4. DISCUSSION
The long-term change of GA and FG in the target watershed affected the watershed hydrology and caused the spatial and temporal decrease of TR. The two factors showed different influence on the watershed hydrology. In case of the decadal change of GA, GF has temporally decreased and GWR has temporally increased while other hydrological components were unaffected. The increasing groundwater use has lowered groundwater level and forced the overlying saturated zone to fill up the deficit which consequently decreased GF and increased GWR. In the aspect of FG, the increase of ET, a natural result of the vegetation growth, is the initial point of hydrologic response. The increased ET led to the decrease of SR and the mass of water infiltrated into the soil layer. Lessened infiltration then reduced the underground components including LF, PE, GF, and GWR. The series of hydrologic response from decadal GA and FG decreased TR in the target watershed as a result.
The significant decrease rate of hydrologic components in June and July is a corresponding result to the climate characteristic of South Korea that shows concentrated rainfall in Summer. According to the concentration of rainfall, cultivation under structure and the growth of vegetation in South Korea are focused on Summer, and the decadal change of GA and FG indicated that the cultivation and growth have been temporally developed. Thus, the hydrological components showed bigger response during Summer. By the way, the additional trend from October to March discovered from the fluctuation of GF can be explained by water curtain cultivation. The cultivation, one of the farming methods utilized during dry period, uses pumped groundwater to reserve water supply and keep vinyl house warm. Regarding that water curtain cultivation is usually performed from October to March in South Korea, and the site of water curtain cultivation has been largely expanded for the past four decades (Chung and Chang, 2016), we can infer that the trend of GF came from water curtain cultivation.
The analysis on flow-duration curves suggested two noticeable trends in terms of the annual streamflow condition of the target watershed. One is the temporal decrease of streamflow which noted that the decadal change of GA and FG has consistently worsened the annual streamflow condition. Another trend is related to the flow rate in different time duration. The flow rate in bigger time duration showed higher value of decrease percentage which indicated that the influence of GA and FG on the streamflow loss became stronger in lower flow rate.
The spatial vulnerability of TR loss in GA scenario and FG scenario followed the spatial increment of groundwater use (Figure 3) and the distribution of forested area in the study area (Figure 1. (c)) which proves that the model simulation made a good agreement with the actual change of groundwater use and forest growth. The vulnerability of TR loss showed that the subbasins with small streams are more vulnerable to GA and FG due to their smaller flow rate so that tributaries and branches of mainstream are more vulnerable to stream drying phenomena. By the way, the spatial vulnerability in GA scenario showed higher degree of streamflow loss than FG scenario. The decadal record of groundwater use showed increase of 299.7 106m3 in the 2010s since the 1980s. From the record, agricultural use from groundwater showed the highest increase of 149.1 106m3 which accounts for 49.8% of the total increment. Furthermore, the monthly hydrologic responses graphically displayed that the GF in June and July has decreased the most due to the water use in farming season. As a result, it can be concluded that the agricultural water use from groundwater was the most influential factor that has decreased streamflow in the target watershed for the last four decades.