Improving the simulations of the hydrological model in the karst catchment by integrating the conceptual model with machine learning models

Hydrological modelling can be complex in nonhomogeneous catchments with diverse geological, climatic, and topographic conditions. In this study, an integrated conceptual model including the snow module with machine learning modelling approaches was implemented for daily rainfall-runoff modelling in mostly karst Ljubljanica catchment, Slovenia, which has heterogeneous characteristics and is potentially exposed to extreme events that make the modelling process more challenging and crucial. In this regard, the conceptual model CemaNeige Génie Rural à 6 paramètres Journalier (CemaNeige GR6J) was combined with machine learning models, namely wavelet-based support vector regression (WSVR) and wavelet-based multivariate adaptive regression spline (WMARS) to enhance modelling performance. In this study, the performance of the models was comprehensively investigated, considering their ability to forecast daily extreme runoff. Although CemaNeige GR6J yielded a very good performance, it overestimated low flows. The WSVR and WMARS models yielded poorer performance than the conceptual and hybrid models. The hybrid model approach improved the performance of the machine learning models and the conceptual model by revealing the linkage between variables and runoff in the conceptual model, which provided more accurate results for extreme flows. Accordingly, the hybrid models improved the forecasting performance of the maximum flows up to 40 % and 61 %, and minimum flows up to 73 % and 72 % compared to the CemaNeige GR6J and stand-alone machine learning models. In this regard, the hybrid model approach can enhance the daily rainfall-runoff modelling performance in nonhomogeneous and karst catchments where the hydrological process can be more complicated.

Migration and transformation of soil mercury in a karst region of southwest China: Implications for groundwater contamination

Guizhou Province is located in the heart of a karst zone in southwest China, which is one of the largest karst areas in the world. Given the fragile surface ecosystem and highly developed underground karst structure, the migration and transformation of soil Hg may impact groundwater quality in karst environments with high Hg background concentrations. This study examines the vertical migration and transformation of soil mercury (Hg) in two karst catchments, Huilong and Chenqi, with the former containing high Hg contents associated with mineralization and the latter representing regional background Hg. The results show that the soil Hg pool in the Huilong catchment was as high as 44.4 ± 4.2 g m^-2, whereas in the Chenqi catchment was only 0.17±0.02 g m^-2. Compared with farmland soil, forest soil showed a significant loss of Hg. The results of L₃ X-ray absorption near edge structure of Hg indicated that α-HgS, the primary mineral of Hg ore, gradually changed to other mineral types during soil formation. In Huilong catchment, the proportion of organic bound Hg(SR)₂ out of total Hg decreased from 44.0% to 20.3% when soil depth increased from 10 cm to 160 cm in farmland soil profile and from 39.3% to 34.5% in forest soil profile, while the proportion of ionic Hg increased with soil depth, from 4.2% to 10.7% in the farmland soil profile and from 6.7% to 11.6% in the forestland soil profile. Results from the triple-mixing isotope model show that soil Hg accounts for more than 80% Hg in groundwater in the two catchments. Results from this study indicate potential risks of soil Hg entering into groundwater in this karst area.

Significant mercury efflux from a Karst region in Southwest China - Results from mass balance studies in two catchments

Karst regions have long been recognised as landscapes of ecological vulnerability, however the mass balance and fate of mercury (Hg) in karst regions have not been well documented. This study focused on the largest contiguous karst area in China and investigated Hg mass balance in two catchments, one with high geological Hg (Huilong) and the other representative of regional background Hg (Chenqi). The mass balance of Hg was calculated separately for the two catchments by considering Hg in throughfall, open field precipitation, total suspended particulate matter (TSP), litterfall, fertilizer, crop harvesting, air-surface Hg⁰ exchange, surface runoff and underground runoff. Results show that litterfall Hg deposition is the largest loading (from atmosphere) of Hg in both catchments, accounting for 61.5% and 38.5% of the total Hg input at Huilong and Chenqi, respectively. Air-surface Hg⁰ exchange is the largest efflux, accounting for 71.7% and 44.6% of the total Hg output from Huilong and Chenqi, respectively. Because both catchments are subject to farm and forest land use, cultivation plays an important role in shaping Hg fate. Mercury loading through fertilizer was ranked as the second largest input (28.5%) in Chenqi catchment and Hg efflux through crop harvest was ranked as the second largest output pathway in both Huilong (27.0%) and Chenqi (52.9%). The net Hg fluxes from the catchments are estimated to be 1498 ± 1504 μg m^-2 yr^-1 and 4.8 ± 98.2 μg m^-2 yr^-1. The significantly greater magnitude of net Hg source in Huilong is attributed to higher air-surface Hg⁰ exchange. The output/input ratio of Hg in this study was much greater than has been reported for other forest or agricultural ecosystems and indicates that the karst region of Southwest China is a significant source of atmospheric Hg. The results of this study should be considered in the development of pollution control policies which seek to conserve fragile karst ecosystems characterised by high geological background of Hg.

Source and flux of anthropogenically enhanced dissolved inorganic carbon: A comparative study of urban and forest karst catchments in Southwest China

Anthropogenic disturbance has become an accelerator of dissolved inorganic carbon flux (F_DIC), but the magnitude by which F_DIC is enhanced by various anthropogenic activities, especially urbanization, remains largely unknown. In this study, rainfall, discharge, pH, and specific conductivity were continuously monitored during a one-year period in two adjacent karst catchments affected by anthropogenic activities in Southwest China. Groundwater, rainwater, and untreated sewage were collected monthly, and their hydrogeochemical characteristics and isotopic compositions (δ¹³C_DIC, δ³⁴S_SO4, δ¹⁸O_SO4, δ¹⁵N_NO3 and δ¹⁸O_NO3) were measured. The results showed that the urban catchment exported 70% more F_DIC than its forest counterpart, with considerable differences in the concentration and isotopic composition of DIC, SO₄^2- and NO₃^- between the two subterranean streams. Based on the contribution by various SO₄^2- or NO₃^- sources estimated with a Bayesian isotope mixing model, carbonate weathering by H₂CO₃, H₂SO₄ and HNO₃ contributed 67.5%, 26.0% and 6.5% of F_DIC in the urban catchment and 93.3%, 3.4% and 3.3% in the forest catchment, respectively. Anthropogenic activities enhanced the F_DIC by 24.0% in the urban catchment and 5.8% in the forest catchment. Anthropogenically enhanced DIC flux (F_DIC-A) primarily arose from carbonate dissolution due to acid precipitation (especially S deposition), which accounted for approximately 64% of the F_DIC-A in the two catchments. Sewage (including manure) and fertilizer applications also enhanced carbonate dissolution through nitrification processes. Furthermore, F_DIC-A was much higher (>83%) during the wet and warm season (May to October) in both catchments. The results of this study suggest that anthropogenic impacts on carbonate dissolution should be considered in global carbon cycle assessments, especially in urban areas.

Assessing soil degradation under land-use change: insight from soil erosion and soil aggregate stability in a small karst catchment in southwest China

Background

Soil erodibility (K factor) and soil aggregate stability are often used to assess soil degradation in an erodible environment. However, their applicability under land-use change is uncertain, especially agricultural abandonment.

Methods

Different land-use types, including cropland, abandoned cropland, and native vegetation land, were converted into the successive stages following agricultural abandonment by space-for-time substitution approach in a small karst catchment, Southwest China. The indexes of soil aggregate stability and K factor of the Erosion Productivity Impact Calculator (EPIC) model in soil profiles were calculated to identify which method is suitable to indicate soil degradation under land-use change.

Results

The indexes of soil aggregate stability in the soils at 0∼30 cm depth under native vegetation land were significantly larger than those under cropland and slightly larger than those under abandoned cropland. The K factor was not significantly different among the three land use examples because the EPIC model does not consider soil permeability. In the soil organic carbon (SOC)-rich soils (>2%), the K factor was significantly correlated with silt and clay content ranging within a narrow scope of near 0.010 t hm² h/hm²/MJ/mm. While in the SOC-poor soils, the K factor was significantly increased with decreasing SOC content and was significantly correlated with soil aggregate stability.

Conclusions

Soil aggregate stability is more suitable to indicate soil degradation under land-use change. Sufficient SOC in erodible soils would restrain soil degradation, while SOC loss can significantly increase soil erosion risk.