• Journal of Korean Society of Disaster and Security
• /
• v.13 no.4
• /
• pp.75-92
• /
• 2020

In most cases of the water balance analysis, the return flow ratio for each water supply was uniformly determined and applied, so it has been contained a problem that the volume of available water would be incorrectly calculated. Therefore, sewage and wastewater among the return water were focused in this study and the data-driven model was developed to forecast the outflow from the sewage treatment plant. The forecasting results of LSTM (Long Short-Term Memory), GRU (Gated Recurrent Units), and SVR (Support Vector Regression) models, which are mainly used for forecasting the time series data in most fields, were compared with the observed data to determine the optimal model parameters for forecasting outflow. As a result of applying the model, the root mean square error (RMSE) of the GRU model was smaller than those of the LSTM and SVR models, and the Nash-Sutcliffe coefficient (NSE) was higher than those of others. Thus, it was judged that the GRU model could be the optimal model for forecasting the outflow in sewage treatment plants. However, the forecasting outflow tends to be underestimated and overestimated in extreme sections. Therefore, the additional data for extreme events and reducing the minimum time unit of input data were necessary to enhance the accuracy of forecasting. If the water use of the target site was reviewed and the additional parameters that could reflect seasonal effects were considered, more accurate outflow could be forecasted to be ready for climate variability in near future. And it is expected to use as fundamental resources for establishing a reasonable river water management system based on the forecasting results.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.26 no.3B
• /
• pp.279-289
• /
• 2006

The reconstruction of low dimension nonlinear behavior from the hydrologic time series has been an active area of research in the last decade. In this study, we present the applications of a powerful state space reconstruction methodology using the method of Support Vector Machines (SVM) to the Great Salt Lake (GSL) volume. SVMs are machine learning systems that use a hypothesis space of linear functions in a Kernel induced higher dimensional feature space. SVMs are optimized by minimizing a bound on a generalized error (risk) measure, rather than just the mean square error over a training set. The utility of this SVM regression approach is demonstrated through applications to the short term forecasts of the biweekly GSL volume. The SVM based reconstruction is used to develop time series forecasts for multiple lead times ranging from the period of two weeks to several months. The reliability of the algorithm in learning and forecasting the dynamics is tested using split sample sensitivity analyses, with a particular interest in forecasting extreme states. Unlike previously reported methodologies, SVMs are able to extract the dynamics using only a few past observed data points (Support Vectors, SV) out of the training examples. Considering statistical measures, the prediction model based on SVM demonstrated encouraging and promising results in a short-term prediction. Thus, the SVM method presented in this study suggests a competitive methodology for the forecast of hydrologic time series.

• Korea Trade Review
• /
• v.48 no.3
• /
• pp.69-88
• /
• 2023

This study analyzes the dynamic characteristics of cargo volume (demand), ship fleet (supply), and freight rate (price) of container, dry bulk, and tanker shipping markets by using the VAR and VECM models. This analysis is expected to enhance the statistical understanding of market dynamics, which is perceived by the actual experiences of market participants. The common statistical patterns, which are all shown in the three shipping markets, are as follows: 1) The Granger-causality test reveals that the past increase of fleet variable induces the present decrease of freight rate variable. 2) The impulse-response analysis shows that cargo shock increases the freight rate but fleet shock decreases the freight rate. 3) Among the three cargo, fleet, and freight rate shocks, the freight rate shock is overwhelmingly largest. 4) The comparison of adjR² reveals that the fleet variable is most explained by the endogenous variables, i.e., cargo, fleet, and freight rate in each of shipping markets. 5) The estimation of co-integrating vectors shows that the increase of cargo increases the freight rate but the increase of fleet decreases the freight rate. 6) The estimation of adjustment speed demonstrates that the past-period positive deviation from the long-run equilibrium freight rate induces the decrease of present freight rate.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.24 no.2
• /
• pp.63-77
• /
• 2022

Our aim was to reduce estimation errors of a wind velocity model used as an early warning system for weather risk management in the agricultural sector. The Rural Development Administration (RDA) agricultural weather observation network's wind velocity data and its corresponding estimated data from January to December 2020 were used to calculate linear regression equations (Y = aX + b). In each linear regression, the wind estimation error at 87 points and eight time slots per day (00:00, 03:00, 06:00, 09.00, 12.00, 15.00, 18.00, and 21:00) is the dependent variable (Y), while the estimated wind velocity is the independent variable (X). When the correlation coefficient exceeded 0.5, the regression equation was used as the wind velocity correction equation. In contrast, when the correlation coefficient was less than 0.5, the mean error (ME) at the corresponding points and time slots was substituted as the correction value instead of the regression equation. To enable the use of wind velocity model at a national scale, a distribution map with a grid resolution of 250 m was created. This objective was achieved b y performing a spatial interpolation with an inverse distance weighted (IDW) technique using the regression coefficients (a and b), the correlation coefficient (R), and the ME values for the 87 points and eight time slots. Interpolated grid values for 13 weather observation points in rural areas were then extracted. The wind velocity estimation errors for 13 points from January to December 2019 were corrected and compared with the system's values. After correction, the mean ME of the wind velocities reduced from 0.68 m/s to 0.45 m/s, while the mean RMSE reduced from 1.30 m/s to 1.05 m/s. In conclusion, the system's wind velocities were overestimated across all time slots; however, after the correction model was applied, the overestimation reduced in all time slots, except for 15:00. The ME and RMSE improved b y 33% and 19.2%, respectively. In our system, the warning for wind damage risk to crops is driven by the daily maximum wind speed derived from the daily mean wind speed obtained eight times per day. This approach is expected to reduce false alarms within the context of strong wind risk, by reducing the overestimation of wind velocities.

• Journal of the Korean earth science society
• /
• v.36 no.2
• /
• pp.139-157
• /
• 2015

We investigated the relative errors of satellite-observed Surface Skin Temperature (SST) data caused by sea ice in the northern hemispheric ocean ($30-90^{\circ}N$) during April 16-24, 2003-2014 by intercomparing MODerate Resolution Imaging Spectroradiometer (MODIS) Ice Surface Temperature (IST) data with two types of Atmospheric Infrared Sounder (AIRS) SST data including one with the AIRS/Advanced Microwave Sounding Unit-A (AMSU) and the other with 'AIRS only'. The MODIS temperatures, compared to the AIRS/AMSU, were systematically up to ~1.6 K high near the sea ice boundaries but up to ~2 K low in the sea ice regions. The main reason of the difference of skin temperatures is that the MODIS algorithm used infrared channels for the sea ice detection (i.e., surface classification), while microwave channels were additionally utilized in the AIRS/AMSU. The 'AIRS only' algorithm has been developed from NASA's Goddard Space Flight Center (NASA/GSFC) to prepare for the degradation of AMSU-A by revising part of the AIRS/AMSU algorithm. The SST of 'AIRS only' compared to AIRS/AMSU showed a bias of 0.13 K with RMSE of 0.55 K over the $30-90^{\circ}N$ region. The difference between AIRS/AMSU and 'AIRS only' was larger over the sea ice boundary than in other regions because the 'AIRS only' algorithm utilized the GCM temperature product (NOAA Global Forecast System) over seasonally-varying frozen oceans instead of the AMSU microwave data. Three kinds of the skin temperatures consistently showed significant warming trends ($0.23-0.28Kyr^{-1}$) in the latitude band of $70-80^{\circ}N$. The systematic disagreement among the skin temperatures could affect the discrepancies of their trends in the same direction of either warming or cooling.

• Journal of Korea Water Resources Association
• /
• v.54 no.12
• /
• pp.1243-1254
• /
• 2021

For water resources operation or agricultural water management, it is important to accurately predict evapotranspiration for a long-term future over a seasonal or monthly basis. In this study, reference evapotranspiration forecast (up to 12 months in advance) was performed using statistically predicted monthly temperatures and temperature-based Hamon method for the Han River basin. First, the daily maximum and minimum temperature data for 15 meterological stations in the basin were derived by spatial-temporal downscaling the monthly temperature forecasts. The results of goodness-of-fit test for the downscaled temperature data at each site showed that the percent bias (PBIAS) ranged from 1.3 to 6.9%, the ratio of the root mean square error to the standard deviation of the observations (RSR) ranged from 0.22 to 0.27, the Nash-Sutcliffe efficiency (NSE) ranged from 0.93 to 0.95, and the Pearson correlation coefficient (r) ranged from 0.97 to 0.98 for the monthly average daily maximum temperature. And for the monthly average daily minimum temperature, PBIAS was 7.8 to 44.7%, RSR was 0.21 to 0.25, NSE was 0.94 to 0.96, and r was 0.98 to 0.99. The difference by site was not large, and the downscaled results were similar to the observations. In the results of comparing the forecasted reference evapotranspiration calculated using the downscaled data with the observed values for the entire region, PBIAS was 2.2 to 5.4%, RSR was 0.21 to 0.28, NSE was 0.92 to 0.96, and r was 0.96 to 0.98, indicating a very high fit. Due to the characteristics of the statistical models and uncertainty in the downscaling process, the predicted reference evapotranspiration may slightly deviate from the observed value in some periods when temperatures completely different from the past are observed. However, considering that it is a forecast result for the future period, it will be sufficiently useful as information for the evaluation or operation of water resources in the future.

• Management & Information Systems Review
• /
• v.30 no.2
• /
• pp.237-255
• /
• 2011

This empirical study is focused on practical application of Range-Based Volatility which is estimated by opening, high, low, closing price of overall asset. Especially proper forecasting period is what I want to know. There is four useful Range-Based Volatility(RV) such as Parkinson(1980; PK), Garman and Klass(1980; GK) Rogers and Satchell(1991; RS), Yang and Zhang(2008; YZ). So, four RV of KOPSI 200 index during 2000.5.22-2009.9.18 was used for empirical test. The emprirical result as follows. First, the best RV which shows the best forecasting performance is PK volatility among PK, GK, RS, YZ volatility. According to estimating period forcasting performance of RV shows delicate difference. PK has better performance in the period with financial crisis of sub-prime mortgage loan. if not, RS is better. Second, almost result shows better performance on forecasting volatility without sub-prime mortgage loan period. so we can say that forecasting performance is lower when historical volatiltiy is comparatively high. Finally, I find that longer estimating period in AR(1) and MA(1) model can reduce forecasting error. More interesting point is that the result shows rapid decrease form 60 days to 90 days and there is no more after 90 days. So, if we forecast the volatility using Range-Based volaility it is better to estimate with 90 trading period or over 90 days.

• The Journal of Fisheries Business Administration
• /
• v.51 no.2
• /
• pp.51-69
• /
• 2020

The purpose of this study is to construct an outlook model that is consistent with the "Fisheries Outlook" monthly published by the Fisheries Outlook Center of the Korea Maritime Institute(KMI). In particular, it was designed as a partial equilibrium model limited to abalone items, but a model was constructed with a dynamic ecological equation model(DEEM) system taking into account biological breeding and shipping time. The results of this study are significant in that they can be used as basic data for model development of various items in the future. In this study, due to the limitation of monthly data, the market equilibrium price was calculated by using the recursive model construction method to be calculated directly as an inverse demand. A model was built in the form of a structural equation model that can explain economic causality rather than a conventional time series analysis model. The research results and implications are as follows. As a result of the estimation of the amount of young seashells planting, it was estimated that the coefficient of the amount of young seashells planting from the previous year was estimated to be 0.82 so that there was no significant difference in the amount of young seashells planting this year and last year. It is also meant to be nurtured for a long time after aquaculture license and limited aquaculture area(edge style) and implantation. The economic factor, the coefficient of price from last year was estimated at 0.47. In the case of breeding quantity, it was estimated that the longer the breeding period, the larger the coefficient of breeding quantity in the previous period. It was analyzed that the impact of shipments on the breeding volume increased. In the case of shipments, the coefficient of production price was estimated unelastically. As the period of rearing increased, the estimation coefficient decreased. Such result indicates that the expected price, which is an economic factor variable and that had less influence on the intention to shipments. In addition, the elasticity of the breeding quantity was estimated more unelastically as the breeding period increased. This is also correlated with the relative coefficient size of the expected price. The abalone supply and demand forecast model developed in this study is significant in that it reduces the prediction error than the existing model using the ecological equation modeling system and the economic causal model. However, there are limitations in establishing a system of simultaneous equations that can be linked to production and consumption between industries and items. This is left as a future research project.

• Journal of Korea Water Resources Association
• /
• v.43 no.1
• /
• pp.81-94
• /
• 2010

Input variable selection is one of the various techniques for improving the performance of artificial neural network. In this study, mutual information is applied for input variable selection technique instead of correlation coefficient that is widely used. Among 152 variables of RDAPS (Regional Data Assimilation and Prediction System) output results, input variables for artificial neural network are chosen by computing mutual information between rainfall records and RDAPS' variables. At first the rainfall forecast variable of RDAPS result, namely APCP, is included as input variable and the other input variables are selected according to the rank of mutual information and correlation coefficient. The input variables using mutual information are usually those variables about wind velocity such as D300, U925, etc. Several statistical error estimates show that the result from mutual information is generally more accurate than those from the previous research and correlation coefficient. In addition, the artificial neural network using input variables computed by mutual information can effectively reduce the relative errors corresponding to the high rainfall events.

• Journal of the Korean earth science society
• /
• v.38 no.7
• /
• pp.522-534
• /
• 2017

The purpose of this study is to verify urban flow and thermal environment by using the simulated Computational Fluid Dynamics (CFD) model in the area of Gangnam Seonjeongneung, and then to compare the CFD model simulation results with that of Seonjeongneung-monitoring networks observation data. The CFD model is developed through the collaborative research project between National Institute of Meteorological Sciences and Seoul National University (CFD_NIMR_SNU). The CFD_NIMR_SNU model is simulated using Korea Meteorological Administration (KMA) Local Data Assimilation Prediction System (LDAPS) wind and potential temperature as initial and boundary conditions from August 4-6, 2015, and that is improved to consider vegetation effect and surface temperature. It is noticed that the Root Mean Square Error (RMSE) of wind speed decreases from 1.06 to $0.62m\;s^{-1}$ by vegetation effect over the Seonjeongneung area. Although the wind speed is overestimated, RMSE of wind speed decreased in the CFD_NIMR_SNU than LDAPS. The temperature forecast tends to underestimate in the LDAPS, while it is improved by CFD_NIMR_SNU. This study shows that the CFD model can provide detailed and accurate thermal and urban area flow information over the complex urban region. It will contribute to analyze urban environment and planning.