• Journal of Wetlands Research
• /
• v.13 no.1
• /
• pp.105-116
• /
• 2011

Changes in climate have largely increased concentrated heavy rainfall, which in turn is causing enormous damages to humans and properties. Therefore, the exact relationship and the spatial variability analysis of hydrometeorological elements and characteristic factors is critical elements to reduce the uncertainty in rainfall -runoff model. In this study, radar rainfall grid resolution and grid resolution depending on the topographic factor in rainfall - runoff models were how to respond. In this study, semi-distribution of rainfall-runoff model using the model ModClark of Inje, Gangwon Naerin watershed was used as Gwangdeok RADAR data. The completed ModClark model was calibrated for use DEM of cell size of 30m, 150m, 250m, 350m was chosen for the application, and runoff simulated by the RADAR rainfall data of 500m, 1km, 2km, 5km, 10km from 14 to 17 on July, 2006. According to the resolution of each grid, in order to compare simulation results, the runoff hydrograph has been made and the runoff has also been simulated. As a result, it was highly runoff simulation if the cell size is DEM 30m~150m, RADAR rainfall 500m~2km for peak flow and runoff volume. In the statistical analysis results, if every DEM cell size are 500m and if RADAR rainfall cell size is 30m, relevance of model was higher. Result of sensitivity assessment, high index DEM give effect to result of distributed model. Recently, rainfall -runoff analysis is used lumped model to distributed model. So, this study is expected to make use of the efficiently decision criteria for configurated models.

• Journal of Korea Water Resources Association
• /
• v.48 no.9
• /
• pp.719-728
• /
• 2015

In this study, urban runoff analyses were performed using high resolution Quantitative Precipitation Estimation (QPE), and variation of rainfall and runoff were analyzed to evaluate QPE data for urban runoff analysis. The five drainage districts (Seocho3, 4, 5, Yeoksam and Nonhyun) around Gangnam station were chosen as study area, the area is $7.4km^2$. Rainfall data from KMA AWS (34 stations), SKP AWS (156 stations) and Gwanduk radar were used for QPEs in Seoul area. Four types of QPE(QPE1: KMA AWS, QPE2: KMA+ SKP AWS, QPE3: Gwangduk radar, QPE4: QPE2+QPE3) of 6 events in July 2013 were generated by using Krigging and conditional merging. The temporal and spatial resolution of QPEs are 10 minutes and 250 m, respectively. The complex pipe network were treated as 773 manholes, 772 sub-drainage districts and 1,059 pipelines for urban runoff analysis as input data. QPE2 and QPE4 show spatial variation of rainfall by sub-drainage districts as 1.9 times bigger than QPE1. The peak runoff of QPE2 and QPE4 also show spatial variation as 6 times bigger than Gangnam and Seocho AWS. Thus, the spatial variation of rainfall and runoff could exist in small area such as this study area, and using high-resolution rainfall data is desirable for accurate urban runoff analysis.

• Journal of Korea Water Resources Association
• /
• v.53 no.12
• /
• pp.1159-1172
• /
• 2020

This study applied deep convolution neural network based on U-Net and SegNet using long period weather radar data to very short-term rainfall prediction. And the results were compared and evaluated with the translation model. For training and validation of deep neural network, Mt. Gwanak and Mt. Gwangdeoksan radar data were collected from 2010 to 2016 and converted to a gray-scale image file in an HDF5 format with a 1km spatial resolution. The deep neural network model was trained to predict precipitation after 10 minutes by using the four consecutive radar image data, and the recursive method of repeating forecasts was applied to carry out lead time 60 minutes with the pretrained deep neural network model. To evaluate the performance of deep neural network prediction model, 24 rain cases in 2017 were forecast for rainfall up to 60 minutes in advance. As a result of evaluating the predicted performance by calculating the mean absolute error (MAE) and critical success index (CSI) at the threshold of 0.1, 1, and 5 mm/hr, the deep neural network model showed better performance in the case of rainfall threshold of 0.1, 1 mm/hr in terms of MAE, and showed better performance than the translation model for lead time 50 minutes in terms of CSI. In particular, although the deep neural network prediction model performed generally better than the translation model for weak rainfall of 5 mm/hr or less, the deep neural network prediction model had limitations in predicting distinct precipitation characteristics of high intensity as a result of the evaluation of threshold of 5 mm/hr. The longer lead time, the spatial smoothness increase with lead time thereby reducing the accuracy of rainfall prediction The translation model turned out to be superior in predicting the exceedance of higher intensity thresholds (> 5 mm/hr) because it preserves distinct precipitation characteristics, but the rainfall position tends to shift incorrectly. This study are expected to be helpful for the improvement of radar rainfall prediction model using deep neural networks in the future. In addition, the massive weather radar data established in this study will be provided through open repositories for future use in subsequent studies.