• Journal of Korea Water Resources Association
• /
• v.40 no.11
• /
• pp.887-898
• /
• 2007

A physically based semi-distributed model, SWAT was applied to the Chungju Dam upstream watershed in order to investigate the spatial and temporal characteristics of watershed sediment yields. For this, general features of the SWAT and sediment simulation algorithm within the model were described briefly, and watershed sediment modeling system was constructed after calibration and validation of parameters related to the runoff and sediment. With this modeling system, temporal and spatial variation of soil loss and sediment yields according to watershed scales, land uses, and reaches was analyzed. Sediment yield rates with drainage areas resulted in $0.5{\sim}0.6ton/ha/yr$ excluding some upstream sub-watersheds and showed around 0.51 ton/ha/yr above the areas of $1,000km^2$. Annual average soil loss according to land use represented the higher values in upland areas, but relatively lower in paddy and forest areas which were similar to the previous results from other researchers. Among the upstream reaches, Pyeongchanggang and Jucheongang showed higher sediment yields which was thought to be caused by larger area and higher fraction of upland than other upstream sub-areas. Monthly sediment yields at the main outlet showed same trend with seasonal rainfall distribution, that is, approximately 62% of annual yield was generated during July to August and the amount was about 208 ton/yr. From the results, we could obtain the uniform value of sediment yield rate and could roughly evaluate the effect of soil loss with land uses, and also could analyze the temporal and spatial characteristics of sediment yields from each reach and monthly variation for the Chungju Dam upstream watershed.

• Economic and Environmental Geology
• /
• v.56 no.2
• /
• pp.139-153
• /
• 2023

This study aimed to understand the spatiotemporal variations in nitrogen content in the Gyeongan stream along the main stream and at the discharge points of the sub-basins, and to identify the origin of the nitrogen. Field surveys and laboratory analyses, including chemical compositions and isotope ratios of nitrate and boron, were performed from November 2021 to November 2022. Based on the flow duration curve (FDC) derived for the Gyeongan stream, the dry season (mid-December 2021 to mid-June 2022) and wet season (mid-June to early November 2022) were established. In the dry season, most samples had the highest total nitrogen(T-N) concentrations, specifically in January and February, and the concentrations continued to decrease until May and June. However, after the flood season from July to September, the uppermost subbasin points (Group 1: MS-0, OS-0, GS-0) where T-N concentrations continually decreased were separated from the main stream and lower sub-basin points (Group 2: MS-1~8, OS-1, GS-1) where concentrations increased. Along the main stream, the T-N concentration showed an increasing trend from the upper to the lower reaches. However, it was affected by those of the Osan-cheon and Gonjiamcheon, the tributaries that flow into the main stream, resulting in respective increases or decreases in T-N concentration in the main stream. The nitrate and boron isotope ratios indicated that the nitrogen in all samples originated from manure. Mechanisms for nitrogen inflow from manure-related sources to the stream were suggested, including (1) manure from livestock wastes and rainfall runoff, (2) inflow through the discharge of wastewater treatment plants, and (3) inflow through the groundwater discharge (baseflow) of accumulated nitrogen during agricultural activities. Ultimately, water quality management of the Gyeongan stream basin requires pollution source management at the sub-basin level, including its tributaries, from a regional context. To manage the pollution load effectively, it is necessary to separate the hydrological components of the stream discharge and establish a monitoring system to track the flow and water quality of each component.

• Journal of Korea Water Resources Association
• /
• v.57 no.7
• /
• pp.437-449
• /
• 2024

Rainfall-runoff prediction studies using deep learning while considering catchment attributes have been gaining attention. In this study, we selected two models: the Transformer model, which is suitable for large-scale data training through the self-attention mechanism, and the LSTM-based multi-state-vector sequence-to-sequence (LSTM-MSV-S2S) model with an encoder-decoder structure. These models were constructed to incorporate catchment attributes and predict the inflow of 10 multi-purpose dam watersheds in South Korea. The experimental design consisted of three training methods: Single-basin Training (ST), Pretraining (PT), and Pretraining-Finetuning (PT-FT). The input data for the models included 10 selected watershed attributes along with meteorological data. The inflow prediction performance was compared based on the training methods. The results showed that the Transformer model outperformed the LSTM-MSV-S2S model when using the PT and PT-FT methods, with the PT-FT method yielding the highest performance. The LSTM-MSV-S2S model showed better performance than the Transformer when using the ST method; however, it showed lower performance when using the PT and PT-FT methods. Additionally, the embedding layer activation vectors and raw catchment attributes were used to cluster watersheds and analyze whether the models learned the similarities between them. The Transformer model demonstrated improved performance among watersheds with similar activation vectors, proving that utilizing information from other pre-trained watersheds enhances the prediction performance. This study compared the suitable models and training methods for each multi-purpose dam and highlighted the necessity of constructing deep learning models using PT and PT-FT methods for domestic watersheds. Furthermore, the results confirmed that the Transformer model outperforms the LSTM-MSV-S2S model when applying PT and PT-FT methods.