deg.) blocks of approx. 8.72*10-5 radian. Upon occurrence of the flood event 4, on affected area (AA) is determined, by calculating the number of network points that are totally flooded.
In one embodiment, the mesh network points measured as flooded are determined using, for example, a machine-learning approach, a double mesh approach, and the like. In addition, the network cells measured as not flooded do not contribute to the measured affected area of the AOI.
Thereafter, the percentage (at reference number 10) of affected area is generated with the relation of AA/AOI. The payout is based on the pre-defined trigger depending upon the percentage of affected area.
With respect to
For determining the payout trigger structure, a parametric coverage for the measured affected area may be generated by covering a possible loss associated with occurrence of the flood event and impacting a geographic area measured by the affected area. The possible loss incurred due to occurrence of the flood event may be determined as per an adjustable risk-transfer structure, a threshold measure. The possible loss incurred may be triggered by a threshold-trigger that is selected from the percentage of the affected area given by the measured affected area to the geographic area (for example, the AOI). By way of an example, the geographic area may include unvarying landscape that represents area covered by dry land and wetland.
In one exemplary embodiment, the payout may vary from 0% to 100%. As shown, the measured affected area is around 75% and a limit of around USD 10 m is determined under the payout scheme. The insured may receive around 0.75×USD 100 m or around USD 75 m.
In another embodiment, an electronic payment transfer may be made to the individual by an electronic payment transfer module. The electronic payment transfer may be done based on generated parametric coverage monetary pay-out parameter values. The electronic payment transfer to be made to the individual is limited and may be pre-defined based on prior discussion or agreement with the insured. In some scenarios, the individual may provide information related to the geographic area to be excluded for coverage under the payout scheme. The triggering the adaptive damage-cover structure 1041 for covering physical damages or loss 32 associated with the measured occurrence of the flood event 4 can e.g. comprise generating a parametric coverage 1031 based on a cover activation signaling or electronic payout function 1046, and transferring, by an electronic payment transfer module, based on the output of the cover activation signaling or electronic payout function 10460 monetary pay-out by electronic payment transfer to cover the physical damages or loss 32 associated with the measured occurrence. The cover activation signaling or electronic payout function 1046 can e.g. comprise a linear payout function 10461 and/or a stepped payout function 10462 and/or a deductibles-based payout function 10463. The cover activation signaling or electronic payout function 1046 can e.g. be selectable based on topography and/or exposure distribution. The cover activation signaling or electronic payout function 1046 can further e.g. be user-specific definable.
In yet another embodiment, the risk-assessment parameters used to determine occurrence of the flood event 4 may be accessed using a Machine Learning (ML) approach (see
While a number of features are described herein with respect to embodiments of the inventions; features described with respect to a given embodiment also may be employed in connection with other embodiments. The following description and the annexed drawings set forth certain illustrative embodiments of the inventions. These embodiments are indicative, however, of but a few of the various ways in which the principles of the inventions may be employed.
| Number | Date | Country | Kind |
|---|---|---|---|
| 070500/2021 | Nov 2021 | CH | national |
The present application is a continuation application of International Patent Application No. PCT/EP2022/080742, filed Nov. 3, 2022, which is based upon and claims the benefits of priority to Swiss Application No. 070500/2021, filed Nov. 3, 2021. The entire contents of oil of the above applications are incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/EP22/80742 | Nov 2022 | US |
| Child | 18229310 | US |
