Effects of rainfall variability and land cover change on groundwater recharge on a volcanic island (Jeju, Korea)

A GIS-based Soil-Water-Balance (SWB) model was used to estimate spatially distributed recharge across Jeju Island (Korea) for a variety of time periods, and climate and land cover scenarios. SWB is based on a modified Thornthwaite-Mather approach that calculates water balance components for each grid cell at a daily timestep. Rainfall input files were interpolated from daily rainfall measurements recorded at 52-gauges from 1992-2009. Net precipitation was estimated using a bucket model approach in which a daily initial interception storage capacity must be satisfied before precipitation can reach the soil surface. Interception losses were estimated from the literature for each land-use type and season (growing/non-growing). Snowfall was assumed to occur when the mean daily temperature minus one-third of the difference between the daily maximum and minimum temperature was less or equal to the freezing point of water. Snowmelt was calculated assuming that 1.5 mm of snow melts per day per degree Celsius when the daily maximum temperature is above the freezing point. Spatially variable potential evapotranspiration was calculated using the Hargreaves-Samani method which requires gridded minimum and maximum air temperature data for each time step. These were computed using temperature lapse rates calculated from daily temperature data recorded at 19 stations from 1992-2009. Surface runoff was modeled for each rain and snowmelt event using the Natural Resources Conservation Service curve number method. SWB incorporates overland flow routing to ensure that runoff from upslope grid cells either infiltrates soils or continues downslope to the steepest downgradient cell. Root zone depths and water-holding capacities of Jeju's hydrologic soil groups were used to compute maximum soil water storage capacities. Any excess water exited from the bottom of the grid cell as groundwater recharge. Calibration comprised the optimization of interception storage capacities and curve numbers to match estimated interception losses and measured direct runoff. Recharge was modeled for: (1) an 18-year (1992-2009) baseline period, (2) a 10-year drought scenario, and (3) an 18-year climate-land use change scenario. Current land use, soil and climate data was used for the baseline scenario. For the drought scenario, rainfall and temperature maps were scaled by daily rainfall and temperature data of an identified drought period (1963-1972). For the climate-land use change scenario, A2 emission scenario predictions of rainfall and temperature changes for the late 21^st century were used to scale baseline data. Additionally, the spatial abundance of impervious, urban land cover was increased. The baseline scenario produced the highest mean island recharge (1,605 x 10⁶ m³/yr), followed by the climate-land use change scenario (1,432 x 10⁶ m³/yr), and the drought scenario (1,073 x 10⁶ m³/yr). The results of each scenario represent a conservative estimate of actual recharge as SWB does not address recharge from fog drip, irrigation or artificial recharge injection.