35 results about "Predictive capability" patented technology

The invention relates to the technical field of marine ecology, in particular to a marine ecology multi-agent construction method and an interaction system thereof. The construction method comprises the following steps: a marine ecology large model construction step: constructing a fine adjustment data set, and training based on the fine adjustment data set to obtain a marine ecology large model; a parallel application module construction step: constructing a domain knowledge base and a marine ecological multi-intelligent body, constructing a multi-modal large model, and deploying the domain knowledge base, the marine ecological multi-intelligent body, the multi-modal large model and the marine ecological large model of the parallel application module in a GPU server, and the marine ecology large model is used for analyzing the user request and distributing the user request to the parallel application module, triggering the parallel application module to generate a response result, feeding the response result back to the marine ecology large model for semantic integration, and outputting the result. According to the invention, the intelligent level in the field of marine ecology is improved, the prediction capability of the model is enhanced, and the prediction result is more accurate and reliable.

The invention discloses a sea surface temperature prediction method based on an adaptive space-time diagram neural network, and the method comprises the following steps: (1), employing an optimal interpolation sea temperature OISST data set provided by an NOAA official website, and selecting two-dimensional sea surface temperature data of sea areas of the Bohai sea and the East China Sea; (2) carrying out data preprocessing on the collected two-dimensional data set to obtain spatio-temporal data with time coding and an initial adjacency matrix; dividing the data into a training set, a verification set and a test set; (3) constructing an SST prediction model based on the adaptive space-time diagram convolutional network; wherein the SST prediction model comprises a space-time prediction module, a graph learning module and an adjacent matrix updating module; (4) inputting a test set into the trained model for prediction, performing anti-standardization processing on a result, and calculating with an observation value to obtain a prediction error; the prediction capability is improved.

The invention relates to a marine meteorological time sequence prediction method based on a DC-LSTM-Attention combination model, and belongs to the field of marine meteorological time sequence prediction, and the method comprises the following steps: (1) data screening and preprocessing: selecting a data set of a marine meteorological scene, and carrying out the cleaning and normalization processing of the data; (2) designing a marine meteorological prediction model based on DC-LSTM-Attention, wherein the model comprises a cavity convolution layer, an LSTM layer and an Attention layer; and (3) model training and verification: inputting the data into a trained model, obtaining a multi-dimensional feature vector through processing, comparing the multi-dimensional feature vector with historical features in a database, and predicting future marine meteorological changes. The method has the advantages that the receptive field is expanded through the cavity convolution, and the features of different scales are better captured. The LSTM can effectively model long-term dependence in a time sequence, and the prediction capability of periodic changes is improved. The Attention mechanism enhances the adaptability of the model, so that the model can dynamically adjust the attention to different time steps and features, and the prediction performance under complex data is optimized.

The invention relates to the technical field of meteorological simulation, in particular to a dynamic three-dimensional grid meteorological simulation model based on historical observation data. Based on abundant historical observation data, the meteorological element change rule can be accurately reflected, the accuracy of meteorological simulation is greatly improved, and a reliable basis is provided for meteorological research. A dynamic three-dimensional grid result shows complex evolution of weather in a three-dimensional space in all directions, limitation of a traditional model is overcome, weather change details in the vertical direction and the horizontal direction are clearly shown, and deep understanding of a weather mechanism is facilitated. Physical driving and data driving models are fused, the advantages of the physical driving and data driving models are brought into full play, the simulation and prediction capability of various meteorological phenomena is enhanced, and particularly for complex and extreme meteorological phenomena, meteorological disaster early warning is better served. Through visualization and interaction functions, a vivid three-dimensional meteorological scene can be generated, a user can conveniently observe and analyze visually, and development and decision making of the meteorological field and related industries such as aerospace and agriculture are powerfully promoted.

The invention provides a high-resolution ocean thermocline depth rapid reconstruction method based on a random forest. The method is used for carrying out rapid reconstruction and quasi-real-time acquisition on the depth of an ocean thermocline by using sea surface data. A random forest model is established, sea surface data such as sea surface temperature, salinity and sea surface height abnormity are utilized, historical data of thermocline depth calculated based on actually measured temperature data are combined, and regression relation fitting is carried out on sea surface elements and thermocline depth, so that the sea surface elements and the thermocline depth are obtained. Therefore, the targets of multi-source ocean data fusion and high-resolution thermocline depth rapid reconstruction are achieved. In order to train the model, a large amount of measured data is used for sample fitting and optimization, so that the prediction accuracy is improved; the method has the advantages of being accurate, efficient and wide in applicability, the temporal-spatial resolution of ocean thermocline depth calculation can be improved, the ability to understand and predict an ocean temperature field is enhanced, and the method has important significance in the fields of ocean engineering, physical ocean and the like.

The invention discloses a rainfall prediction method and system for the Beijing, Tianjin and Hebei region, and relates to the technical field of rainfall prediction, and the method comprises the steps: obtaining multi-source data, carrying out the preprocessing of the data, extracting the core characteristics of the data, carrying out the dynamic rainfall modeling according to the core characteristics of the data, and simulating a rainfall evolution path; and constructing a rainfall prediction model, performing rainfall intensity prediction based on the rainfall evolution path, performing extreme weather early warning based on a prediction result, and generating a rainfall distribution diagram of a high-risk area. Through fusion of multi-source data, the generative adversarial network and spectral analysis, the spatial resolution of the data and the rainfall evolution prediction capability are improved, the real-time performance and the accuracy of rainfall prediction are improved in combination with dynamic rainfall modeling, a fractional order differential equation and a multiple regression model, and the prediction accuracy is improved. The rainfall prediction challenge under complex weather conditions can be effectively dealt with, and accurate prediction of rainfall intensity and extreme weather events is realized.

The invention discloses an ENSO prediction method, system and device based on a space-time fusion neural network, and belongs to the field of deep learning model prediction. The invention designs a time-space fusion Transform network, the network designs a parallel encoder-decoder structure, and the time-space information can be effectively extracted from the ocean surface temperature data and the # imgabs0 # index at the same time, so that the prediction precision of the # imgabs1 # index is improved. Meanwhile, the network extracts global features by using an attention mechanism in an encoder of the network, and establishes a remote dependency relationship on a target. Compared with a comparison method, the method has better prediction capability in multiple prediction advance periods, the problem that an existing ENSO prediction method underestimates the ENSO event amplitude along with the increase of the prediction advance periods is solved, and the ENSO prediction precision is improved.

The present invention proposes a flood monitoring and early warning method and system based on the fusion of multi-source remote sensing satellite data. The present invention collects various types of remote sensing satellite data of flood areas; pre-processes the collected remote sensing satellite data; integrates the pre-processed multiple remote sensing satellite data with features of different scales to obtain feature fusion; generalizes the feature fusion based on the revisited channel attention module from the same-layer scene to the cross-layer scene; utilizes the fused remote sensing images to identify floods and extract the boundaries and scope of the flood area; conducts flood warnings based on the identified flood areas and historical flood data, combined with meteorological and hydrological information, to obtain warning results; and conducts a comprehensive analysis of the causes, processes, and impacts of flood events. By integrating multi-source satellite data, the present invention aims to improve the accuracy of large-scale flood detection and enhance the ability to predict flood trends, thereby ensuring the accuracy and reliability of spatiotemporal flood warnings.

The invention discloses a sea ice concentration space prediction method based on an MLR model, and the method comprises the steps: calculating lead-lag correlation coefficients between the Wild sea ice concentration and global sea surface temperature, average sea level pressure and surface temperature in different lead periods, so as to determine atmosphere and ocean prediction factors which affect Wild sea SIC; an MLR space prediction model is constructed based on the prediction factors, and SIC data of past 12 months are obtained in the prediction factors; and according to a predicted object, the MLR model carries out multiple linear regression on the SIC of each lattice point in the South Pole Wild sea area according to the six prediction factors, and monthly prediction of sea ice space distribution is realized. According to the method, the Wildhai SIC space monthly prediction model is established based on the MLR statistical method, remote atmosphere and ocean climate signals can be brought into sea ice prediction, the influence of the continuity of atmosphere, ocean and sea ice on prediction can be considered respectively, and the MLR model shows good prediction capacity.

This invention relates to the field of meteorological services and provides a method and system for correcting subseasonal model precipitation based on climate zoning and sea surface temperature (SST) background constraints. The method includes: climate zoning, adding SST background constraints, reanalysis data modeling, model historical return data modeling, real-time prediction model matching, and prediction result correction. The system includes: a data preprocessing unit, a climate zoning unit, an SST background constraint unit, a reanalysis data modeling unit, a model historical return data modeling unit, and a real-time model matching and prediction result correction unit. This invention considers the importance of regional differences in precipitation and SST background constraints, solving the problems of significant regional differences in daily summer precipitation in my country, the tendency for overfitting in single-point modeling, and the lack of physical constraints that make correction methods unsuitable. It effectively utilizes physically meaningful predictability sources to modify the cumulative probability density correction method based on percentile mapping, thereby improving my country's summer subseasonal precipitation prediction capabilities.

The invention belongs to the technical field of water environment monitoring and early warning, and discloses a lake and reservoir chlorophyll a concentration prediction method based on an SO-KNN model. According to the invention, multi-time scale meteorological cumulative effect features are introduced to enrich information representation, and an SO-KNN intelligent prediction model is constructed. According to the method, under the conditions of data scarcity and non-equilibrium, the chlorophyll a concentration, especially the high-precision and strong-generalization prediction capability of the water bloom risk critical point, is remarkably improved. The model is simple in structure and efficient in calculation, the common defects of overfitting, insufficient generalization ability and the like of a complex machine learning model in the scene are effectively overcome, and a reliable and practical innovative technical solution is provided for early water bloom warning of northern reservoirs and water areas with similar data conditions.

The present application relates to the technical field of geological exploration, in particular to a coal mine GIS transparent geological management system fusing multi-source geological exploration data, comprising a multi-source geological and gas monitoring data acquisition module: obtaining structural geological data and dynamic gas data, forming a structure-gas joint observation data set; a structure-gas coupling modeling module: constructing a three-dimensional geometric model, dynamically adjusting the fracture development zone and the stress concentration area, and generating a structure-gas double-field coupling three-dimensional geological model; a model dynamic updating and local feedback correction module: when the gas sensor detects abnormal changes or new drilling data is accessed, triggering an automatic updating mechanism; a GIS transparent visualization and risk labeling module: using a multi-layer fusion and asynchronous drawing mechanism for real-time rendering, and labeling and displaying high-risk areas. The present application can truly reflect the dynamic correlation between geological structure and gas evolution, and has stronger prediction ability and safety evaluation adaptability.

This application relates to the field of marine ecological technologies, particularly to a construction method of marine ecological multi-agents and its interaction system. Among them, the construction method includes: a step of constructing a marine ecological large model, constructing a fine-tuning data set, and training a marine ecological large model based on the fine-tuning data set; a step of constructing a parallel application module, constructing a domain knowledge base, marine ecological multi-agents, and building a multi-modal large model, deploying the domain knowledge base, marine ecological multi-agents, multi-modal large model of the parallel application module and the marine ecological large model on a GPU server, where the marine ecological large model is used to parse user requests and distribute them to the parallel application module, triggering the parallel application module to generate response results and feedback them to the marine ecological large model for semantic integration and then output. Through this application, the intelligent level in the field of marine ecology is improved, the prediction ability of the model is enhanced, and the prediction results are made more accurate and reliable.

The invention discloses a method for predicting meteorological disasters based on weather classification, and relates to the technical field of weather prediction. Acquiring spatial meteorological data of the target area, and performing three-dimensional processing on the spatial meteorological data to obtain training data; obtaining a terrain distribution diagram of the target area, and inputting the training data and the terrain distribution diagram into a weather prediction model for training to obtain a weather prediction result; and determining a weather type of the target area according to the weather prediction result, and performing meteorological disaster prediction according to the weather type. According to the method, the spatial meteorological data of the target area are comprehensively obtained, three-dimensional processing is carried out, and the training data are input into an advanced weather prediction model in combination with detailed terrain distribution map information, so that the prediction capability of the model can be remarkably improved. According to the process, the weather type of the target area can be more accurately determined, powerful support is provided for meteorological disaster prediction, and the prediction accuracy is effectively enhanced.

The present invention provides a method and device for predicting chlorophyll concentration in the ocean surface, and relates to the technical field of ocean data analysis and processing; the method is to first import water body chlorophyll data and satellite remote sensing Rrs reflectivity, then perform time matching, standardization processing and logarithmic conversion, then screen satellite data set features to construct a target data set, and then integrate a chlorophyll concentration prediction model based on multiple gradient models, and finally input the satellite target data set and the in-situ data set as independent variables and dependent variables respectively into the model and MDN network prediction, and merge the results using a regression method. The present invention effectively extracts key information from complex data, improves model capture and prediction capabilities, makes up for feature deficiencies, reduces the impact of data distribution, improves the limitations of fusion algorithms, and uses data reasonably, the model is highly targeted, and the prediction results are scientific and practical.

The application discloses a time-frequency cooperative significant wave height prediction method and system based on cross attention and adaptive weighting, and the method specifically comprises the following steps: acquiring hourly historical observation data of an observation station; constructing an ATF-CAN model based on cross attention and adaptive weighting, wherein the model comprises a learnable frequency domain decomposition module, a multi-path encoder group, a two-stage cross attention module, an adaptive weighting fusion module and a position embedding decoder; training the ATF-CAN model, verifying the trained ATF-CAN model by using a verification set; and predicting the significant wave height of a test set by using the trained ATF-CAN model. The application improves the generalization prediction ability of the model under multiple sea conditions.

The invention relates to the technical field of climate event analysis, in particular to a regional composite thermal wave population exposure risk analysis method based on multi-source data. The method comprises the following steps: acquiring day-by-day temperature observation data, month-by-month reanalysis data of various meteorological elements and population distribution data in a target area; judging a composite heat wave event by adopting a percentile threshold method, and counting the composite heat wave frequency; on the basis of an inverse algorithm, calculating a whole-layer atmospheric heat source according to monthly meteorological reanalysis data; analyzing the correlation between the composite heat wave frequency in the target area and an atmospheric heat source, and establishing a linear regression equation; and calculating the population exposure degree of the composite heat wave in the target area caused by the abnormal change of the atmospheric heat source, and quantitatively evaluating the population exposure risk level. The method disclosed by the invention reveals the formation mechanism of the composite heat wave event from a new perspective, assists in improving the prediction capability, provides a solid scientific basis for accurately identifying the distribution of related high-risk crowds, and has important scientific significance and application prospects.

The application discloses a meteorological satellite cloud map extrapolation method based on a geographical attention mechanism, characterized by a deep learning framework constructed by GeoAttX, the introduction of a geographical attention mechanism to enhance the modeling ability of spatial features, and the synergistic effect of geographical semantic modeling, prediction path optimization and distribution correction mechanism, using the observation data of meteorological satellites to realize high-precision and low-error accumulation cloud prediction. The method uses a reverse greedy algorithm to realize flexible forecast time by reducing the autoregressive step, and proposes a second-order standardization method to normalize the pixel distribution, to maintain the consistency of brightness statistics and reduce error propagation. Compared with the prior art, the application has the advantages of using real-time stationary meteorological satellite observation data, having error suppression capability, geographical structure modeling capability, long-term stable prediction capability, high precision and stability, and can effectively improve the cloud motion extrapolation capability, and has important application value for weather forecasting and disaster warning.

The present invention discloses a spatial prediction method of sea ice density based on an MLR model, including calculating the lead-lag correlation coefficient between the Weddell Sea sea ice density and the global sea surface temperature, the mean sea level pressure and the surface air temperature in different lead times, so as to determine the atmospheric and oceanic prediction factors affecting the SIC of the Weddell Sea; constructing an MLR spatial prediction model based on the prediction factors, wherein the SIC data of the past 12 months are respectively included in the prediction factors; according to the predicted object, the MLR model performs multiple linear regression on the SIC of each grid point in the Antarctic Weddell Sea region based on the above six prediction factors, so as to realize the monthly prediction of the spatial distribution of sea ice. The present invention is to establish a spatial monthly prediction model of the Weddell Sea SIC based on the MLR statistical method, which can incorporate the climate signals of the distant atmosphere and ocean into the sea ice prediction, and can consider the influence of the persistence of the atmosphere, ocean and sea ice itself on the prediction respectively, and the MLR model shows good prediction ability.

The invention discloses a multi-time-efficiency short-temporary rainfall prediction method based on time-space joint feature extraction, and the method comprises the steps: carrying out the down-sampling processing of preprocessed multi-dimensional meteorological data through employing an encoder based on a U-Net architecture, and carrying out the semantic embedding of input information through employing a time information embedding module, and the sub-module has the time sensing capability. And the features after down-sampling processing are input into a Transform-Mama hybrid module for efficiently extracting space and time features, and window division is used in front of each sub-module for parallel processing. And performing up-sampling reduction on the feature data by using a decoder, complementing information lost due to down-sampling by using jump connection, and finally inputting the information into a multi-layer perceptron for prediction to obtain a precipitation result of the target area. The method can effectively carry out multi-time local high-precision rainfall prediction in a short critical time, realizes the complexity of a lightweight model, and has excellent prediction capability.

The invention discloses a ConvLSTM sea level height change prediction method based on reduced gravity hypothesis, and the method is suitable for the prediction of global sea level height anomaly (SLA). The method comprises the following steps: S1, acquiring and preprocessing global satellite altimeter data; s2, based on a ConvLSTM model, extracting spatio-temporal characteristics of sea level height anomaly; s3, introducing wind stress data, and combining physical constraints of a reduced gravity model to enhance the prediction capability of the model for wind-driven Rogobby waves; and S4, outputting a high-precision sea level height prediction result through error evaluation and optimization. According to the method, global satellite altimeter data in 1993-2014 are used for training, experiments in a black tide extension sea area show that the prediction correlation coefficient reaches 0.98, the effective prediction time efficiency can reach 365 days, the performance is more excellent in a low-frequency change area, and the accuracy, stability and timeliness of sea level prediction are remarkably improved.

The present invention discloses a flood forecasting method and system based on a support vector machine model, which belongs to the field of runoff forecasting; the method and system include S1, collecting hydrological and meteorological data of a study area; S2, determining the parameter CN value of a runoff generation module of an SCS-CN model based on the collected hydrological and meteorological data; S3, improving the runoff generation module of the SCS-CN model by using rainfall characteristic factors and the parameter CN value of the runoff generation module; S4, establishing an SCS-CN three-water source confluence model by coupling the runoff generation module improved in S3 with a three-water source confluence module; S5, constructing a dynamic confluence hysteresis parameter through a support vector machine model to improve the SCS-CN three-water source confluence model; S6, forecasting the flood process line of a watershed outlet section by using measured rainfall data; considering the influence of rainfall characteristic factors, improving the prediction ability of the model in areas with relatively dry climates; considering factors such as the spatial distribution of rainfall events and previous rainfall, so as to reduce the risk of prediction deviation of the hydrological model due to uneven spatial distribution of rainfall.

The present invention proposes a high-resolution ocean thermocline depth rapid reconstruction method based on random forest, which is used for rapid reconstruction and quasi-real-time acquisition of the ocean thermocline depth using sea surface data; by establishing a random forest model, using sea surface data, such as sea surface temperature, salinity, sea surface height anomaly, etc., combined with historical data of thermocline depth calculated based on measured temperature data, the sea surface elements and the thermocline depth are fitted with a regression relationship, thereby achieving the goals of multi-source ocean data fusion and high-resolution thermocline depth rapid reconstruction; in order to train the model, a large amount of measured data is used for sample fitting and optimization to improve prediction accuracy; the present invention has the advantages of accuracy, high efficiency and wide applicability, can improve the spatiotemporal resolution of ocean thermocline depth calculation, enhance the understanding and prediction ability of ocean temperature field, and is of great significance to the fields of ocean engineering and physical oceanography.

The invention belongs to the technical field of marine environment monitoring and prediction, and discloses a non-linear sea wave reconstruction and prediction method and system based on coherent radar, and the method comprises the steps: obtaining sea wave data through the coherent radar; processing the sea wave data to obtain the radial speed of the sea surface; constructing a propagation operator matrix based on a linear wave theory; introducing a second-order nonlinear correction term to correct the linear wave theoretical model; solving a nonlinear equation containing the second-order nonlinear correction term to obtain a model coefficient; and reconstructing and predicting the wave height time calendar based on the model coefficient. The sea surface elevation is predicted by using the coherent radar, so that more reliable sea wave data can be obtained; after the radial velocity of the sea waves is obtained through a spectral distance method, band-pass filtering is carried out on the radial velocity, and therefore the more accurate radial velocity of the sea waves is obtained. By introducing a second-order nonlinear term, the method has higher reconstruction and prediction capabilities on nonlinear sea waves, and finally, integration of sea wave reconstruction and prediction is realized.

The present invention relates to the field of meteorological detection technology, specifically to a method and system for detecting an average electric field in the air atmosphere, comprising the following steps: based on real-time meteorological data, historical trend analysis is performed on the temperature, humidity, and wind speed data in the atmosphere, and by comparing with the real-time data, key meteorological factors that affect the measurement results of the electric field strength are identified, and a list of key meteorological factors is generated. In the present invention, by combining real-time meteorological data and historical trends to analyze key meteorological factors, the accuracy and timely response capability of atmospheric electric field strength measurement are enhanced, by identifying and calculating the sensitivity of meteorological factors and adjusting model parameters, the accuracy and adaptability of electric field strength prediction are optimized, by establishing an atmospheric electric field dependency model, the model's ability to predict weather changes is improved, the ability to predict electric field change trends is enhanced, and the estimation of the fluctuation range is optimized, providing data support for predicting extreme weather events and conducting climate research.

The invention relates to the field of meteorological services, and provides a sub-seasonal mode rainfall correction method and system based on climate partitioning and sea temperature background constraint, and the method comprises the steps: climate partitioning, sea temperature background constraint addition, reanalysis data modeling, mode historical return data modeling, real-time prediction model matching, and prediction result correction. The system comprises a data preprocessing unit, a climate partitioning unit, a sea temperature background constraint unit, a reanalysis data modeling unit, a mode historical return data modeling unit and a real-time model matching and prediction result correction unit. According to the method, the importance of rainfall regional differences and sea temperature background constraints is considered, and the problem that the correction method is not applicable due to the fact that the daily rainfall regional differences in summer in China are obvious, overfitting is likely to occur to single-point modeling and physical constraints are lacked is solved; a predictable source with physical significance is effectively utilized to transform the cumulative probability density correction method of percentile mapping, and the rainfall prediction capacity of China in the summer sub-seasons is improved.

The invention belongs to the technical field of marine environment meteorological analysis, and particularly relates to a temperature and humidity data model construction method based on a K-means clustering algorithm and a Bayesian network, the K-means clustering process is optimized by introducing three evaluation indexes of SilhouetteCoffeient, Clinski-HarabaszIndex and Davies-BouldinIndex, the rationality and accuracy of the clustering result are ensured, and the accuracy of the temperature and humidity data model is improved. Therefore, different temperature and humidity change modes can be distinguished more finely. Besides, a local causal network structure is constructed by using a K2 Bayesian network structure learning algorithm in combination with a Bayesian-Dirichlet equivalent scoring function, so that the learning efficiency of the model is improved, and the prediction capability of the temperature and humidity change trend under a specific meteorological condition is enhanced. Through statistics of sample frequency and Laplacian smooth estimation of conditional probability distribution, high-precision prediction of future temperature and humidity changes is realized. According to the method, the limitation in the traditional technology is overcome, and powerful technical support is provided for meteorological monitoring and early warning in coastal areas.

The invention discloses a meteorological element forecasting method and device based on a genetic algorithm, equipment and a medium. The method comprises the following steps: acquiring actual meteorological data and initial meteorological forecasting data of different climate modes; obtaining target weather forecast data of each observation station according to the initial weather forecast data and the position information of each observation station; searching an optimal solution set in the plurality of initial weight candidate solution sets by using a genetic algorithm to obtain a target weight of each climate mode; and performing weighted average according to the predicted meteorological data in each climate mode and the target weight to obtain target predicted meteorological data. According to the method, the optimal weight combination is searched in a plurality of initial weight candidate solution sets by using the genetic algorithm, so that the optimal weight combination more reasonably reflects the prediction capability of different climate modes, weighted averaging is performed through the optimal weight combination and the weather forecast data, more accurate target prediction weather data is obtained, and the prediction efficiency is improved. Therefore, the accuracy of weather forecast data is improved.

The invention discloses a northwest region summer drought prediction method, and relates to the technical field of regional drought prediction, and the method comprises the steps: 1, building a northwest region summer drought prediction weather dynamics conceptual model under the synergistic effect of multiple factors; 2, establishing a rainfall dynamic-statistical forecasting model based on CFSv2 forecasting data under the multi-factor synergistic effect; 3, forecasting technology research of the next season extension period under the multi-factor synergistic effect is carried out; according to the method for predicting the summer drought in the northwest region, large-scale circulation and sea temperature in spring are gradually influenced through heat-sensitive heating and air-sea interaction from front winter foreign aid forcing factor change and mutual influence, and then direct influence factors of the summer drought in the northwest region are influenced through high-low altitude configuration adjustment and air-sea interaction; and finally, a weather dynamics conceptual model influencing drought and flood in northwest summer improves the predictability of drought events, and realizes the comprehensive prediction capability of full-time-scale drought.

The invention belongs to the technical field of marine science and data processing, and provides a base model construction method and system for marine observation data, and the method comprises the steps: constructing a marine base model; inputting the acquired ocean time series data into an ocean foundation model, constructing a Token sequence of a timestamp granularity level of the ocean time series data, and connecting an offset delay feature vector and a covariable vector before a timestamp into a timestamp Token vector; in combination with a causal self-attention mechanism and rotation position embedding, the features of the Token sequence are mapped to a decoder of a Transform, and a next timestamp sequence is generated; and defining a loss function, and optimizing model parameters to obtain a trained ocean foundation model. According to the method, universal feature representation and time sequence modes are learned on various ocean data sets, so that an ocean universal model with high generalization ability is constructed, large-scale and multi-source ocean data can be processed, time sequence characteristics of the data can be captured, and real-time data processing and prediction ability can be provided.