• Journal of the Korean association of regional geographers
• /
• v.16 no.4
• /
• pp.339-354
• /
• 2010

In this study, diurnal and intra-seasonal variations of summertime temperature lapse rate (TLR) by synoptic weather conditions in a mountainous basin are examined based on hourly temperature data observed in 2009 summer at an Automatic Weather Station (AWS) network deployed in Haean basin (called Punch Bowl), Yanggu in the Republic of Korea. Summertime average TLR between the top and bottom of the basin is $-0.53^{\circ}C$/100m. Due to its diurnal variations, TLR shows the lowest by $-0.25^{\circ}C$/100m at 6AM, while it maximizes up to $-0.85^{\circ}C$/100m between 4PM~5PM. Comparisons of daily average TLRs by synoptic weather patterns reveal that the magnitude of TLRs is greatest in the order of rainy days ($-0.63^{\circ}C$/100m), heavy rainfall days ($-0.53^{\circ}C$/100m), partly cloudy days ($-0.47^{\circ}C$/100m), and sunny days ($-0.39^{\circ}C$/100m). At dawn on sunny days in summer, strong cooling pools accompanying temperature inversion layers are formed within the basin, while on heavy rainfall days, warming pools are observed due to relatively low TLRs associated with the reduction of surface radiation cooling by clouds.

• Journal of Korea Water Resources Association
• /
• v.45 no.7
• /
• pp.631-641
• /
• 2012

In this study, satellite data (MTSAT-1R), a numerical weather prediction model, RDAPS (Regional Data Assimilation and Prediction System) output, ground weather station data, and artificial neural networks were used to improve the accuracy of summer rainfall forecasts. The developed model was applied to the Seoul station to forecast the rainfall at 3, 6, 9, and 12-hour lead times. Also to reflect the different weather conditions during the summer season which is related to the frontal precipitation and the cyclonic precipitation such as Jangma and Typhoon, the neural network models were formed for two different periods of June-July and August-September respectively. The rainfall forecast model was trained during the summer season of 2006 and 2008 and was verified for that of 2009 based on the data availability. The results demonstrated that the model allows us to get the improved rainfall forecasts until lead time of 6 hour, but there is still a large room to improve the rainfall forecast skill.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.13 no.4
• /
• pp.176-184
• /
• 2011

Cold air on sloping surfaces flows down to the valley bottom in mountainous terrain at calm and clear nights. Based on the assumption that the cold air flow may be the same as the water flow, current models estimate temperature drop by regarding the cold air accumulation at a given location as the water-like free drainage. At a closed catchment whose outlet is blocked by man-made obstacles such as banks and roads, however, the water-like free drainage assumption is no longer valid because the cold air accumulates from the bottom first. We developed an empirical model to estimate quantitatively the effect of cold pool on nocturnal temperature in a closed catchment. In our model, a closed catchment is treated like a "vessel", and a digital elevation model (DEM) was used to calculate the maximum capacity of the cold pool formed in a closed catchment. We introduce a topographical variable named "shape factor", which is the ratio of the cold air accumulation potential across the whole catchment area to the maximum capacity of the cold pool to describe the relative size of temperature drop at a wider range of catchment shapes. The shape factor is then used to simulate the density profile of cold pool formed in a given catchment based on a hypsometric equation. The cold lake module was incorporated with the existing model (i.e., Chung et al., 2006), generating a new model and predicting distribution of minimum temperature over closed catchments. We applied this model to Akyang valley (i.e., a typical closed catchment of 53 $km^2$ area) in the southern skirt of Mt. Jiri National Park where 12 automated weather stations (AWS) are operational. The performance of the model was evaluated based on the feasibility of delineating the temperature pattern accurately at cold pool forming at night. Overall, the model's ability of simulating the spatial pattern of lower temperature were improved especially at the valley bottom, showing a similar pattern of the estimated temperature with that of thermal images obtained across the valley at dawn (0520 to 0600 local standard time) of 17 May 2011. Error in temperature estimation, calculated with the root mean square error using the 10 low-lying AWSs, was substantially decreased from $1.30^{\circ}C$ with the existing model to $0.71^{\circ}C$ with the new model. These results suggest the feasibility of the new method in predicting the site-specific freeze and frost warning at a closed catchment.

• Korean Journal of Remote Sensing
• /
• v.36 no.5_1
• /
• pp.679-692
• /
• 2020

We coupled a CFD model to the WRF-Chem model (WRF-CFD model) and investigated the characteristics of flows and carbon monoxide (CO) distributions in a building-congested district. We validated the simulated results against the measured wind speeds, wind directions, and CO concentrations. The WRF-Chem model simulated the winds from southwesterly to southeasterly, overestimating the measured wind speeds. The statistical validation showed that the WRF-CFD model simulated the measured wind speeds more realistically than the WRF-Chem model. The WRF-Chem model significantly underestimated the measured CO concentrations, and the WRF-CFD model improved the CO concentration prediction. Based on the statistical validation results, the WRF-CFD model improved the performance in predicting the CO concentrations by taking complicatedly distributed buildings and mobiles sources of CO into account. At 04 KST on May 22, there was a downdraft around the AQMS, and airflow with a relatively low CO concentration was advected from the upper layer. Resultantly, the CO concentration was lower at the AQMS than the surrounding area. At 15 KST on May 22, there was an updraft around the AQMS. This resulted in a slightly higher CO concentration than the surroundings. The WRF-CFD model transported CO emitted from the mobile sources to the AQMS measurement altitude, well reproducing the measured CO concentration. At 18 KST on May 22, the WRF-CFD model simulated high CO concentrations because of high CO emission, broad updraft area, and an increase in turbulent diffusion cause by wind-shear increase near the ground.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.