Ph.D. Seminar: Rasika Ramesh, Maj. Prof, Dr. Latif Kalin

Title: Understanding wetland hydrology and water quality through data/process based modelling

Location: Classroom 1207

Date: Wednesday, April 11, 2018

Time: 9:00 a.m.

Abstract:

Rapid coastal development has led to loss/alteration of wetlands, streams, riparian vegetated areas and headwater areas that buffer coastal waterways from pollution. Small wetlands, besides being easily altered, have also shown to have higher capacity for nonpoint source amelioration. Consequently, the protection and restoration of small wetlands and their vegetated buffer systems are critical in regulating flows and enhancing water quality on the landscape. This requires good understanding of their functionality so that appropriate steps can be taken for their management and preservation. This study evaluates headwater slope wetlands in Alabama’s coastal plain using field data and process-based modeling, as well as improves existing relationships describing sediment removal by vegetated buffers through data-based modeling. Very little data exists for headwater slope wetlands (groundwater-fed wetlands above and alongside 1st order streams) in the region; to address this knowledge gap we observed hydrology and dissolved inorganic nitrogen (DIN) trends in select wetlands, addressed challenges associated with modeling their hydrology and lastly, identified nitrogen inputs pertinent to predicting nitrate export through a sensitivity analysis. Delineated watersheds were small (<2 sq km); flashy flows followed level of urbanization in the watershed, with the least altered wetland having stable and damped flows. Despite watershed alterations, wetlands still showed DIN load reductions ranging from 9% to 50%. One of the study wetlands showed unusually large flows, indicating the presence of a larger ground watershed relative to the extent the delineated surficial watershed; a common issue in coastal plain regions where topography is flat and water tables are shallow. Using this as a case study, we investigated different approaches of modelling flow using popular watershed model SWAT (Soil and Watershed Assessment Tool) as a simpler alternative to complex groundwater models. Since flows in SWAT are limited by watershed precipitation, simulated flows were several times smaller in magnitude than observed flows. Calibration approaches involved manual amplification of baseflow with a multiplier (ENASH = 0.66), tweaking parameter RCHRGE_DP to allow extra water to be added to the system (ENASH = 0.75), and incorporating ANN (Artificial Neural Network) with SWAT to further improve calibration performance (ENASH = 0.88). These approaches provide managers and modelers useful tools to navigate similar flow calibration challenges in other groundwater dominant watersheds. Since data for models aimed at understanding wetland function are especially scarce for smaller wetlands (e.g., headwater slope wetlands), optimizing data collection to include only those most valuable for model predictions is a pressing need. Taking the case of nitrate, we conducted a sensitivity analysis to assess if detailing surface inputs of organic nitrogen and ammonia (whose fluxes are linked with nitrate) were necessary to predict nitrate export from study headwater slope wetlands. Nitrate export, modelled by model WetQual, showed negligible sensitivity to organic nitrogen and ammonia inputs. Perhaps low residence times in study headwater slope wetlands, which are typically gaining wetlands with no depressional storage, afforded too little time for N transformations to effect nitrate export leading us to conclude that organic nitrogen and ammonia input data at high resolution are not as important as detailing nitrate inputs in low residence time, groundwater interacting wetlands such as headwater slope wetlands. Wetland management also involves revitalizing streamside vegetation which are crucial in mitigating nonpoint pollution, such as sediment pollution. With the objective of improving existing relationships describing sediment removal, we compiled data from 54 studies (including online BMP database) concerning sediment trapping by vegetated buffers and recorded buffer characteristics (such as buffer width, slope, area, vegetation type, sediment and runoff loading, runoff rates, residence time, roughness and sediment removal efficiency). An exponential regression model best described the relationship between sediment removal efficiency and volume ratio, residence time and width further increased (R2 = 40.5%). This model was compared with performances derived from applying other sediment reduction regression models reported in literature namely those in White and Arnold (2009), Liu et al. (2008) and Zhang et al. (2010) to our database. Of these, only the model presented by White and Arnold (2009) was statistically significant presumably because of the inclusion of runoff reduction in their study. The results of this study point towards the importance of considering flow in buffer design.