USING AUTOMATIC CAVE-DRIP MONITORING TO CHARACTERIZE GROUNDWATER INFILTRATION AT NATURAL BRIDGE CAVERNS IN TEXAS

In south-central Texas, the karst groundwater of the Edwards and Trinity aquifers is an important water source. Natural Bridge Caverns (NBC) is located within the Edwards (Balcones Fault Zone) Aquifer recharge area and Trinity Aquifer contributing zone. The caverns are situated within the early Cretaceous limestone, marl, and shale units of the upper Glen Rose Formation and the lower Kainer Formation and provide a suitable environment to study subsurface water flow paths.

This study evaluates water infiltration through limestone formations at the caverns, over one hydrological year using automatic drip rate logging systems. Twenty loggers were installed at actively dripping sites, throughout the Castle of the White Giants and the Hall of the Mountain King chambers and the connecting passageways, covering a range of elevation gradients. The drip time series was evaluated with respect to rainfall, soil water content, and observed flow patterns, using a series of investigative methods including Functional principal component analysis (FPCA), multidimensional scaling (MDS), and k-means clustering. Similarities between drip time series are interpreted in terms of cave chamber morphology, host lithology, and overburden thickness.

Analysis reveals significant findings related to surface-subsurface interactions. Some loggers display correlated (instantaneous or delayed) responses to rainfall events, suggesting the presence of ample fractures. Others indicate that cave drip is not directly connected to rainfall but driven by overflow of stored water in the overlying limestone or groundwater flow in the vadose zone. A comparison of limestone thickness found that drip rate did not increase with depth nor was there any observable relationship between overburden, coefficient of variation (COV) and skewness, indicating that variance does not change with depth. However, all sites were positively skewed, suggesting that the dataset reflects steady discharge at lower rates interrupted by smaller intervals of elevated discharge. These results highlight the heterogeneous nature of the relationship between surface infiltration, drip discharge and the response to hydrological factors.