Integrating Data Science and Biogeochemistry: Inductive-Deductive Analysis of Carbon and Nutrients
Examples of recent work
Adler T, Underwood KL, Rizzo DM, Harpold AA, Sterle G, Li L, Wen H, Stinson L, Bristol C, Shanley J, Lini A, Perdrial N, Perdrial, J.N. (2021). Drivers of Dissolved Organic Carbon (DOC) in Forested Headwater Catchments: A Multi Scale Approach. Frontiers in Biogeoscience. https://doi.org/10.3389/frwa.2021.578608
We did a full pattern-process iteration for this study authored by Thomas Adler, a MS student in my group: for pattern investigation, we analyze long-term trends of flow-adjusted dissolved organic carbon (DOC) concentrations and wet deposition trends using Seasonal Kendall tests in USGS headwater catchments, comparing these trends to catchment attributes. Our analysis suggests that soil depth and catchment connectivity modulates DOC concentrations.
For process investigation, we experimentally simulate increased and decreased acid deposition on soil cores from catchments with contrasting long-term DOC responses. Results indicate that soils from catchments with long-term increases in DOC release more DOC, and scanning electron microscope imaging reveals a significant DOC contribution from destabilized soil aggregates, but only in the location where where stream DOC is still increasing. |
|
Underwood KL, Rizzo DM, Hanley JP, Sterle G, Harpold AA, Adler T, Li L, Wen H, Perdrial, J.N. 2023. Machine-Learning Reveals Equifinality in Drivers of Stream DOC Concentration at Continental Scales. Water Resources Research. 59 (3). https://doi.org/10.1029/2021WR030551.
To better understand patterns of dissolved organic carbon (DOC) in streams, we used an inductive approach where we applied a novel evolutionary algorithm (EA) to identify geologic, topographic, meteorologic, hydrologic, and land use attributes correlated with mean stream DOC concentration across forested catchments in the continental United States. The EA reduced the dimensionality of our attribute dataset, revealing the combination of factors and their specific value ranges driving membership in High or Low mean DOC clusters. High mean DOC concentrations were linked to two distinct geographic locations with varying climatic and vegetative conditions, highlighting equifinality.
|
Our findings emphasize the critical role of the critical zone structure in moderating hydrological and biogeochemical processes that govern DOC dynamics at the catchment scale.
|
Ruckhaus, M., Perdrial, J.N., Underwood, K.L., Stewart, B., Kincaid, D., Seybold, E., Li, L., Shanley, J., 2023. Disentangling the separate and intersecting pathways of soil and stream water carbon and nitrogen in response to atmospheric drivers. Frontiers in Water. 5. https://doi.org/10.3389/frwa.2023.1065300.
Another example is the investigation of dissolved organic carbon (DOC), dissolved organic nitrogen (DON), and dissolved inorganic nitrogen (DIN) levels in the stream in combination with soil leaching experiments simulating high and low acid deposition. This study was authored by Manya Ruckhaus, a MS student in my group. Manya used data from a stream in a Vermont catchment and we found DOC and DON concentrations typically vary together over time. Winter soil leachates had the highest concentrations of these solutes, which aligns with spring snowmelt flushing organic materials into streams. Interestingly, acid deposition had different effects on DOC/DON compared to DIN. Low acid deposition (similar to present-day precipitation) led to higher DOC/DON concentrations, while high acid deposition had the opposite effect, increasing DIN concentrations. These trends match the increasing DOC and decreasing DIN levels observed in the long-term stream data from 1991 to 2018.
|
|
Seybold, E.C., Kincaid, D., Musselman, K., Schroth, A., Adair, C., Perdrial, J.N., Dwivedi, R., Classen, A. (2022). Winter runoff events pose an unquantified continental-scale risk of high wintertime nutrient export. Environmental Research Letters. DOI 10.1088/1748-9326/ac8be5
Manyas work connects to work at larger scales led by Erin Seybold (KU). Here we looked at data from different sources across the contiguous United States. This work shows that especially winter events, such as rain-on-snow cause significant flooding and supplies a large portion of the country's nitrogen and phosphorus to groundwater and surface water. Importantly, this work emphasizes that our previous assumptions about low water flow and nutrient transport in snowy regions during winter may not be accurate anymore.
|
|
