• Korean Journal of Agricultural and Forest Meteorology
• /
• v.25 no.4
• /
• pp.368-375
• /
• 2023

In this study, we examined the error characteristic and bias correction method for one-month temperature forecast data produced through joint development between the Rural Development Administration and the H ong Kong University of Science and Technology. For this purpose, hindcast data from 2013 to 2021, weather observation data, and various environmental information were collected and error characteristics under various environmental conditions were analyzed. In the case of maximum and minimum temperatures, the higher the elevation and latitude, the larger the forecast error. On average, the RMSE of the forecast data corrected by the linear regression model and the XGBoost decreased by 0.203, 0.438 (maximum temperature) and 0.069, 0.390 (minimum temperature), respectively, compared to the uncorrected forecast data. Overall, XGBoost showed better error improvement than the linear regression model. Through this study, it was found that errors in prediction data are affected by topographical conditions, and that machine learning methods such as XGBoost can effectively improve errors by considering various environmental factors.

• Journal of Energy Engineering
• /
• v.27 no.4
• /
• pp.103-110
• /
• 2018

In this paper, we introduce evaluation method of time series prediction model with new approach of Mean Absolute Percentage Error(hereafter MAPE) and Symmetric Mean Absolute Percentage Error(hereafter sMAPE). There are some problems using MAPE and sMAPE. First MAPE can't evaluate Zero observation of dataset. Moreover, when the observed value is very close to zero it evaluate heavier than other methods. Finally it evaluate different measure even same error between observations and predicted values. And sMAPE does different evaluations are made depending on whether the same error value is over-predicted or under-predicted. And it has different measurement according to the each sign, even if error is the same distance. These problems were solved by Maximum Mean Absolute Percentage Error(hereafter mMAPE). we used the absolute maximum of observed value as denominator instead of the observed value in MAPE, when the value is less than 1, removed denominator then solved the problem that the zero value is not defined. and were able to prevent heavier measurement problem. Also, if the absolute maximum of observed value is greater than 1, the evaluation values of mMAPE were compared with those of the other evaluations. With Beijing PM2.5 temperature data and our simulation data, we compared the evaluation values of mMAPE with other evaluations. And we proved that mMAPE can solve the problems that we mentioned.

• Journal of the Korean Solar Energy Society
• /
• v.40 no.5
• /
• pp.13-22
• /
• 2020

Day ahead forecast is necessary for the electricity market to stabilize the electricity penetration. Numerical weather prediction is usually employed to produce the solar irradiance as well as electric power forecast for longer than 12 hours forecast horizon. Korea Meteorological Administration operates the UM-LDAPS model to produce the 36 hours forecast of hourly total irradiance 4 times a day. This study interpolates the hourly total irradiance into 15 minute instantaneous irradiance and then compare them with observed solar irradiance at four ground stations at 1 minute resolution. Numerical weather prediction model employed here was produced at 00 UTC or 18 UTC from January to December, 2018. To compare the statistical model for the forecast horizon less than 3 hours, smart persistent model is used as a reference model. Relative root mean square error of 15 minute instantaneous irradiance are averaged over all ground stations as being 18.4% and 19.6% initialized at 18 and 00 UTC, respectively. Numerical weather prediction is better than smart persistent model at 1 hour after simulation began.

• Journal of Korea Water Resources Association
• /
• v.54 no.12
• /
• pp.1285-1294
• /
• 2021

In recent years, as rainfall is concentrated and rainfall intensity increases worldwide due to climate change, the scale of flood damage is increasing. Rainfall of a previously unobserved magnitude falls, and the rainy season lasts for a long time on record. In particular, these damages are concentrated in ASEAN countries, and at least 20 million people among ASEAN countries are affected by frequent flooding due to recent sea level rise, typhoons and torrential rain. Korea supports the domestic flood warning system to ASEAN countries through various ODA projects, but the communication network is unstable, so there is a limit to the central control method alone. Therefore, in this study, an artificial intelligence-based flood prediction model was developed to develop an observation station that can observe water level and rainfall, and even predict and warn floods at once at one observation station. Training, validation and testing were carried out for 0.5, 1, 2, 3, and 6 hours of lead time using the rainfall and water level observation data in 10-minute units from 2009 to 2020 at Junjukbi-bridge station of Seolma stream. LSTM was applied to artificial intelligence algorithm. As a result of the study, it showed excellent results in model fit and error for all lead time. In the case of a short arrival time due to a small watershed and a large watershed slope such as Seolma stream, a lead time of 1 hour will show very good prediction results. In addition, it is expected that a longer lead time is possible depending on the size and slope of the watershed.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.25 no.4
• /
• pp.404-414
• /
• 2023

In this study, digital climate maps with high-resolution (1km, daily) for the period of 1981 to 2020 were produced for the use as reference data within the procedures for statistical downscaling of climate change scenarios. Grid data for the six climate variables including maximum temperature, minimum temperature, precipitation, wind speed, relative humidity, solar radiation was created over Korean Peninsula using statistical downscaling model, so-called IGISRM (Improved GIS-based Regression Model), using global reanalysis data and in-situ observation. The digital climate data reflects topographical effects well in terms of representing general behaviors of observation. In terms of Correlation Coefficient, Slope of scatter plot, and Normalized Root Mean Square Error, temperature-related variables showed satisfactory performance while the other variables showed relatively lower reproducibility performance. These digital climate maps based on observation will be used to downscale future climate change scenario data as well as to get the information of gridded agricultural weather data over the whole Korean Peninsula including North Korea.

• Journal of Digital Contents Society
• /
• v.19 no.4
• /
• pp.789-799
• /
• 2018

This article presents machine learning based approach on Big data to analyzing time series data for anomaly detection in such industrial complex system. Long Short-Term Memory (LSTM) network have been demonstrated to be improved version of RNN and have become a useful aid for many tasks. This LSTM based model learn the higher level temporal features as well as temporal pattern, then such predictor is used to prediction stage to estimate future data. The prediction error is the difference between predicted output made by predictor and actual in-coming values. An error-distribution estimation model is built using a Gaussian distribution to calculate the anomaly in the score of the observation. In this manner, we move from the concept of a single anomaly to the idea of the collective anomaly. This work can assist the monitoring and management of Smart Factory in minimizing failure and improving manufacturing quality.

• Structural Engineering and Mechanics
• /
• v.65 no.5
• /
• pp.547-556
• /
• 2018

Based on the shaking table tests of a 1:3.52 scale one-bay and one-story ancient wooden structure, a simplified structural mechanics model was established, and the structural state equation and observation equation were deduced. Under the action of seismic waves, the damage rule of initial stiffness and yield stiffness of the joint was obtained. The force hammer percussion test and finite element calculations were carried out, and the structural response was obtained. Considering the 5% noise disturbance in the laboratory environment, the stiffness parameters of the mortise-tenon joint were identified by the partial least squares of singular value decomposition (PLS-SVD) and the Extended Kalman filter (EKF) method. The results show that dynamic and static cohesion method, PLS-SVD, and EKF method can be used to identify the damage degree of structures, and the stiffness of the mortise-tenon joints under strong earthquakes is reduced step by step. Using the proposed model, the identified error of the initial stiffness is about 0.58%-1.28%, and the error of the yield stiffness is about 0.44%-1.21%. This method has high accuracy and good applicability for identifying the initial stiffness and yield stiffness of the joints. The identification method and research results can provide a reference for monitoring and evaluating actual engineering structures.

• Atmosphere
• /
• v.32 no.2
• /
• pp.149-162
• /
• 2022

This study investigated the prediction skill of the Asian dust seasonal forecasting model (GloSea5-ADAM) on the Asian dust and meteorological variables related to the dust generation for the period of 1991~2016. Additionally, we evaluated the prediction skill of those variables depending on the combination of the initial dates in the sub-seasonal scale for the dust source region affecting South Korea. The Asian dust and meteorological variables (10 m wind speed, 1.5 m relative humidity, and 1.5 m air temperature) from GloSea5-ADAM were compared to that from Synoptic observation and European Centre for medium range weather forecasts reanalysis v5, respectively, based on Mean Bias Error (MBE), Root Mean Square Error (RMSE), and Anomaly Correlation Coefficient (ACC) as evaluation criteria. In general, the Asian dust and meteorological variables in the source region showed high ACC in the prediction scale within one month. For all variables, the use of the initial dates closest to the prediction month led to the best performances based on MBE, RMSE, and ACC, and the performances could be improved by adjusting the number of ensembles considering the combination of the initial date. ACC was as high as 0.4 in Spring when using the closest two initial dates. In particular, the GloSea5-ADAM shows the best performance of Asian dust generation with an ACC of 0.60 in the occurrence frequency of Asian dust in March when using the closest initial dates for initial conditions.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.16 no.2
• /
• pp.195-202
• /
• 1998

In this study, we utilized various type of GPS observation measurements to get a camera projection center of the aerial triangulation and consequently to determine which type is acceptable. For the accuracy and the error analysis, comparison between a projection center from the conventional model adjustment and the result determined by the kinematic DGPS positioning which is fitted to the conventional model adjustment using 3D conformal transformation method has been made. The camera projection center is located within a $\pm{2m}$ for C/A code range measurements, $\pm{14cm}$ for L1 phase measurements and $\pm{10cm}$ for L1/L2 phase measurements with $1\sigma$. In this way, the accuracy of the camera projection center by the bundle block adjustment can be predicted.

• Journal of the Korea Convergence Society
• /
• v.9 no.3
• /
• pp.15-21
• /
• 2018

The ocean drifter is a device for observing the ocean weather by floating off the sea surface. The data observed through the drifter is utilized in the ocean weather prediction and oil spill. Observed data may contain incorrect or missing data at the time of observation, and accuracy may be lowered when we use the data. In this paper, we propose a data correction model using recurrent neural networks. We corrected data collected from 7 drifters in 2015 and 8 drifters in 2016, and conducted experiments of drifter moving prediction to reflect the correction results. Experimental results showed that observed data are corrected by 13.9% and improved the performance of the prediction model by 1.4%.