Model Output Data Research Articles

A Privacy-Preserving Scheme for a Traffic Accident Risk Level Prediction System

Due to the expansion of Artificial Intelligence (AI), especially Machine Learning (ML), it is more common to face confidentiality regulations about using sensitive data in learning models generally hosted in cloud environments. Confidentiality regulations such as HIPAA and GDPR seek to guarantee the confidentiality and privacy of personal information. Input and output data of a learning model may include sensitive data that must be protected. Adversaries could intercept and exploit this data to infer more sensitive data or even to determine the structure of the prediction model. To guarantee data privacy, one option could be encrypting data and making inferences over encrypted data. This strategy would be challenging for learning models that now must receive encrypted data, make inferences over encrypted data, and deliver encrypted data. To address this issue, this paper presents a privacy-preserving machine learning approach using Fully Homomorphic Encryption (FHE) for a model that predicts risk levels of suffering a traffic accident. Despite the limitations of experimenting with FHE on machine learning models using a low-performance computer, limitations that are undoubtedly overcome by using high-performance computational infrastructure, we built some encrypted models. Among the encrypted models based on Decision Trees, Random Forests, XGBoost, and Fully Connected Neural Networks (FCNN), the model based on FCNN reached the highest accuracy (80.1%) for the lowest inference time (8.476 s).

Integrated effects of pavement simulation models and scale differences on the thermal environment of tropical cities: physical and numerical modeling experiments

Simulation methods attempt to explain what happens in full-scale environments. However, as simplification procedures, they also have their limitations and opportunities. One of the applications is to use the output data of a physical model to calibrate numerical simulation, or even to use outputs of numerical simulations to analyze urban scale studies. But it is uncertain the error in the interaction between these models. This study aims to analyze the impact of scale analysis and pavements simulation model modification on ambient and surface temperature of asphalt pavement in a physical model of a tropical city street canyons. Therefore, a scaled outdoor experiment was conducted, and a numerical simulation model, using ENVI-met software, was used to investigate the spatiotemporal variation of air and pavement surface temperature, in urban (1:1) and reduced (1:15) scales. For studies on the surface temperature of pavements, within the temperature range of 12 ºC to 37 ºC, it is recommended to calibrate physical models using as input, data derived from numerical simulation models, yielding a mean absolute percentage error (MAPE) of 4.9%. For estimating data in real-world urban scale, within the air temperature range of 15 ºC to 37 ºC, it is proposed to use output data from simulated models in ENVI-met, that presented a mean absolute error (MAE) of ± 0.59 or physical models (MAE = ± 0.66). These results would be useful for the development of urban surface temperatures parametrizations.

Model for generating data from navigation sensors for assessing the spatial position of objects

When selecting or developing new algorithms for assessing the spatial position of objects based on data from consumer navigation equipment (CNA) of global navigation satellite systems (GNSS) and data from inertial MEMS modules, the task of generating test model input data that meets certain conditions arises. Such conditions include, for example, the requirements for limited time increments of elements of spatio-temporal trajectories and ease of formation of their characteristic sections. The characteristic sections of spatio-temporal trajectories can be sections of a stationary position – sections of “Rest” or sections of rectilinear movement. The importance of forming these areas is due to the need to test the potential capabilities of compared or developed algorithms for assessing the spatial position of objects to ensure an improvement in their accuracy characteristics in the presence of such areas in various conditions. A data generation model is proposed for comparing algorithms for estimating the spatial position of objects, which ensures that the time increments of elements of spatio-temporal trajectories are limited in time and the simple formation of their characteristic sections is met. The model is based on the selection of a set of specified parameters that would allow, on the one hand, to form characteristic sections of spatio-temporal trajectories, and, on the other hand, to obtain model output data of NAP GNSS and inertial MEMS modules. The selected set of specified model parameters is determined through a set of analytically specified functions with sequential calculation of their parameters, which is much simpler compared to traditional approaches that involve solving boundary value problems of optimal control with restrictions on the type of control functions or the construction and application of genetic models and algorithms. The use of the developed model makes it possible to generate test model input data for the selection or development of new algorithms for assessing the spatial position of objects, implemented in devices that include CNA of GNSS and inertial MEMS modules and produce results both in real time and in delayed time. The effectiveness of the proposed solution is confirmed by the criterion of suitability under the considered conditions.

Conducting volcanic ash cloud exercises: practising forecast evaluation procedures and the pull-through of scientific advice to the London VAAC

The London Volcanic Ash Advisory Centre (VAAC) provides forecasts on the expected presence of volcanic ash in the atmosphere to mitigate the risk to aviation. It is fundamentally important that operational capability is regularly tested through exercises, to guarantee an effective response to an event. We have developed exercises which practise the pull-through of scientific advice into the London VAAC, the forecast evaluation process, and the decision-making procedures and discussions needed for generating the best possible forecasts under real-time conditions. London VAAC dispersion model forecasts are evaluated against observations. To test this capability in an exercise, we must create observation data for a hypothetical event. We have developed new methodologies for generating and using simulated satellite and lidar retrievals. These simulated observations enable us to practise our ability to interpret, compare, and evaluate model output and observation data under real-time conditions. Forecast evaluation can benefit from an understanding of how different choices of model setup (input parameters), model physics, and driving meteorological data impact the predicted extent and concentration of ash. Through our exercises, we have practised comparing output from model simulations generated using different models, model setups, and meteorological data, supplied by different institutions. Our exercises also practise the communication and interaction between Met Office (UK) scientists supporting the London VAAC and external experts, enabling knowledge exchange and discussions on the interpretation of model output and observations, as we strive to deliver the best response capability for the aviation industry and stakeholders. In this paper, we outline our exercise methodology, including the use of simulated satellite and lidar observations, and the development of the strategy to compare output generated from different modelling systems. We outline the lessons learnt, including the benefits and challenges of conducting exercises which practise our ability to provide scientific advice for an operational response at the London VAAC.

Prediction of RUL of lubricating oil based on Deep Learning Technique

This study uses condition monitoring (CM) to determine the lubricating oil's remaining usable life (RUL). First, Atomic emission spectroscopy (AES) was used to determine the lubricating oil's element makeup and content. Under use were quantitatively examined. The precision and efficacy of the RUL prediction model might be diminished in light of the wide range of oil data acquired using AES. A thorough it is recommended to screen oil data using a method for choosing parameters based on correlation analysis, lubricant degradation analysis, and information entropy. In order to address this issue. The lubricant's RUL prediction model was then developed using a support vector machine (SVM). It is shown that the SVM prediction model's efficacy and accuracy by contrasting the experimental findings with the prediction model's output data.

Projecting marine fish distributions during early life stages under future climate scenarios

AbstractChanges to Earth's climate affect organisms globally; in marine systems, these impacts are seen through warming water temperatures, ocean acidification, hypoxia and frequent marine heatwaves. These effects may lead to the movement of species to more favourable conditions. While climate‐driven movement is well studied at the adult stage, how the early life stages of marine fish will respond to future variability is less clear. Many fish species are constrained by specific spawning locations or phenology. Spawning in certain locations allows for local retention of offspring, while precise timing can facilitate transport of offspring to nursery locations through seasonal circulation patterns. Our research investigates how changing oceans impact the location and timing of spawning of Bering Sea groundfishes over the next century. We used ROMS SST and SSS model output and NOAA survey data in species distribution models to hindcast and project distributions and centre of gravity for eggs and larvae of six groundfish species. Our analyses found that most of our study species exhibit flexible geography. However, the speed and direction of egg and larval movement did not track the speed and direction of their respective thermal niches. Hence, the projected distributional patterns of adult stages may be limited by their early life stages. This response is likely to be mirrored globally by other species with planktonic eggs and larvae. These results indicate that life history considerations are critical for the management of commercially important species, as effects on early life stages are strongly connected to the success or failure of adult populations.

СТАТИСТИЧЕСКАЯ КОРРЕКЦИЯ ДОЛГОСРОЧНЫХ ПРОГНОЗОВ ПРИЗЕМНОЙ ТЕМПЕРАТУРЫ ВОЗДУХА ПО МОДЕЛИ ПЛАВ ДЛЯ ТЕРРИТОРИИ СЕВЕРНОЙ ЕВРАЗИИ

For the territory of Northern Eurasia, a scheme for the statistical correction of surface air temperature forecasts has been developed for periods of 1-4 months on the basis of the SL-AV model using the MOS concept. For statistical correction of operational temperature forecasts, the regression parameters and EOF expansion coefficients obtained by cross-validation on historical forecasts were used. Due to the internal relationships of the model output data, the proposed scheme allows improving the skill of surface characteristic forecasts. A significant improvement in the skill of deterministic air temperature forecasts by using statistical correction is manifested in transition seasons. The scheme of statistical correction is constantly evolving. Further development of the statistical correction technology involves the use of neural networks and forecast indices of atmospheric circulation.

Аналитическая и компьютерная программная математические модели шума выходного сигнала волоконно-оптического гироскопа, анализ и верификация

The goal of this work is to identify the main source of noise in the output signal of a fiber-optic gyroscope (FOG) and ways to reduce it. To achieve the stated goal, two mathematical models of the noise component of the output signal of the FOG were developed. These models are based on fundamental expressions defining shot and thermal noise, as well as an expression for the excess noise of an optical radiation source. The general expression of the analytical model takes into account the four fundamental sources of noise in FOG. Their influence on the output signal noise is expressed through the relation between the measured optical radiation power and the angular velocity acting on the FOG. The computer model was developed in the “Octave” environment. Model performs cyclic execution of the FOG operation algorithm with a closed feedback loop for measuring angular velocity. In the computer model, noise from four independent sources is added to a given useful signal. An array of model output data is generated. The noise power spectral density is calculating. The paper analyzes the influence of FOG circuit parameters on the noise level. The noise source having the greatest significance in the output signal was identified. To confirm the correctness of the models, a comparison was made between the simulation results and the noise of commercially produced FOG. The model error is no more than 10%.

Identification of rheological parameters for shallow water flows of viscoplastic fluids using elevation hydrographs

In this paper, rheological parameters, in particular yield stress and consistency index, for viscoplastic fluids are inferred from elevation hydrographs derived from experiments. The direct model consisting of shallow water equations with a Herschel–Bulkley rheology is used to simulate a fluid flowing down an inclined plane and past a cylindrical occlusion. Numerical simulations are validated with experimental and related results from the literature. The aim is to infer the unknown rheological parameters using hydrograph measurements in the contact line region between the fluid and the occlusion. The rheological identification problem is formulated to minimize an objective functional that measures the discrepancy between the elevation hydrographs from the model output and experimental data. The inverse solver is tested on both synthetic and laboratory data. The set of rheological parameters inferred is compared with the values measured on a rheometer for the fluid used in the experiments. Inference of the unknown flow quantities from the wetting free-surface data has direct applications not only in industrial settings, to predict the wetting dynamics, but also in geophysical ones for risk assessments and management plans.

Identifying and characterising trapped lee waves using deep learning techniques

Trapped lee waves, and resultant turbulent rotors downstream, present a hazard for aviation and land‐based transport. Though high‐resolution numerical weather prediction models can represent such phenomena, there is currently no simple and reliable automated method for detecting the extent and characteristics of these waves in model output. Spectral transform methods have traditionally been used to detect and characterise regions of wave activity in model and observational data; however, these methods can be slow and have their limitations. Machine‐learning (ML) techniques offer a new and potentially fruitful method of tackling this problem. We demonstrate that a deep‐learning model can be trained to accurately recognise and label coherent regions of lee waves from vertical velocity data on a single level from a high‐resolution numerical weather prediction (NWP) model. Using transfer learning, wave characteristics (wavelength, orientation, and amplitude) can be extracted from the trained segmentation model. The use of synthetic wave fields with prescribed wave characteristics makes this transfer learning possible without the need to characterise real complex wave fields. Addition of noise to the synthetic data makes the models more robust when applied to more complex and noisy NWP data. The collection of trained models produced provides a valuable tool to investigate the prevalence and nature of lee wave activity, as well as a new way for forecasters to detect resolved waves. The deep‐learning model was more capable and quicker at detecting and characterising lee waves than a spectral technique was. This work is just one example of how already established ML techniques can be used to detect and characterise complex weather phenomena from NWP model output and observational data, and how the careful use of synthetic data can reduce the requirements for large volumes of hand‐labelled training data for ML models.

Patterns of Water Consumption in Longleaf Pine Restoration Areas and the Relationship with Cone Production

Monitoring water consumption dynamics across the geographic range of an ecosystem may indicate the possible variation and stress in a biome. Here, model output data based on remote sensing (1979–2022) were used to examine the water consumption dynamics and effects on cone production in three geographic margins in the longleaf pine’s range (i.e., Bladen Lake State Forest, Escambia Experimental Forest, and Kisatchie National Forest) under varying climatic conditions. Results indicated that the mean annual transpiration at Escambia was approximately 431 mm and that at Bladen and Kisatchie was 500 mm. Mean monthly transpiration peaked twice (June and October) at Escambia but only once (August) at Bladen and Kisatchie. The mean annual evapotranspiration ranged from approximately 900 mm at Kisatchie to about 791 mm at Escambia and Bladen. The mean annual transpiration/evapotranspiration ratio was about 0.65 at Bladen and 0.55 at Escambia and Kisatchie. A significant correlation existed between evapotranspiration and specific humidity across the sites on a monthly scale but not on a yearly scale. Significant negative relationships existed between precipitation and the ratios of transpiration/precipitation and evapotranspiration/precipitation on the yearly scale across the sites. Negative power relationships were observed between precipitation and the specific humidity/precipitation ratio on monthly and yearly scales. Cone production was generally highest in years with moderate water consumption. These results provide baseline information on how hydrological and ecological processes of longleaf pine forests interact with climate across broad spatial and temporal scales.

Nonlinear Effects of the Stratospheric Quasi‐Biennial Oscillation and ENSO on the North Atlantic Winter Atmospheric Circulation

AbstractPrevious studies have shown that both the stratospheric Quasi‐Biennial Oscillation (QBO) and the El Niño‐South Oscillation (ENSO) can influence the North Atlantic winter circulation. Here we use reanalysis and model output data to show that the QBO and ENSO interact to produce a nonlinear effect on the North Atlantic winter circulation. Specifically, during El Niño winters, the QBO teleconnection mainly takes a subtropical pathway with changes in the North Pacific–Atlantic subtropical jet (STJ); during La Niña winters, the QBO connects with the troposphere predominantly through a polar pathway, that is, stratospheric polar vortex (SPV) changes. Further, the QBO‐induced STJ changes in El Niño lead to anomalous Rossby wave propagation toward the North Atlantic, and the QBO‐induced SPV during La Niña anomaly exerts a downward effect on the North Atlantic. Hence, the various interactions between ENSO and QBO teleconnections result in nonlinear, and even synergistic, impacts on the North Atlantic circulation.

Bulanık Mantık Yaklaşımı Kullanılarak Kızılırmak Nehri Aylık Ortalama Akımlarının Tahmini

River flow values are used in the design and operation of hydraulic structures. Determining the correct flow value is important in terms of controlling water movements in the operation of hydraulic structures, irrigation of agricultural lands, hydroelectric production, environmental protection and flood control. In the literature, different methods are used to predict possible river flows using the available data. The fuzzy logic approach is a kind of intelligent system method used in solving problems involving uncertainty. The method has been widely used in the modeling of hydrological data for 2000’s. In this study, the fuzzy logic method was applied to estimate the flow data of Yamula Station on the Kızılırmak River in the Kızılırmak basin, one of the largest basins in Turkey. In addition to these flow station data, the monthly average temperature and monthly total precipitation data of the Kayseri meteorology station, which affects the station flows, were also used for modeling. Three different models were created for the flow estimates. In these models, temperature and precipitation data were selected as input values and river flow data were chosen as output values. In the models, 1982-2012 data of the stations were used. Model output data were tested with data set of 2013, 2014 and 2015. As a result, it has been seen that the fuzzy logic approach gave healthy results when both temperature and precipitation data were used as inputs.

Evaluation of soft soil foundation reinforcement effect of prefabricated building based on BP neural network

The use of traditional reinforcement methods in construction sites often causes problems such as pore water pressure, which can not effectively form a solid foundation. Aiming at this problem, the evaluation model of soft soil foundation reinforcement effect of prefabricated buildings is established based on BP neural network, combined with the geological characteristics of soft soil and the elements of foundation reinforcement; The L-M algorithm is used to optimize the slow convergence problem of BP neural network, and finally its evaluation effect is verified through practical application. The results show that the strengthening effect of 1550 kN⋅m/m2 is better than that of 2000 kN⋅m/m2 with the more times of tamping for marine and river facies, and there is a positive correlation between the times of strengthening and the effect. At the same time, similar qualitative conditions also show that the greater the burial depth, the worse the reinforcement effect. When the overlying soil layer is soft, the shallow buried soil layer can be reinforced by laying a cushion to improve the overall reinforcement effect. The laws reflected in the final model output data are the same as those reflected in the construction, and the accuracy of the proposed model is up to 87%, indicating that the model has superior performance in the reinforcement effect evaluation.

Data assimilation for urban stormwater and water quality simulations using deep reinforcement learning

Hydrological models have been used to understand the transportation of water quantity and quality in drainage systems, and the stormwater management model (SWMM) is one of the most widely-used models for runoff quantity and quality simulations in urban areas. Although significant efforts have been made to identify the appropriate input parameters of the SWMM model in various watersheds, it is difficult to reflect the variability of the real environment using a fixed input parameter. Data assimilation (DA) is a compatibility method that improves prediction accuracy by merging observations and simulation outputs. It is useful to reduce the forecast error from temporal transferability with information interaction between the model output and observation data. However, conventional DA approaches cannot completely overcome their unrealistic assumptions such as linearity, normality, and error covariances. To address these challenges, we used deep reinforcement learning (RL) to develop an automatic assimilation model that interactively optimizes the SWMM parameters in a real environment (SWMM-RL). In the SWMM-RL model, the agent is trained by rewarding and/or punishing the action (modulated SWMM input parameter) in real time according to the state changes. The model was constructed to minimize the error of runoff and pollutant load simulation of suspended solids (SS), total nitrogen (TN), and total phosphorus (TP) in each stormwater monitoring event. The results demonstrated that the SWMM-RL model primarily outperformed the fixed input parameter SWMM model (SWMM-PS) in simulating the runoff and the three different pollutant loads (the median value of Nash–Sutcliffe efficiency (NSE) for 10 stormwater events was increased by 0.11, 0.31, 0.36, and 0.07 for runoff, SS, TN, and TP, respectively). The SWMM-PS has disadvantages in simulating low rainfall events because of the high sensitivity of runoff peaks to heavy rainfall conditions. Furthermore, the effects of extreme rainfall events were estimated using the SWMM-RL model. This study showed how SWMM-RL combined with the DA method increased forecast accuracy by providing sensitivity and temporal transitions of input parameters.

Physics-Guided Neural Network Model for Aeroengine Control System Sensor Fault Diagnosis under Dynamic Conditions

Sensor health assessments are of great importance for accurately understanding the health of an aeroengine, supporting maintenance decisions, and ensuring flight safety. This study proposes an intelligent framework based on a physically guided neural network (PGNN) and convolutional neural network (CNN) to diagnose sensor faults under dynamic conditions. The strength of the approach is that it integrates information from physics-based performance models and deep learning models. In addition, it has the structure of prediction–residual–generation-fault classification that effectively decouples the interaction between sensor faults and system state changes. First, a PGNN generates the engine’s non-linear dynamic prediction output because the PGNN has the advantage of being able to handle temporal information from the long short-term memory (LSTM) network. We use a cross-physics–data fusion scheme as the prediction strategy to explore the hidden information of the physical model output and sensor measurement data. A novel loss function that considers physical discipline is also proposed to overcome the performance limitations of traditional data-driven models because of their physically inconsistent representations. Then, the predicted values of the PGNN are compared with the sensor measurements to obtain a residual signal. Finally, a convolutional neural network (CNN) is used to classify faults for residual signals and deliver diagnostic results. Furthermore, the feasibility of the proposed framework is demonstrated on an engine sensor fault dataset. The experimental results show that the proposed method outperforms the pure data-driven approach, with the predicted RMSE being reduced from 1.6731 to 0.9897 and the diagnostic accuracy reaching 95.9048%, thereby confirming its superior performance.

DETERMINATION OF THE INFLUENCE OF THE SOLAR CELLS DESIGN ON THEIR FUNCTIONING EFFICIENCY

An essential scientific and practical problem of sustainable development of the country’s energy sector is predicting the parameters and foreseeing the conditions of functioning of solar cells and solar elements in normal and emergency situations. It is emphasized that this allows to provide solar energy with high efficiency indicators, including return on invested capital in the solar panels construction. The main specific research method used is regression analysis aimed at building a model for predicting the total amount of generated energy from solar cells in ground-based installations under variable conditions. The distribution of the model output data by the number of solar cell modules is analyzed on the example of ground-mounted solar installations. To obtain empirical data, we selected 31 facilities in the Dnipro and Zaporizhzhia regions that use solar cells with different numbers of modules. This allows to calculate the weighted average amount of energy generated during operation in variable conditions. We distinguished 10 intervals of frequency values, with the largest range of values being in the range of 10000-20000 solar cell modules. The regression coefficients were checked by two criteria. According to the criterion of “t-statistics value”: the coefficients are significant at a significance level of 0.05. The criterion of asymptotic significance determines the rejection of the null hypothesis in a statistical test. The random value is 170.3694, which is significantly higher than the critical value of the F-statistic of 3.33. A model of the dependence of the total amount of generated energy on the number of solar modules and the weighted average amount of generated energy was built on the basis of regression analysis. It is determined that the model influence factor “number of solar modules” has a positive effect on the resultant factor (solar cell productivity). Whereas the influence factor is “weighted average amount of generated energy”. However, the influence factor “number of solar modules” is more significant. The obtained results allow us to assert the possibility of their implementation in the energy sector.

Data‐driven iterative learning trajectory tracking control for wheeled mobile robot under constraint of velocity saturation

AbstractConsidering the wheeled mobile robot (WMR) tracking problem with velocity saturation, we developed a data‐driven iterative learning double loop control method with constraints. First, the authors designed an outer loop controller to provide virtual velocity for the inner loop according to the position and pose tracking error of the WMR kinematic model. Second, the authors employed dynamic linearisation to transform the dynamic model into an online data‐driven model along the iterative domain. Based on the measured input and output data of the dynamic model, the authors identified the parameters of the inner loop controller. The authors considered the velocity saturation constraints; we adjusted the output velocity of the WMR online, providing effective solutions to the problem of velocity saltation and the saturation constraint in the tracking process. Notably, the inner loop controller only uses the output data and input of the dynamic model, which not only enables the reliable control of WMR trajectory tracking, but also avoids the influence of inaccurate model identification processes on the tracking performance. The authors analysed the algorithm's convergence in theory, and the results show that the tracking errors of position, angle and velocity can converge to zero in the iterative domain. Finally, the authors used a simulation to demonstrate the effectiveness of the algorithm.

Research on the damage diagnosis model verification method of cable-stayed bridges

To solve the problem that the damage diagnosis model of cable-stayed bridges cannot quantitatively evaluate its validity, the verification method of the diagnosis model is investigated. A model validation method [Root-Mean-Square Error (RMSE)-Statistical Hypothesis Test (SHT), R–S] that combines the RMSE evaluation index and the SHT is proposed and applied to the reliability evaluation of the cable-stayed bridge data mining damage diagnosis model. First, the R–S method evaluates the diagnostic effect of the model via the evaluation index. Second, the SHT method is used to quantitatively compare the consistency between the output data of the diagnostic model and the actual structure output data. Finally, the validity of the model is quantitatively evaluated. The application process of the proposed method is demonstrated by an example. The results show that the diagnostic model has strong reliability and that the confidence level Pδ exceeds 90%, which indicates the effectiveness of the diagnostic model verification method.

Study of the Utilization of WRF Model Output Data to Produce Daily Flood Forecast Maps in the North Sumatra Region

Flood is a natural disaster that often occurs in Indonesia. Currently, flood events are relatively difficult to predict because floods generally occur suddenly in uncertain periods. Extreme rainfall is a major factor for the occurrence of floods. Considering that floods can be caused by heavy rainfall events within a few hours, it is necessary to produce daily flood forecasts for flood disaster mitigation. This study aims to test the accuracy of utilizing rainfall forecast data from the Weather Research and Forecasting (WRF) model to create daily flood forecast maps. The data used in this study include Global Forecast System (GFS) data, BMKG rainfall measurement data which spread across several points in North Sumatra Province, and flood incident reports from BNPB. Data processing is carried out by Geospatial Information System (GIS) using Quantum-GIS, which includes weighting and scoring the parameters of soil type, slope, land elevation, river density, and land cover to produce Flood Prone Maps, then integrated with rainfall data to produce Daily Flood Forecast Maps. The case studies of flood events in this study are August 28 and November 28, 2022. The results showed that the spatial forecast of flood potential (WRF) has a pattern in accordance with the flood event area. Therefore, the WRF model output rainfall prediction data can be used to create a daily flood forecast map in the North Sumatra region.