Spatially Explicit, Regional‐Scale Simulation of Lake Carbon Fluxes

Abstract

Lakes are areas of intense biogeochemical processing in the landscape, contributing significantly to the global carbon cycle despite their small areal coverage. However, current large‐scale estimates of lake biogeochemical fluxes are all generated by multiplying a mean observed areal rate by regional or global lake surface area, which ignores important heterogeneous spatial and temporal processes that regulate lake carbon cycling. We have developed a process‐based model that integrates core scientific knowledge in hydrology, biogeochemistry, and ecology that is specifically designed to be applied over large geographic regions to hindcast or forecast regional lake carbon fluxes. We used our model to simulate daily carbon fluxes and pools for 3,675 lakes in the Northern Highlands Lake District from 1980–2010 and produced spatial and seasonal patterns consistent with observations. Variabilities in lake carbon fluxes were well predicted by relatively simple hydrologic metrics, such as the fraction of hydrologic export as evaporation (FHEE). Overall, lakes with a high FHEE processed a greater percentage of carbon inputs in the simulations than lakes with a low FHEE, but low‐FHEE lakes ultimately processed more total carbon because of greater carbon inputs. Large lakes with low FHEE and high external loading of dissolved inorganic carbon contributed most to total CO2 emissions for the Northern Highlands Lake District, and our model estimated that 78% of total CO2 emissions from lakes to the atmosphere originated from external loads of dissolved inorganic carbon. By better characterizing the unique biogeochemical processes for each individual lake, regional estimates of carbon fluxes are more accurately determined.

Integrated, Regional‐Scale Hydrologic Modeling of Inland Lakes

“…A direct application of our integrated modeling framework to the ecological field has been detailed in Zwart et al. (), which used the hydrologic fluxes simulated in this study to investigate lake carbon processing for the same set of inland lakes. In this study, we also showed that our model can easily take advantage of new data sources to improve model performance.…”

“…Our final dataset included 3,692 lakes and reservoirs within the NHLD region (see Zwart et al. for more information on dataset inspection). We simulated each lake one time within its respective subdomain across the modeling time period and synthesized the results of all subdomains into a single regional dataset for subsequent analysis of the NHLD.…”

“…To avoid problems, we manually inspected each lake and reservoir within the NHLD, comparing the NHD classification with Google Earth Imagery and the World Imagery Basemap in ArcGIS. Our final dataset included 3,692 lakes and reservoirs within the NHLD region (see Zwart et al 2018 for more information on dataset inspection). We simulated each lake one time within its respective subdomain across the modeling time period and synthesized the results of all subdomains into a single regional dataset for subsequent analysis of the NHLD.…”

Cross‐Scale Interactions Dictate Regional Lake Carbon Flux and Productivity Response to Future Climate

“…The lake heat budget, constituent load, and biogeochemical models were driven by the hydrologic output from the coupled surface water and groundwater model and parameterized exactly as described in Zwart et al (2018) with altered driving data from the climate change scenarios as described above. Our lake heat budget model used mass balance equations (Lenters et al, 2005) and functions from the R package LakeMetabolizer to model lake water temperature during the open-water period as in Zwart et al (2018). The constituent loading model uses land cover, long-term data, and simulated water budgets from our coupled hydrologic model to estimate inflowing dissolved organic carbon (DOC), dissolved inorganic carbon (DIC), terrestrial particulate organic matter, and dissolved inorganic phosphorus to each lake Zwart et al (2018).…”

“…Our lake heat budget model used mass balance equations (Lenters et al, 2005) and functions from the R package LakeMetabolizer to model lake water temperature during the open-water period as in Zwart et al (2018). The constituent loading model uses land cover, long-term data, and simulated water budgets from our coupled hydrologic model to estimate inflowing dissolved organic carbon (DOC), dissolved inorganic carbon (DIC), terrestrial particulate organic matter, and dissolved inorganic phosphorus to each lake Zwart et al (2018). For this analysis, we focused on the impact of climate change scenarios on lake C fluxes and productivity.…”