Impact of land-use management and climate variability on hydrological responses of representative groundwater spring typologies in Eastern Himalaya

In a first-of-its-kind study from Eastern Himalaya, we analysed the relative controls of vegetation, precipitation, soil properties, and hydrogeology on the diurnal and seasonal variability in three representative groundwater springs typologies using high-resolution discharge data. The three springs, Gaddi- a Fracture spring with forest land-use, Mamley - a Karst spring with forest and agriculture land-use and Kamrang - a Depression spring with agriculture land-use, together provide water security to over 600 households in South District, Sikkim, India, and are managed by local community organisations.

Based on hydrogeological traverse mapping and master recession curve (MRC) analysis, we categorised the Kamrang as a high-discharge depression spring fed by a relatively homogenous aquifer with uniformly high transmissivity (T_s) and storage (S_t) displaying a gradually decreasing smooth recession curve. Conversely, Mamley has a relatively larger springshed area fed by a smaller homogenous aquifer with two components: a low T_s and low S_t component possibly situated in the upper phyllite-quartzite beds and a high T_s and low S_t sitting in the karst environment. Similarly, Gaddi has the largest springshed area covered with dense oak forests overlaying a nearby shallow aquifer with high T_s which empties faster than the main distant aquifer body with low T_s. All three springs were classified as highly variable (Coefficient of variability, C_v > 40 %). Annually, Kamrang (95±22 %) showed the highest variability followed by Gaddi (88±7 %) and Mamley (72±41 %). However, in winter Kamrang and Gaddi showed very stable flows (C_v > ~20 %) whereas Mamley had higher variability (C_v > 30%). In summers, Gaddi showed much higher fluctuations (C_v > 40%) than Kamrang and Mamley.

Strong yet contrasting diel fluctuations in discharge were observed with significantly higher amplitude in the depression spring (Kamrang, 19±16 l min^-1) and the fracture spring (Gaddi, 12±10 l min^-1) than in the karst spring (Mamley, 7±14 l min^-1). The daily troughs in diel discharge occurred early in Mamley (1600 h) followed by Gaddi (1700 h) and Kamrang (2000 h), largely attributed to daily evapotranspiration-related abstraction. Mamley recovered almost instantaneously compared to Kamrang and Gaddi, both of them peaking in the morning (1000 h). Among agriculture-dominated springsheds, relatively high T_s along with moderate saturated soil hydraulic conductivity (K_sat.soil) resulted in lower lag-time in Mamley than Kamrang, which had similarly high T_s but lower K_sat.soil. On the contrary, the lag-time was longest in forest-dominated Gaddi, with the lowest T_s and K_sat.soil. The forest land-use may also have influenced the contrasting observations of Gaddi spring discharge responding faster to high-intensity and low-moderate volume rainfall than agriculture-dominated Kamrang and Mamley. Thus, any land-use change negatively affecting K_sat.soil, such as compaction through grazing and topsoil erosion, is likely to have strong negative effects on the longevity of the spring. On the other hand, fracture springs like Gaddi, fed by a larger catchment, are likely to be immune from the effects of small-scale land-use changes. Our results suggest that any future changes in the precipitation patterns and land-use may significantly impact spring behaviour in the Himalaya.