36 results about "Nowcasting" patented technology

A system and method for combining the delivery of advertising with weather predictions that are limited in geo-graphical area and time, and hence which are much more precise but also more time sensitive than regular weather forecasts. The present invention is preferably implemented with “nowcasting”.

In one embodiment, a wind power ramp event nowcasting system includes a wind condition analyzer for detecting a wind power ramp signal; a sensor array, situated in an area relative to a wind farm, the sensor array providing data to the wind condition analyzer; a mesoscale numerical model; a neural network pattern recognizer; and a statistical forecast model, wherein the statistical model receives input from the wind condition analyzer, the mesoscale numerical model, and the neural network pattern recognizer; and the statistical forecast model outputs a time and duration for the wind power ramp event (WPRE) for the wind farm.

A system and method for combining the delivery of advertising with weather predictions that are limited in geographical area and time, and hence which are much more precise but also more time sensitive than regular weather forecasts. The present invention is preferably implemented with “nowcasting”.

The present document describes a method for generating and displaying a succession of short-term weather forecasts, also called nowcasts, in selectable time increments. A system for preparing nowcasts, called nowcaster, is used for preparing short-term forecasted weather values with a default time increment between each one of them. The method receives a chosen time increment from a user and the prepared forecasted weather values. The method comprises an aggregator that re-packages the forecasted weather values in the chosen time increments. A succession of short-term weather forecasts, which is a collection of forecasted weather values at the chosen time increment, is then outputted.

The invention discloses a highland severe convection weather short-term nowcasting and pre-warning system. A short-term nowcasting and pre-warning system is used for carrying out short-term nowcasting on the severe convection weather; a severe convection weather monitoring and forecasting system is used for monitoring the severe convection weather on the basis of meteorological observation materials; and a severe convection weather forecasting system is used for pre-warning and forecasting the severe convection weather by using the meteorological observation materials of an automatic meteorological station and the conventional meteorological observation materials. According to the highland severe convection weather short-term nowcasting and pre-warning system, a V-3Theda model of the severe convection weather is established to pre-warn the severe convection weather and distribute visual pre-warning information in time; a safety certification and authority management mechanism is provided for realizing the query, downloading, conversion and service of information under network environment in allusion to information service platforms of users with different authorities; and the forecasting level for the severe convection weather such as highland rainstorm, hailstone, snowstorm and the like is improved.

The invention relates to the technical field of information, and provides a rainfall nowcasting method and device based on deep learning. The objective of the invention is to solve the problems of insufficient attention of traditional model training on strong echo prediction and difficulty in radar echo space-time sequence prediction. According to the main scheme, the method comprises the following steps: S1, preprocessing weather radar base data; S2, dividing the preprocessed radar echo data into a training set, a verification set and a test set for deep learning network training; S3, performing space-time coding prediction network model training by using the training set, the verification set and the test set, performing radar echo extrapolation by using the trained space-time coding prediction network model to obtain predicted echo data, and comparing the predicted echo data with actual observation data; and S4, finally, carrying out radar quantitative rainfall estimation on the predicted radar echoes, and comparing the radar echoes with ground real rainfall data to carry out rainfall forecast detection.

The invention discloses a CGRU-based strong space-time characteristic radar echo proximity prediction method, which comprises the following steps: (1) obtaining a continuous radar echo image related to weather proximity prediction, preprocessing the continuous radar echo image, and constructing tensor data with unified time dimension and space dimension; (2) constructing and training a 3DCNN-CGRUnetwork training model to obtain a 3DCNN-CGRU coding prediction network model; (3) inputting the tensor data of the continuous radar echo image sequence for weather proximity prediction in the step (1) into the 3DCNN-CGRU network model to generate a weather proximity prediction result; according to the invention, the 3DCNN-CGRU network model is provided, the transmission capability of spatial-temporal features is enhanced, the spatial-temporal feature correlation of continuous radar echo images is captured and learned more effectively, and the problems that spatial-temporal information is easyto lose and the prediction accuracy is low are solved.

The invention relates to a meteorological radar echo map prediction method and device, computer equipment and a storage medium. The meteorological radar echo map prediction method comprises the steps:acquiring historical meteorological radar echo maps of a target area at multiple continuous historical moments; acquiring continuous dynamic information according to the historical meteorological radar echo diagrams of the multiple continuous historical moments, and acquiring content information according to a target historical meteorological radar echo diagram; fusing the continuous dynamic information and the content information, inputting the fused information into a generation network, and obtaining a predicted meteorological radar echo map of the target area at a plurality of continuousprediction moments, wherein the generative network is obtained by training based on a gradient penalty training mode. By adopting the method, the prediction accuracy of the meteorological radar echo map can be improved, the accuracy of strong weather proximity prediction based on the meteorological radar echo map is improved, and the training speed and model stability of the generative network areimproved.

The invention belongs to the technical field of nowcasting and early warning, and discloses a thunderstorm and gale early warning method and system, equipment and a terminal. The method comprises the steps of: preprocessing single radar data, and recognizing a potential thunderstorm and gale region; combining ground thunderstorm and gale information observed by an automatic station to establish a potential thunderstorm and gale area identification model and a thunderstorm and gale parameter inversion model; applying the models to a real-time thunderstorm and gale early warning service; and in the real-time service, within the single-radar identification potential thunderstorm and gale area at each time, carrying out one-hour extrapolation by calling a thunderstorm and gale parameter model, and then forming a thunderstorm and gale early warning product in the future one hour. According to the thunderstorm and gale early warning method, the identification technology of the dual-polarization radar is fully utilized, potential identification is carried out on the thunderstorm gale, the falling area of the potential thunderstorm gale in the next one hour is obtained through the extrapolation technology, and compared with an existing thunderstorm gale early warning method, the method has better advance and accuracy.

The invention discloses a real-time forecasting system for ocean short-term circulation. The system comprises a marine environment real-time detection subsystem and a forecasting subsystem. Local short-term change data of meteorological and marine environment parameters of an operation area can be monitored in real time; and the environmental risks of the marine operation under the circulating current are estimated through comprehensive evaluation including tidal power flow forecasting, weather forecasting, sea wave forecasting, ocean current forecasting and proximity forecasting units, and the time periods of all elements meeting the construction conditions at the same time are selected, so that the marine operation can make an operation time plan more conveniently and accurately.

The invention discloses an automatic nowcasting method of a multi-monomer convection system short-time heavy rainfall event. The method comprises the following steps of identifying and tracking convection monomers; carrying out convection monomer identification by using a multi-threshold adaptive algorithm so as to acquire a convection monomer identification result simultaneously reserving a convection monomer core and surrounding related information; tracking the convection monomer by using an optical flow algorithm, and acquiring speeds of the convection monomers; identifying multi-monomer convection system; providing position prediction of the convection monomers at a next moment according to space-time correlation among the convection monomers in the multi-monomer convection system andthe monomer speeds, calculating a superposition coefficient among the convection monomers, establishing a correlation matrix according to the superposition coefficient, and obtaining an identification result of the multi-monomer convection system by using a transfer closure clustering method; and carrying out multi-monomer convection system diagram model establishment and short-time heavy rainfall event identification. By using the method, automatic multi-monomer convection system short-time heavy rainfall event nowcasting is realized, disasters are warned in time, and economic losses and casualties are reduced.

The invention relates to a meteorological radar echo map prediction method and device, computer equipment and a storage medium. The meteorological radar echo map prediction method comprises the steps:obtaining historical meteorological radar echo maps of a target area at multiple different historical moments; and obtaining predicted meteorological radar echo maps of the target area at a pluralityof different prediction moments according to the historical meteorological radar echo maps and a target generator network, wherein the target generator network is obtained by training according to aninitial generator network and an initial discriminator network. With application of the method, the prediction accuracy of the meteorological radar echo map can be improved, and the accuracy of strong weather proximity prediction based on the meteorological radar echo map is improved.

The invention relates to a disaster weather forecasting method based on an energy generation adversarial predictor. The disaster weather forecasting method comprises the following steps: S1, acquiringa radar echo image sequence for weather short-term and short-term forecasting; and S2, inputting the radar echo image sequence into an energy-based EBGAN predictor to generate a weather short-term and short-term forecasting result, and inputting a radar echo image sequence used for carrying out disastrous weather proximity prediction into a model of the EBGAN predictor. According to the method, the image feature information in the radar echo image sequence is acquired through the model, so that the image feature information can be analyzed, the weather short-term and short-term forecast can be acquired more conveniently, and the forecast is more accurate.

The invention relates to a radar proximity prediction method based on heavy rainfall identification and numerical atmospheric mode driving. The method comprises the following steps: 1, identifying convection nuclear grid points based on phase state partition; 2, identifying the convective nuclear grid points again by adopting the gradients of the radar reflectivity in the horizontal direction andthe vertical direction or the gradients of the radar reflectivity in the horizontal direction and the radial direction; 3, searching for convection nuclear grid points based on a three-dimensional region growing method until all the grid points are searched for; 4, determining all grid point sets as convection regions; 5, continuously monitoring convection nuclear grid points, superposing wind field information, and determining a rainfall falling area; step 6, implementing rainfall proximity prediction. According to the method, the recognition speed and accuracy of the severe convection area are improved, the approach prediction precision of heavy rainfall is improved, and reliable technical support is provided for sudden rainstorm and flood disaster prevention.

The invention relates to a radar data display method and device, computer equipment and a storage medium. The method comprises the steps that the computer equipment obtains at least two radar featureproducts, wherein the radar feature products are data which are obtained according to base data collected by a weather radar and used for providing weather monitoring; then, the at least two radar feature products are processed to obtain at least two feature product image layers; the at least two feature product image layers are displayed in an overlapped mode. By adopting the method, the accuracyof short-time approaching forecast can be improved.

The invention discloses a thunderstorm strong wind storm body identification method and system based on ground observation, and relates to the field of meteorology and computers. The method comprisesthe following steps: step 1, acquiring a radar puzzle, a thunderstorm strong wind storm body and related attribute information of the thunderstorm strong wind storm body output by a disastrous weathershort-time proximity forecasting and early warning service system, and acquiring ground wind speed observation information received by a ground automatic station; 2, preprocessing the radar puzzle, the related attribute information of the thunderstorm strong wind storm body and the ground wind speed observation information to obtain thunderstorm strong wind storm body data to be identified; and 3, performing fuzzy logic calculation processing on the thunderstorm strong wind storm body data to be identified to obtain a matching result of the thunderstorm strong wind storm body and strong windobserved on the ground in the ground wind speed observation information, and completing identification. The problem of reasonable matching of strong wind observed by a ground station and a storm bodyobserved by a Doppler radar can be solved.

The invention relates to a meteorological proximity prediction method and device, computer equipment and a storage medium. The computer equipment acquires a radar echo map of a target area; then, inputting the radar echo map into a proximity prediction model to obtain meteorological proximity prediction data of the target area; wherein the proximity prediction model is obtained by carrying out migration training on a base model according to a region sample set of a target region; wherein the base model is obtained through training by adopting a meta-learning mechanism and a universal sample set; wherein both the regional sample set and the universal sample set comprise historical radar echo maps and proximity prediction results corresponding to the historical radar echo maps. By adopting the method, the accuracy of meteorological proximity prediction can be improved, and the quick learning ability of a proximity prediction model can be improved.

The invention discloses a novel squall line system forecasting method and forecasting system and belongs to the field of squall line forecasting and disaster prevention and reduction. The invention provides a forecasting method and a corresponding forecasting system through steps of quality control, narrow band echo identification, convergence line identification, squall line wind positioning andearly warning and squall line wind forecasting. Based on quality control of radar original base data, structural features and geometrical features of narrow-band echoes of an intensity field are fullyutilized, and a bidirectional gradient algorithm is designed to identify the narrow-band echoes; an abrupt change of wind is identified through radial convergence in a velocity field; a squall line positioning algorithm is used for integrating bidirectional gradient and shear convergence to obtain a result; and an accurate position of squall line wind is determined, vector field extrapolation isperformed by combining an optical flow method with a neural network technology so that a two-hour short-time proximity prediction service is provided, a squall line weather system in a radar echo canbe identified and positioned, 0-2-hour short-time proximity forecasting can be realized, and a forecasting effect is accurate.

The invention relates to a radar proximity prediction method based on heavy rainfall identification and numerical atmospheric mode driving. The method comprises the following steps: 1, identifying convection nuclear grid points based on phase state partition; 2, identifying the convective nuclear grid points again by adopting the gradients of the radar reflectivity in the horizontal direction andthe vertical direction or the gradients of the radar reflectivity in the horizontal direction and the radial direction; 3, searching for convection nuclear grid points based on a three-dimensional region growing method until all the grid points are searched for; 4, determining all grid point sets as convection regions; 5, continuously monitoring convection nuclear grid points, superposing wind field information, and determining a rainfall falling area; step 6, implementing rainfall proximity prediction. According to the method, the recognition speed and accuracy of the severe convection area are improved, the approach prediction precision of heavy rainfall is improved, and reliable technical support is provided for sudden rainstorm and flood disaster prevention.

The invention discloses an integrated intelligent grid forecasting service system for weather forecasting. The system includes a comprehensive detection module, an intelligent grid forecasting module, a forecasting making module, a rain condition fast report making module, an inspection and evaluation module, an intelligent display module, a decision service module, a data management platform and a predictor behavior analysis module, all the modules being designed based on a system architecture. Compared with the prior art, the method has the advantages that the research and development of a short-time proximity monitoring and early warning technology and an intelligent grid element forecasting support technology are enhanced, the business process of real-time updating, synchronous sharing and forecasting cooperation is constructed and completed, and an integrated intelligent grid forecasting business system and an integrated short-time proximity forecasting and early warning system are preliminarily built; therefore, a seamless, precise and intelligent forecasting service system is perfected.

The invention belongs to the field of short temporary rainfall forecasting, and particularly relates to a method for improving the precision of short temporary rainfall forecasting, which comprises the following steps: collecting a continuous radar echo image for weather nowcasting, and converting the radar echo image into a tensor; performing superposition processing on the tensor through a four-layer MFI-LSTM network to obtain a network output tensor at the current moment; converting the network output tensor at the current moment into a corresponding radar echo map; acquiring short temporary rainfall forecast information from the newly acquired radar echo map; the MFI-LSTM network is composed of a multi-scale feature interaction module, a convolution LSTM gating mechanism and a memory feature interaction module. On the basis of the LSTM model, two brand-new structures of the multi-scale feature interaction module and the memory feature interaction module are added, so that interaction of feature information of input data, a hidden state and a memory unit is enhanced; the problem that the final radar echo intensity prediction is inaccurate due to prediction error accumulation in the radar echo extrapolation task of the original LSTM can be prevented.

The invention relates to a real-time scheduling method for rainwater storage facilities based on real-time meteorological information. By constructing a regional SWMM model, different plans are established according to meteorological early warning or real-time nowcasting; The upper computer software scheduling control platform, in the control center, obtains the meteorological early warning data released by the relevant national departments in real time, and uses this as a basis to call the corresponding plan; or obtains the impending rainfall forecast information in real time, and uses the genetic algorithm for real-time decision-making optimization Realize the real-time scheduling control of rainwater storage facilities based on real-time weather information; in addition, through the set of information collection components, executive components and lower computers, the implementation of platform plans and real-time data collection are realized. After accumulating a certain amount of operating data, modify the parameters of the SWMM model and re-simulate, and adjust the plan according to the simulation results to ensure the reliability of the plan.

The invention discloses a nowcasting and short-term forecasting fusion system for a wind field, relates to the crossing field of meteorology and engineering disaster prevention, and solves the problem that the existing fusion integrated forecasting method is relatively high in difficulty in realizing wind speed forecasting accuracy. The system comprises a quality control module, an error feature analysis module, a model construction module, an inspection module and a comprehensive effect evaluation module. According to the nowcasting and short-term forecasting fusion system, a model is established through a BP neural network based on a large amount of site live data, nowcasting data and short-term numerical forecasting data, and the model can simply and quickly generate fusion data. The business real-time operation effect of the fusion system in the Beijing area is verified to be capable of effectively improving the accuracy rate of the wind speed.

The invention provides a convective development identification method based on networking wind profile radars, and relates to the technical field of weather forecast, the convective development identification method based on networking wind profile radars comprises the following steps: S1, networking wind profile radars: firstly screening the wind profile radars for networking, the number of the networking wind profile radars being four or more; according to the method, atmospheric power parameters such as divergence, vorticity, wind shear and vertical speed with higher spatial-temporal resolution can be inverted by utilizing mesoscale net observation of the wind profile radar, atmospheric vertical power change characteristics before convection triggering can be more accurately captured, divergence and vorticity distribution of a real-time three-dimensional space can be obtained, and the real-time vertical power change characteristics can be obtained. According to the method, weather predictors can understand the fine structure of a weather system and quickly find a strong rising motion area, the method is used for short-time nowcasting of heavy rainfall, important reference is provided for severe convection weather monitoring and early warning and short-time nowcasting, practicability is high, and progress is remarkable.

Embodiments of the present invention provide a method and device for convective weather nowcasting based on multi-source observation data, the method comprising: acquiring preprocessed multi-source observation data of convective weather; A forecast model is used to obtain the generation and consumption evolution characteristics reflecting the real-time changes of the convective weather with time and space; according to the generation and consumption evolution characteristics and the forecast model, the nowcasting results of the convective weather are obtained. The apparatus performs the method described above. The method and device for convective weather nowcasting based on multi-source observation data provided by the embodiments of the present invention obtains the generation and consumption evolution characteristics of multi-source observation data of convective weather through the forecast model, and based on the generation and consumption evolution characteristics and the forecast model Weather nowcasting can effectively predict the evolution process of convective weather generation and consumption, so that convective weather can be accurately and effectively nowcasted.

The embodiment of the invention discloses an identification method and device for a spatial distribution area of a meteorological element field, and belongs to the technical field of weather forecast. Based on the method provided by the invention, a field space distribution area of meteorological elements such as high and low temperatures and rainfall can be quickly fitted and identified, and on the basis of time parameters and meteorological observation data, nowcasting of convective weather in key periods and key areas can be realized; and the weather forecast is obtained by deducing the evolution process of the meteorological element observation object based on a time sequence instead of a probabilistic forecast result in a regional range, so that the accuracy is high, weather change can be coped and early warning can be carried out based on an identification result, and the method can also be used for researching a mechanism influencing weather change formation.

The invention discloses a satellite retrieval method for night rainfall probability, which comprises two steps of rain area division and rainfall estimation. A rain area division method comprises the steps of establishing an RPIM (rainfall possibility identification matrix) and then obtaining rainfall probability of all kinds of combination of brightness temperature Tb1 and brightness temperature Tb4-1 in the RPIM, wherein the RPIM is established based on an IR1-BTD41 two-dimensional spectral space. Therefore, an estimated rainfall value obtained through the rain area division method and a rainfall estimation method provided by the invention can be better related to and be more slightly deviated from an actual rainfall value, i.e. the estimated rainfall value is better consistent with the actually measured value, so the satellite retrieval method for night rainfall probability has the advantage of better application to multi-spectral satellite image night rainfall retrieval and nowcasting.

The invention belongs to the field of strong wind prediction manufacturing, and particularly provides a strong convection strong wind approaching prediction method and system based on an upstream live wind speed, and the method comprises the steps: calculating an optical flow field (v) of combined reflectivity radar echoes under different time dimensions through an optical flow method based on continuous multi-scene radar observation; radar echo observation of 6 min is mapped to radar observation of 10 min, optical flow fields of different time dimensions are fused through dynamic weights, and finally radar echo extrapolation of the next 2 h and 10 min interval is obtained. And through an XGBOOST-CNN algorithm, real-time radar observation at a 10-min interval after mapping and ground 10-min-level strong wind real-time station monitoring are carried out, topographic information and underlying surface information are fused, a small sample echo feature-wind speed inversion model based on a machine learning technology is established, and a real-time wind speed inversion model is obtained. The invention innovatively provides a lattice nowcasting technology for strong convection wind. And thunderstorm gale products can be provided for 10 minutes in future 2 hours.