2020-2024 Theses and dissertations

Full text at CU

Abstract

I present quantitative Pliocene-Pleistocene terrestrial tropical temperature estimates along with the refinement of organic and stable isotope geochemical proxies in the northern tropical Andes of Colombia. The Pliocene epoch, 5 to ~2.6 million years ago (Ma), is often cited as the last time when Earth’s mean temperature was ~2.5-4°C warmer than today, and CO2 concentrations may have been higher than preindustrial levels. Although the tropics play an important role in regulating global climate, a recent summary of Pliocene temperature terrestrial records includes no site within 10º of the equator. The Sabana de Bogotá in the Eastern Cordillera of Colombia offers unique sedimentary archives from the tropics (~4ºN), including sediment from an extinct lake preserved in the Funza-II core that dates back to late Pliocene, which allows an opportunity to apply quantitative geochemical proxies for temperature reconstructions.

Branched glycerol dialkyl glycerol tetraethers (brGDGTs) are bacterial cell membrane lipids that, when preserved in sedimentary archives, can be used to infer continental paleotemperatures. I present an in situ regional calibration of soil brGDGTs along altitudinal transects on both flanks of the Eastern Cordillera of Colombia that spans ~3,200 m in elevation for soil and air temperatures. These calibrations yield RMSEs of 1.5°C and 1.9°C, respectively, and allow for more precise and reliable reconstructions of past temperatures in the tropics than global calibrations. Along with refining brGDGTs in the northern tropical Andes, I also evaluate the efficacy of stable isotopes of precipitation and plant waxes as proxies for paleoaltimetry studies in the region. I use monthly hydrogen (δ²H_p) and oxygen (δ¹⁸O_p) isotope values of precipitation for an annual cycle, as well as hydrogen isotope values of plant waxes (δ²H_wax) in top soils along the same elevation transects as for the brGDGTs calibrations. The δ²H_wax values along the eastern flank of the Eastern Cordillera follow a simple Rayleigh distillation, with the average δ²H_wax values of n-C29, n-C31, and n-C33 alkanes showing an R² = 0.65 when regressed against elevation. In contrast, because of the lack of correlation with elevation in modern precipitation on the western flank, neither δ²H_p nor δ¹⁸O_p, and therefore δ²H_wax, offer reliable estimates of past elevations.

Regressions of surface temperatures in the Eastern Cordillera of Colombia with sea-surface temperatures (SSTs) in the equatorial Pacific show that the Eastern Cordillera warms or cools by half of the amplitude of the variation of SSTs in the eastern Tropical Pacific. Because Pliocene SSTs in the eastern Tropical Pacific resemble those during major El Niño events, when SSTs warm by ~4°C, the Pliocene Eastern Cordillera warms by ~2ºC at both high and low elevations. To evaluate how temperature changed in the Sabana de Bogotá during the Pliocene-Pleistocene, I estimated brGDGTs-based temperatures in the Funza-II core. New geochronology based on zircon U-Pb dates from ash layers place the base of the core at around ~4 Ma. I show that Pliocene temperatures were ~2.2 ± 2.0°C warmer than mid-late Pleistocene temperatures. This ~2°C warm in temperature could be explained by a permanent El Niño-like teleconnection to the Eastern Cordillera of Colombia, rather than a pantropical change in temperature. These temperature estimates are the only terrestrial tropical record within 5° of the equator for Pliocene time.

Full text at CU

Abstract

Uncertainty in climate predictions arises from three distinct sources: the internal variability of the climate system, which refers to natural fluctuations in climate that occur even in the absence of external forcing, model or structural uncertainty, as different models make different assumptions and hence simulate somewhat different changes in climate in response to the same forcing, and scenario uncertainty, which represents humankind’s free will concerning future climate change. In this thesis, we evaluate projections of Arctic sea ice in the context of these different sources of uncertainty. In particular, we show that internal variability, not scenario uncertainty, will ultimately determine the year of first summer ice-free conditions in the Arctic, in addition to the contribution from model uncertainty. Moreover, the increased inter-annual variability in late historical biomass burning forcing is found to cause a strong acceleration in sea ice decline in the early 21st century in several CMIP6 models, with model uncertainty affecting how different CMIP6 models respond to this forcing. Finally, we focus on the implications of scenario uncertainty on the changing sea ice cover through the lens of trans-border exchange of sea ice between the exclusive economic zones of the Arctic states. These different perspectives on climate model uncertainty allow for an improved understanding of the processes, variability and implications of a diminishing Arctic sea ice cover.

​Full text at CU

Abstract

Plucking is a common erosional mechanism in steep bedrock rivers with well-jointed, layered bedrock. Where active, plucking is one of the most efficient erosional mechanisms. However, the influence of bedrock layering and jointing on the evolution of rivers carved into layered landscapes has yet to be properly addressed at the process level. Rivers carving into fractured and jointed rock commonly display sharp steps in the bed separated by flat reaches between the steps. At the edges of these steps, blocks are vulnerable to plucking by both sliding and toppling. In this work, I seek to constrain the roles of block geometry and flow physics on the susceptibility of such blocks to entrainment. I use numerical modeling to demonstrate how blocks play an important role in setting the pace of river evolution, and therefore should be accounted for in landscapes where plucking is active. I first employ a computational fluid dynamics model to constrain the pressure and shear forces on blocks under different flow conditions. I use the results of this model to inform calculations of the susceptibility of blocks to entrainment and find that accounting for the pressure differences around blocks in force calculations significantly reduces entrainment thresholds. I then use these numerical results to inform a process-based 1-D model of bedrock river evolution that accounts for the entrainment probability of individual blocks in a jointed bed. I find that in the absence of external forcing, jointed beds will self-organize into a series of steps that is set by the baselevel lowering rate and block heights. Adding layers consisting of larger blocks stalls erosion at the contact between the large blocks and smaller blocks. Further, the large blocks prevent any signal of changes in baselevel lowering from being transmitted upstream until the large blocks are able to be plucked. Finally, I test this model with a spillway erosion case study from Canyon Lake, Texas, where an 8 m deep canyon was carved during a three-day flood event. The work presented here demonstrates the importance of properly accounting for block physics in river evolution models.

Advisor

Robert S. Anderson

Full text at CU

Abstract

Quantifying snowmelt behavior is valuable for understanding snow’s implications for water availability, flood risk, and ecosystem health. Building off of ongoing analyses of observed lateral movement of liquid water through snow, this research used repeat ground penetrating radar (GPR) derived liquid water content (LWC) maps and terrestrial Light Detection and Ranging (LiDAR) derived snow depth maps to explore the effects of wind and topography on the distribution of liquid water within an alpine snowpack on Niwot Ridge, Colorado. Observations of wind speed and direction were combined with a simplified linear turbulence model to produce distributed estimates of wind speed for this alpine basin. In this paper, we assess how wind-driven snow accumulation and redistribution impact the subsequent patterns of liquid water in snow during the melt season. Using linear regression models relating topography and liquid water estimates, we explore the physical controls on snowmelt water availability. In areas of deep snow, wind flow patterns play a critical role in determining the storage and transmission of liquid water in the snowpack. The GPR and LiDAR data indicate that meltwater tends to collect at the interfaces of the snow layers, largely governed by wind dynamics, and this water moves rapidly along these interfaces to streams and rivers. We discuss how improved model representation of in-snow liquid water may be achieved through the aforementioned detailed measurements and analyses. 

Abstract

Mountain snowpack is one of the primary surface water sources for about one-sixth of the global population. Accurately monitoring the spatial and temporal distribution of mountain snowpack – often measured as snow water equivalent (SWE) – is crucial for effective water management. While existing SWE estimation approaches remain highly uncertain, particularly when applied over large mountainous regions, the remotely-sensed snow data provide new opportunities to better characterize the spatial distributions of mountain snowpacks. This dissertation investigates the approaches that optimally blend satellite, airborne, and ground snow observations to improve (near) real-time SWE estimation over mountainous terrain.

The second chapter of this dissertation evaluates the accuracy of existing SWE estimation models in Sierra Nevada California. Five large-scale SWE datasets at fine spatial resolutions (<= 1000 m) are comprehensively validated and compared with the Airborne Snow Observatory (ASO) SWE data and ground snow pillow and snow course SWE observations. These SWE datasets include REC-INT, REC-ParBal, a Sierra Nevada SWE reanalysis (REC-DA), and two operational SWE datasets from the Snow Data Assimilation System (SNODAS) and the National Water Model (NWM-SWE), respectively. The results show that the REC-DA overall provides the most accurate SWE estimates across the Sierra Nevada (R2 = 0.87, MAE = 66 mm, PBIAS = 8.3%), followed by the REC-ParBal (R2 = 0.73, MAE = 83 mm, PBIAS = -6.4%), which is the least biased SWE estimates. Generally, SNODAS (R2 = 0.66, MAE = 106 mm, PBIAS = 9.3%) and REC-INT (R2 = 0.61, MAE = 131 mm, PBIAS = -28.3%) exhibit comparable but lower accuracy than the earlier mentioned two datasets, while NWM-SWE (R2 = 0.49, MAE = 142 mm, PBIAS = -25.2%) shows the least accuracy among the five SWE datasets.

Given that REC-DA is not applicable in real-time, in the third chapter, a SWE data-fusion framework is developed, which integrates the historical SWE patterns derived from REC-DA into a statistically-based linear regression model (LRM) to estimate SWE in real-time. To investigate the influence of satellite-observed daily mean fractional snow-covered area (DMFSCA) on SWE estimation accuracy, two LRMs are compared: a baseline regression model (LRM-baseline) in which physiographic data and historical SWE patterns are used as independent variables, and an FSCA-informed regression model (LRM-FSCA) in which the DMFSCA from MODIS satellite imagery is included as an additional independent variable. By incorporating DMFSCA, LRM-FSCA outperforms LRM-baseline with improved R2 from 0.54 to 0.60, and reduced PBIAS from 2.6% to 2.2% in snow pillow cross-validation. The improvement in LRM-FSCA’s performance is more significant during snow accumulation periods than during the snowmelt seasons. Compared to the ASO SWE, the LRM-FSCA explains 85% of the variance on average, which is at least 21% higher than the operational SNODAS (R2 = 0.64) and NWM-SWE (R2 = 0.33) in comparison.

In chapter 4, a SWE bias correction framework (SWE-BCF) is developed that incorporates the ASO SWE and machine learning (ML) algorithms to further improve LRM SWE estimates in real-time. The performance of a wide range of commonly used machine learning algorithms is examined in the SWE-BCF including Gaussian Process Regression (GPR), Support Vector Machine (SVM), Bayesian Regularized Neural Networks (BRNN), Random Forest (RF), and Gradient Boosting Machine (GBM). The results indicate that all ML algorithms are capable of improving LRM-SWE accuracy substantially. While no single model performs significantly better than others, GPR, overall, shows the best performance with a 20% (0.14) increase in mean R2 value, a 31% (51 mm) reduction in mean RMSE, and a 61% (18.0%) reduction in absolute PBIAS compared with the original LRM using ASO SWE data for model validation. RF shows the most robust and stable performance in SWE bias correction with a 10% (0.08) increase in median R2 and a 41% (50 mm) reduction in median RMSE compared with the original LRM.

​Full text at CU

Abstract

Stable isotopes of hydrogen and oxygen in polar ice cores provide information about local temperature and atmospheric circulation. We use a multi-taper method (MTM) of spectral analysis on a continuous high-resolution (i.e. mm-scale) Greenland water isotope record, recently recovered from the East Greenland Ice Core Project (EGRIP), to determine how interannual and decadal temperature variability changed throughout the past 50 thousand years. We are specifically interested in trends across the most recent glacial-interglacial transition and across millennial scale Dansgaard-Oeschger (i.e. stadial-interstadial) cycles to elucidate how large temperature changes affect variability around the mean in Greenland. To further understand global relationships in variability, we later make comparisons with mm-scale ice core records from the South Pole (SPC) and the West Antarctic Ice Sheet Divide (WDC). Our results reveal a strong coupling between mean temperature and high-frequency (i.e. 7-15 year) climate variability at EGRIP. On average, the Last Glacial Period (LGP; 11.7-50 ka bp) exhibits 2.5 times greater variability than the Holocene and within the context of the LGP, cold stadial periods are 1.5 times more variable than warm interstadial periods. We provide a plausible mechanism for the trend we observe across Dansgaard-Oeschger (DO) cycles in northeast Greenland: a larger sea ice area coupled with a more variable sea ice front may explain the increased isotopic variability during cold stadial periods. In contrast, neither Antarctic site (SPC or WDC) exhibit changes in high-frequency variability across millennial scale warm phases, known as Antarctic Isotope Maxima (AIM) events, that occur with each DO Event. While elucidating exact forcing mechanisms for observed trends in high-frequency variability is outside the scope of this study, we provide critical benchmarks and reasonable hypotheses to test in future climate modeling research.

Abstract

Polar ice cores contain multiple proxies which record vast amounts of climate information over hundreds of thousands of years. Water isotopes in ice cores can be used to infer past temperature and atmospheric circulation patterns, and with recent advances in technology can be measured at very high resolution. Here, I present analyses of continuous water isotope records from two new Greenland ice cores, and use snow surface experiments to work towards improving our interpretation of ice core records.

The Renland Ice Cap is located in East-Central Greenland, and an ice core drilled in 2015 contains a seasonally-resolved water isotope signal through 2.6 ka and sub-decadal signals through 8 ka. Using spectral analysis, I find that decadal (i.e. 15-20 year) isotope variability at Renland co-varies with North Atlantic sediment core indicators of ocean circulation patterns throughout the Holocene. Furthermore, analysis of the seasonal signal reveals that a decreasing trend in the winter isotope signal may correspond to an increase in Arctic sea ice cover and a decrease in total annual insolation over the last 2.6 ka. Together, these findings show that coastal Greenland climate may be closely tied to regional sea surface conditions.

The East Greenland Ice Core Project (EastGRIP) is an ongoing drilling campaign in Northeast Greenland, with a record currently extending to 50 ka. EastGRIP has sub-decadal isotope signals preserved through the Glacial period, and for the first time we have a continuous Glacial record for multiple isotope parameters (i.e. δ18O and deuterium excess). Through comparison to a North-Central Greenland ice core record, I demonstrate the importance of high-resolution sampling and expand on our understanding of spatial variability in Greenland climate. Additionally, an analysis of abrupt climate events throughout the Glacial period indicates that rapid warming is preceded by a shift in North Atlantic atmospheric circulation patterns; this has important implications for our understanding of mechanisms occurring during abrupt climate changes.

Many aspects of the isotope-climate relationships are well constrained through decades of research; however, there remain gaps in our understanding of processes taking place at the surface of the ice sheet between depositional precipitation events, and how they influence the recorded climate signal. To address this I use a series of laboratory experiments to show that in a controlled environment the snow isotopic composition changes rapidly due to sublimation, with this finding supported by model results. Complementary field experiments demonstrate that in a natural setting, the top 4 cm of the snow surface evolves on an hourly timescale due to sublimation and exchange with atmospheric vapor. These results suggest that water isotopes may effectively integrate across multiple parameters to record a more continuous climate signal, which may improve isotope-enabled climate models and inform our interpretation of ice core water isotope records.

Full text at CU

Abstract

Lagerman Reservoir is a saline, closed evaporite basin subject to changing hydrologic regime, algal blooms, and fish kills. This ecological survey was conducted to obtain scientific data for future management decisions and recreational safety. Weekly shore samples and monthly depth profile samples were taken, along with data synthesis of EPA's CyAN satellite monitoring system. Dissolved oxygen was low, indicating unfavorable conditions for fish and macroinvertebrates. Transparency and nutrient concentrations were also low, with nitrogen limiting the system. The algal biomass was dominated by Synechococcus, as well as Dolichospermum, Lyngbya, Oscillatoria, and Merismopedia. Chlorophyll-a analysis indicated similar trends to the CyAN satellite data, with a peak of algal growth around late June, followed by a die off of algal biomass. This reservoir ranged from eutrophic to hypereutrophic. Lagerman Reservoir is considered to be unsuitable for fish and vulnerable to toxic algal blooms that could be hazardous to human health.

​Full text at Opin vísindi

Abstract

As high latitude regions continue a decades-long trend of warming at roughly twice the rate of the global average, an understanding of their climatic histories becomes increasing important for predicting their future. Organic molecular proxies preserved in lake sediment archives offer one avenue for reconstructing key elements of such past climates, including their temperature, precipitation, and vegetation regimes. In particular, a class of bacterial membrane-spanning lipids called branched glycerol dialkyl glycerol tetraethers (brGDGTs) form the basis for a paleothermometer that can be applied to reconstruct temperatures as far back as the Cretaceous in sedimentary archives across the globe.

Despite these successes, challenges remain that complicate the development and application of brGDGT-based proxies. First, while they correlate best with temperature and pH, other environmental parameters can influence brGDGT distributions, including seasonality, conductivity, and oxygen availability. Second, it is unknown whether these empirical correlations are the result of a direct physiological response of brGDGT-producing organisms to their environment or an indirect effect resulting from variations in bacterial community composition. Finally, an incomplete understanding of where brGDGTs are produced on the landscape and how they contribute to the sedimentary record hinders our ability to interpret proxies in mixed-source archives.

Herein, I present research addressing each of these three challenges with an emphasis on the Eastern Canadian Arctic and Iceland. First, I develop a technique for grouping brGDGTs based on structural characteristics and show that it can be used to deconvolve the effects of temperature and pH/conductivity. I further find a warm-season bias in brGDGT-derived temperatures and develop calibration equations for temperature and conductivity. Next, I compile >2500 samples from a dozen sample types across the globe and find near-universal trends in the relationships between brGDGTs and temperature, pH, and one another. These commonalities support a physiological basis for observed environmental trends. Finally, by measuring brGDGTs in their intact, polar form, I find that lipid sources in lake catchments can be distinguished and suggest novel applications down core. By advancing our understanding of brGDGTs, my results further our ability to reconstruct key climatic variables from sedimentary archives, especially at high latitudes.

Full text at CU

Abstract

Wetlands serve as important locations of disproportionately high biogeochemical activity and plant productivity in many lowland regions. However, little is known about the function of alpine wetlands, or about how their biogeochemical cycling compares with the broader alpine landscape, which is usually composed of thin, rocky soils and tundra vegetation. In my thesis research, I compared the soil and water biogeochemistry among three types of alpine wetlands at Niwot Ridge, Colorado: alpine wet meadow, periglacial solifluction lobe, and subalpine wetland. Each wetland type exhibited unique biogeochemical characteristics and higher concentrations of carbon and sulfur than dry alpine meadow. These findings suggest that wetlands may have a disproportionate effect on biogeochemical and ecological processes at Niwot Ridge, and serve as important sites in predicting the effects of global change on alpine landscapes.

Abstract

The past several decades have witnessed a fascinating evolution in the sulfur (S) cycle, resulting in an unprecedented increase in agricultural S use. However, little research has explored how intensive agricultural S applications alter S biogeochemistry across a range of agricultural settings and scales— essential for constraining the cascade of environmental effects of using S in agriculture. In this dissertation, I identify and trace the biogeochemical “fingerprint” of agricultural S from field-to-watershed scales.

Asking first how patterns of S chemistry change across a watershed with intensive agricultural S inputs, I contrasted S stable isotope (δ34S) and sulfate (SO42-) concentration measurements collected within agricultural areas and surrounding forests and grasslands with background (atmospheric and geologic) S sources. Stable S isotope results showed that agricultural S has a robust and distinctive biogeochemical fingerprint that is traceable beyond fields. I then delved deeply into processes affecting organic S composition—the largest pool of S within soils. I developed a novel method to directly measure δ34S of dissolved organic matter (DOM) and combined this approach with techniques to measure organic S speciation and molecular composition. Agricultural S applications increased DOM S-content by two-fold compared to forests and grasslands, and I found that a suite of molecules unique to agricultural areas have the potential to be used as agricultural S tracers. Finally, I investigated how wildfire disturbance affects agricultural S and its interactions with other elemental cycles. Though not appearing to strongly affect the agricultural S fingerprint, wildfire did enhance organic carbon leaching, producing a potentially potent cocktail for stimulating toxin production in downstream aquatic ecosystems. These results reinforce the importance of considering integrative studies at watershed scales to evaluate the transport and fates of agricultural S and point to the increasing role of climate change as an additional control on agricultural S biogeochemistry.

Combined, this research (1) reveals agricultural changes to the modern S cycle at multiple scales, (2) establishes tools and techniques to trace agricultural S through watersheds, and (3) provides a critical first step towards fully constraining the environmental fates and unintended consequences of S inputs to agricultural systems.

Full text at CU

Abstract

As water flows downstream, it is transported to and from environments that surround the visible stream. Along with surface water, these laterally and vertically connected environments comprise the stream corridor. Stream corridor connectivity influences many ecosystem services, including retention of excess nutrients. The subsurface area where stream water and groundwater mixes—the hyporheic zone—represents one of the most biogeochemically active parts of stream corridors.

The goal of my research is to advance understanding of how connectivity between different parts of a stream corridor controls the availability and retention of nitrogen (N), a primary nutrient that can negatively impact water quality. First, I developed and applied a new machine learning method to objectively characterize the extent and variability of hyporheic exchange using geophysical data. In applying this method to a benchmark dataset, I found that hyporheic extent does not scale uniformly with streamflow and that changes in the heterogeneity of connectivity differ over small (<10 m) distances. Next, I leveraged the relative simplicity of ephemeral streams of the McMurdo Dry Valleys (MDVs), Antarctica, to isolate stream corridor processes that influence the fate of N. Through intensive field sampling campaigns, I found that the hyporheic zone can be a persistent source of N even in this low nutrient environment. Next, I combined historic sample data and remote sensing analysis to estimate how much N is stored in an MDV stream corridor. My results indicate that up to 104 times more N is stored in this system than is exported each year, with most of this storage in the shallow (< 10 cm) hyporheic zone. Lastly, I examined 25 years of data for 10 streams to assess how stream corridor processes control concentration-discharge relationships. I found that in the absence of hillslope connectivity, stream corridor processes alone can maintain chemostasis – relatively small concentrations changes with large fluctuations in streamflow – of both geogenic solutes and nutrients. My analysis also revealed that solutes subject to control by biological processes exhibit more variability within chemostatic relationships than weathering solutes.

Altogether, this research advances characterization of processes that are difficult to measure or are often overlooked in typical studies of temperate stream corridors. These studies provide insight into the surprising ways in which N is mobilized, transformed, and retained due to stream corridor connectivity in intermittent stream systems with few N inputs.

​Full text via ProQuest

Abstract

The Chukchi Sea is a key region for shipping and other growing economic activities in the Arctic. Seasonal sea ice conditions in the Chukchi Sea are strongly determined by the oceanic heat transport into the Chukchi Sea from the north Pacific Ocean via the Bering Strait (see Chapter 1). In Chapter 2, we statistically model Bering Strait heat transports and then use these models to forecast sea ice retreat and advance dates in the Chukchi Sea. In Chapter 3, we further investigate the interannual variability of spring Bering Strait water temperatures. We find that June Bering Strait water temperatures are set upstream the preceding autumn and winter by ocean temperatures in the southwestern Bering Sea shelf, and then advected by the Anadyr current towards the Bering Strait. In Chapter 4, this research is summarized and avenues for future work are discussed.

Abstract

The impacts of anthropogenic climate change, including sea-level rise, warming temperatures, and ocean deoxygenation, are expected to significantly alter marine and terrestrial ecosystems, though the exact nature of future changes are not well constrained. Intervals in Earth history that have experienced similar conditions, such as the Cretaceous Oceanic Anoxic Event 2 (OAE2; ~94 Ma), can serve as case studies for investigating the responses of global biogeochemistry to carbon cycle perturbations, sea-level rise, and expanded and intensified ocean anoxia.

A rock core from southern Utah (SH#1 core) presents an expanded sedimentary record of OAE2 from the western margin of the Western Interior Seaway that contains abundant organic matter derived from marine and terrestrial environments. In this dissertation, we developed and established several analytical organic chemical methods to characterize lipid biomarkers in the SH#1 core as a means of investigating biogeochemical responses to global climate change during OAE2. Results indicate that sea-level rise and CO2 input drove increased marine productivity, alterations in marine ecology, and expanded and intensified ocean deoxygenation. Ocean anoxia was persistent in bottom waters, and periods of severe anoxia, including euxinia in the upper water column, occurred throughout the event. Bacterial and algal populations were altered in response to sea-level changes, ocean deoxygenation, and increased productivity, with the frequent appearance of green sulfur bacteria, methanotrophic bacteria, and several periodic changes in dominant algal groups. Records of biomarkers produced by forest fires show an increase in the frequency of forest fires during OAE2, likely due to increased atmospheric oxygen from widespread marine organic carbon burial at the onset of the event. Biomarkers and carbon isotope mass balance equations indicate that forest fires may have been part of a global-scale positive feedback between terrestrial nutrient input to the oceans, marine productivity, ocean anoxia, and marine organic carbon burial. 

Results from carbon isotope analyses of individual marine and terrestrial biomarkers provide constraints on the carbon cycle during OAE2, including a high resolution atmospheric pCO2 record, and estimates of the isotopic composition of oceanic dissolved inorganic carbon and atmospheric CO2. Nitrogen isotopic analyses of bulk organic matter in the SH#1 core and another core from the southern Western Interior Seaway indicate that nitrogen cycling was sensitive to changes in water column deoxygenation and sea level, and varied across oceanographic settings. Results from ongoing porphyrin-specific nitrogen isotope analyses will inform isotope mixing models to decipher dominant sedimentary organic nitrogen sources and nitrogen transformation in anoxic oceans. This dissertation elucidates the nature of ecosystem responses to global environmental change, which can provide a historical context for understanding anthropogenic climate change, and inform projections for future changes in ocean circulation, chemistry, biology, and carbon burial.

Full text at CU

Abstract

Primary producing algae form the basis of carbon fixation, oxygen production, and food webs in aquatic ecosystems globally. However, human activities disrupt climate and freshwater physicochemistry. These impacts alter the health of algal communities and the ecosystem services algae provide. Meanwhile, spatial processes like dispersal and landscape characteristics like geology also influence algal structure and function. Diatoms are indicators of stream health and are model organisms for understanding the processes underlying microbial biogeography. Benthic cyanobacteria present risks to human health through the proliferation of toxin-producing blooms. With this dissertation, I investigate the ecosystem processes that influence diatom and cyanobacterial community composition and taxon distributions. My goal is to advance the understanding of ecosystem controls on algal biogeography and to characterize taxon-specific autecology for use in environmental management. First, I measured the extent of wind-mediated dispersal of benthic diatoms across aquatic habitats to better understand how community composition is structured by spatial processes across the McMurdo Dry Valleys polar desert in Antarctica. I found that inter-habitat dispersal is common but less influential on community composition than intra-habitat factors such as environmental conditions. I then used non-linear, multivariable modeling to assess the relative influences of climate, watershed characteristics, and in-stream stressors on the relative abundances of 268 diatom taxa across gradients of human impact in the northeast United States. My results indicate diatom taxa are affected by different suites of environmental conditions but that taxa belong to ecological guilds based on shared responsiveness to environmental factors. Finally, I applied multivariable modeling towards understanding the effects of aquatic stressors, including herbicides and persistent organic pollutants, on the distributions of benthic cyanobacteria across northeast U.S. streams. I found that watershed characteristics, streamflow, and herbicides were more influential than light availability, water temperature, and nutrients on the distributions of potentially toxigenic cyanobacterial genera. Collectively, this research expands the knowledge of how benthic algal communities and taxon distributions are structured at large spatial scales along gradients of unimpacted and human-altered environmental conditions. I provide a novel modeling framework and taxon-specific autecological information that can be applied to environmental assessments of stream health and future algal research.

​Full text via ProQuest

Abstract

A deep-water aerial sampling technique has been designed and implemented for sampling pit lakes, which are dangerous water bodies for boats to approach. The approach involves an unmanned aerial vehicle (UAV) hexacopter, DJI Matrice 600, that can attach a 1.2L Niskin sample bottle or conductivity, temperature, and depth profiler. The bottom attachment on the UAV consists of two servo motors that allow the detachment of a 1 kg messenger that triggers a 1.2L Niskin bottle shut and provides an emergency payload release. We first tested this technology on September 20, 2016 and obtained a conductivity, temperature and depth (CTD) profile of Dillon Reservoir and then collected a water sample from 25 m depth. The CTD profile showed a possible layer of water originating from an inflow stream, the Snake River, which has high particulate and dissolved metals from acid rock drainage. In addition, this UAV deep-water sampling approach has been tested at seven pit lakes from 2016 – 2017, the deepest sample retrieved was 80 meters, and is seen to be an effective method of water sampling at depth. It is important to gain an understanding of water-quality conditions to determine the extent to which pit lakes are effecting the environment. The water-quality characteristics of pit lakes also provide a better understanding of remediation requirements. This methodology of sampling with a heavy lifting drone is seen to be an effective approach to sampling dangerous water bodies and a municipality’s water supply. The benefits and field considerations of a UAV water sampler include improved safety, reduced sampling costs, and improved efficiency.

Full text at CU

Abstract

Nearly 80% of the United States’ freshwater originates in forested landscapes at risk of wildfires, which influence both the terrestrial landscape and hydrologic regime by introducing a heterogeneous spectrum of thermally altered carbon compounds, known as pyrogenic carbon (PyC). Given the projected increase in both wildfire frequency and intensity, understanding the coupling of hydrologic transport and chemical fractionation that wildfires impose on water sources is critical. New research has begun to show that PyC can be quite mobile and reactive with turnover time of decades or years in soils rather than previously assumed millennia timescales, emphasizing the importance of dissolved PyC (DPyC) translocation from soils to rivers. While riverine PyC transport has been identified as a key component of the global PyC cycle, the extent to which photodegradation contributes to both short-term and long-term DPyC chemical fraction has yet to be resolved. This research investigates the role of photodegradation as a major driver altering aquatic DPyC physical and chemical properties using fluorescence spectroscopy. Artificial PyC was created by burning organic matter at various temperatures to isolate distinct portions of the PyC spectrum. The organic matter, comprised of leaves and soils, was collected from Great Smoky Mountain National Park. Each temperature range of the PyC spectrum was separately leached, filtered, and the dissolved fraction was placed outside and exposed to natural sunlight for various exposure times ranging from zero to 28 days. This photodegradation experiment took place in Boulder, Colorado during the summer months to maximize daily sun exposure. Photochemistry was confirmed by monitoring the photochemical formation of hydrogen peroxide via constant wavelength fluorescence spectroscopy. The dissolved organic matter was characterized using ultraviolet-visible absorption and excitation- emission matrix fluorescence spectroscopy. By isolating distinct portions of the PyC spectrum, we will better be able to anticipate the fate of PyC in watersheds effected by wildfires.

Full text at CU

Abstract

Dissolved oxygen (DO) concentrations in rivers are critical for aquatic habitat and controlled by biological generation and uptake, and physical factors. One important physical factor is hydrology: not only streamflow dynamics (changing amounts of water), but also changes in surface-groundwater exchanges. Over a period of 15 months in East River, Colorado from August 2017 (a somewhat "average" flow year) to October 2018 (a low flow year), high frequency (5 minute) DO and temperature data were collected in the water column of the river and directly in the streambed at depths of 10 cm, 20 cm, and 35 cm. Using the VFLUX2 model, temperature data were used to estimate vertical upwelling and downwelling vertical fluxes of water. We find that there was downwelling throughout both years, and increased fluxes into the bed during peak flows. From relating vertical flux to steam discharge and groundwater tables we find that stream discharge is a control of streambed DO during low flow. We calculated DO removal from the channel to the bed, finding enhanced removal rates in 2018. We observed an extended hyporheic anoxic period throughout the summer and fall of 2018 due to increased DO removal rates. The three subsurface locations were found to not all be on the same flow path, which may account for some of the DO differences in 2017 while increased removal due to low flow conditions are the primary factor in 2018. This research has advanced our understanding of the dependence of DO in both the streambed and open channel on stream-groundwater exchanges by showing periods of stream discharge control on DO dynamics.

Full text at CU

Abstract

In the McMurdo Dry Valleys (MDVs), foehn winds are a principal vector of landscape connectivity that facilitate movement of materials between glaciers, streams, soils, lakes and other parts of the ecosystem. While previous publications show that turbulent, warm and dry foehn winds indirectly relate to an increase in lake level rise via an increase in degree days above freezing (DDAF), the direct quantified impact of foehn winds to streamflow and lake level rise remains unclear. The MDVs are the largest ice-free region of Antarctica, which experience minimal precipitation. Valley bottoms contain permanently ice-covered closed basin lakes filled with meltwater from outlet glaciers via stream channels. In Taylor Valley, several meteorological stations and lake monitoring stations record average measurements of weather conditions and lake conditions on 15 to 20-minute intervals. In this thesis, the meteorological definition of foehn winds is refined and hydrologic response to foehn winds is evaluated. During the austral summer streamflow season (November-February), foehn winds are predicted to increase meltwater generation and closed-basin lake level rise. Past publications have shown that foehn wind events contribute to lake ice sublimation year-round, whereas melt does not typically occur in nonsummer months. Analysis of non-summer lake ice ablation utilizing recent lake stage and ablation data is also explored herein. Although a significant correlation was not found, summer foehn winds appear to promote above average daily lake level rise given sufficient air temperatures. Daily average lake level rise is greater for longer periods (i.e., 4-day average daily rise > 3-day average daily rise, etc.) indicating that there is at least a 4-day post-foehn impact on lake level rise during the summer. Lake ice ablation in non-summer months is shown to have a significant relationship with increasing foehn wind occurrence and wind-run. Because foehn winds are expected to increase with global warming, these hydrologic relationships aid in predicting the future of the McMurdo Dry Valley ecosystem in a warming world.

Full text at CU

Abstract

This thesis aims to assess the validity of bacterial branched glycerol dialkyl glycerol tetraethers (brGDGTs) temperature reconstructions in Arctic lake settings from a Holocene (~11,700 BP) lacustrine sediment core from Baffin Island, Eastern Canadian Arctic. The distribution of brGDGTs in peats, soils, and lake sediments has been shown to correlate with mean annual air temperature (MAAT) and this proxy has been widely applied to sedimentary archives for paleotemperature reconstructions. However, the production and distribution of brGDGTs are impacted by confounding environmental variables that are currently not well understood. Here I study the distribution of brGDGTs preserved in a high-Arctic lake setting and apply the most up-to-date brGDGT-inferred temperature reconstruction calibrations. This thesis specifically investigates the role of changing oxygen levels on reconstructed brGDGT paleotemperatures. Comparisons with other soil and lacustrine samples from Baffin Island suggest that brGDGTs in Upper Gnarly are primarily sourced from within the lake over the Holocene, and estimated temperatures from surface sediments using recently published lake-specific calibrations compare favorably with measured summer air and water temperatures from the region. The downcore reconstruction from Upper Gnarly exhibits a trend opposite of what is expected with a cool Early Holocene followed by warming towards the present. Importantly, two intervals of cooler reconstructed temperatures are observed during intervals of supposed suboxia in the lake. Overall, my results present further evidence that suboxic conditions generate a cold-bias in brGDGT paleotemperature reconstructions, and ultimately need to be considered in future research in paleotemperature reconstruction in high-Arctic lake settings.

Full text at CU

Abstract

Seasonal sea ice transitions dates are under-utilized in evaluating climate model projections of Arctic sea ice loss despite long, pan-Arctic satellite-based observational records of seasonal transition dates. In this thesis, we show how the limitations that have prevented their widespread use can be overcome through the use of large ensembles and a novel sea ice satellite simulator, and how seasonal sea ice transitions dates can benefit model evaluations.

In particular, we quantify the uncertainty related to definition differences and internal variability, allowing us to use seasonal sea ice transitions as process-based metrics to understand model biases in sea ice simulations. In addition, we demonstrate that a sea ice satellite simulator can take model evaluation a step further, providing novel and direct comparisons between satellite observations and model simulations as well as insights into the physical processes captured by sea ice remote sensing algorithms. These direct comparisons enable more accurate climate model assessment and improve the evaluation of seasonal Arctic sea ice transitions in models.

​Full text via ProQuest

Abstract

El Nino Southern Oscillation (ENSO) is Earth's largest source of interannual climate variability; however, its future remains difficult to predict. Evaluation of past ENSO may help improve our basic understanding of the phenomenon and help resolve discrepancies among models tasked with simulating future climate. Individual foraminifera analysis (IFA), a tool that allows continuous down-core records of ENSO-related temperature variability through the construction and comparison of paleotemperature distributions, provides an opportunity to extend records of past ENSO beyond the most recent geologic past. However, accurately interpreting ENSO from IFA requires (1) an understanding of how partial dissolution affects temperature distributions derived from individual planktic foraminifera, (2) confirmation that a population of planktic foraminifera is capable of capturing site-specific temperature variability, and when employing down-core, (3) consideration of the complexity of ENSO diversity.

In Chapter 1 of this dissertation, I evaluate how partial dissolution affects the Mg/Ca-temperature proxy. I show that temperatures derived from individual measurements of Mg/Ca in two species of planktic foraminifera, Globigerinoides ruber and Neogloboquadrina dutertrei, support a percent Mg loss model which does not affect the shape or variability of a temperature distribution and is ideal for reconstructing past ENSO. In Chapter 2 of this dissertation, I investigate whether Mg/Ca-based IFA is capable of capturing site-specific temperature variability using nine core-tops across the equatorial Pacific. Using quantile-quantile analysis, I show that individual measurements of Mg/Ca in G. ruber and N. dutertrei reflect site-specific temperature distribution shape and variance when accounting for regional differences in depth habitat of both species of planktic foraminifera. In Chapter 3, this dissertation transitions to investigate ENSO during the Last Glacial Maximum (LGM). I use individual measurements of Mg/Ca in two species of planktic foraminifera, G. ruber and N. dutertrei, from two locations in the eastern equatorial Pacific to show that the center of ENSO activity was farther east during the LGM than during the late Holocene suggesting that ENSO diversity can affect interpretations of IFA records and emphasizing the necessity for multi-site reconstructions.

Abstract

Soils contain the largest pool of terrestrial organic carbon and the organic content of soil is strongly influenced by climate, vegetation, land use, time, and physicochemical properties of the soil. In order to better understand factors that control soil carbon, I seek to understand the mechanisms of stabilization of soil organic matter (SOM) across a diverse range of soil types. To integrate SOM stabilization and vulnerability across different spatial scales, it is important to understand the mechanisms of stabilization and how broad (climate) and fine scale (soil physicochemical properties) controls impact SOM dynamics under climate change conditions. The goals of my research are to (1) develop fluorescence spectroscopy models designed for SOM; (2) understand how the molecular composition of SOM varies as a function of fine and broad scale controls; (3) build a predictive model for SOM composition; and (4) relate metrics of composition between fluorescence spectroscopy and mass spectrometry. Using fluorescence spectroscopy and PARAllel Factor Analysis (PARAFAC), I created a universal 6 component soil model to compare and interpret fluorescence signals. Using sequential extraction methods, I can determine the composition of the SOM that is hydraulically active (non-stabilized) and which fraction loosely held in soils (labile). To understand how the molecular composition of SOM varies across a diverse range of soil types, I will be coupling fluorescence spectroscopy and Fourier-transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) of soil extracts. Combining environmental factors, physicochemical properties of the soil, vegetation indices, and FT-ICR-MS data I determined that SOM composition in the hydraulically active and labile fraction (water extract) is related to by climate and physio-chemical properties of soil. In the methanol extractable fraction, the SOM composition is related to vegetation indices, and in the chloroform extract was related to soil texture, precipitation, and vegetation. I use this work to gain insight into the interplay between mineralogy, SOM composition, stabilization and vulnerability to changes in broad scale controls.

​Full text at CU

Abstract

The intensity, duration, and spatial distribution of frozen soil influences hydrologic flow paths, soil biogeochemistry, and slope geomorphology. In mountain environments, steep topography produces strong gradients in solar insolation, vegetation, and snowpack dynamics that lead to large differences in soil temperature over short distances, suggesting a need for high-resolution, process-based models that quantify the influence of topography. Surface energy balance calculations and a physical snowpack model based on the Utah Energy Balance have been coupled with PFLOTRAN-ICE, a subsurface thermo-hydrologic model that simulates water and energy transport in the subsurface, including freeze-thaw processes. A thermo-hydrologic modeling study is presented against the backdrop of field observations from Gordon Gulch and Niwot Ridge, seasonally snow-covered catchments in the headwaters of the Boulder Creek watershed. Despite a persistent snowpack on the north-facing slope at Gordon Gulch, seasonally frozen ground is more prevalent and persistent there because of low solar insolation and a thin snowpack. The south-facing slope experiences significantly higher incoming solar radiation that prevents the persistence of frozen ground. Representation of the snowpack and surface energy balance significantly improves soil temperature estimates compared to model forcing based on air temperature alone. At Niwot Ridge, deep (>1m depth) frozen soil underlying bare ground impeded groundwater recharge, and shallow frozen ground (<1m depth) beneath seasonal snow limited infiltration. Modeled alpine and subalpine snowcover exerted a positive effect on soil temperatures but did not prevent or eliminate frozen ground completely. Shallow freezing beneath snow-covered ground exerted a much stronger effect on infiltration than shallow freezing beneath bare ground because the soil beneath the snow remained frozen while the snowpack was melting, whereas solar insolation thawed bare patches by the time they received excess snowmelt “run-on”. In projections of seasonally frozen ground, simulations forecast two additional months of unfrozen soils by the end of the 21st century compared to the 1952-1970 time period. A permafrost analysis provides support for the occurrence of permafrost above 3800m and suggests that the deep soil thaw that has taken place over the last several decades is small compared to deep soil thaw that should be expected throughout the current century.

​Full text via ProQuest Open Access

Abstract

Glaciers are an integral part of polar and alpine landscapes, providing water, inorganic, and organic material subsidies to downstream ecosystems. These subsides regulate downstream temperature, streamflow, and sediment supplies. Warming in high altitude and high latitude environments due to climate change is resulting in rapid and substantial mass loss of glaciers. In order to better predict impacts and future change to glaciers and downstream environments, we endeavor to better understand glacier physical and biogeochemical processes. Glaciers in the McMurdo Dry Valleys (MDVs) of Antarctica are very sensitive to slight changes in the energy balance. Small temperature or solar radiation increases can result in outsize increases in glacier melt, which is the main source of water for the MDV ecosystem. Sediment on the glacier surface is thought to be a key factor driving both melt and biogeochemical cycling on glaciers. This dissertation examines the distribution of sediment on the MDVs glacier surfaces, how it may have driven recent glacier morphological change, and identifies sediment-driven biogeochemical processes on the MDV glaciers. To do so, we carried out field data collection, field- and lab-based nutrient uptake experiments, geospatial analysis, and coupled sediment and energy balance modeling. We find that the glacier surfaces have changed in response to recent warm events by increasing roughness and the density of meltwater channels on the glacier surface. The increase in roughness occurred in already rough areas that serve as collection points for wind- and water-transported sediment. The rough surfaces and sediment have low albedo and can absorb a higher amount of energy, spurring additional melt. The distribution of sediment on the surface and in the top meter of ice is a reflection of patterns of wind deposition and seasonal melt on the glacier. The total amount of sediment in the top meter of ice agrees with previously measured rates of sediment deposition and follows a valley-wide pattern. The depth of the peak sediment concentration in the top meter of ice is a function of the thermal history of the glacier – both summer energy balance and winter sublimation rates. We also find that the biota living in the sediment is capable of removing nutrients from glacier melt water, modulating the amount and form of nutrients delivered to downstream ecosystems. This research clarifies the role of glaciers within the larger MDV ecosystem. It also advances our understanding of surficial glacier melt and biogeochemistry, which can improve predictions of how the functional role of glaciers within their larger ecosystems will evolve due to climate change.

Abstract

Each winter, snow whitens up to a third of the land on Earth, plus the ice-covered tenth of the ocean. In wind-swept areas, this snow collects in shifting bedforms such as ripples, dunes, and waves. These bedforms cover 10-20% of the surface of the Earth, and increase the surface roughness and energy fluxes, but their formation is little understood. Here, I present the first observations of snow bedform movement and growth, drawn from a three-year field study in the Colorado Front Range (Ch. 2). These observations show snow dunes accelerating minute-by-minute in response to gusts; migrating sastrugi stripping a layer of snow; and eroding surfaces organizing into upwind-facing steps. These changes are functions of wind, melt, and time since snowfall (Ch. 3). Flat snow exists but briefly in the Front Range, in gentle winds. After identifying the processes that drive bedform evolution, I developed the first numerical model of snow bedform growth (Ch. 4). The model uses a cellular automaton to simulate saltation, snowfall, and sintering, producing realistic dunes and waves. I discuss model non-dimensionalization, calibration, acceleration, and emulation.Finally, I simulated the growth of snow dunes in a range of wind and snowfall conditions (Ch. 5). The simulated dunes grew tall and widely-separated when winds were strong and snow fell slowly; these trends agreed with field observations. I then quantified the impact of the simulated bedforms on radiative and conductive heat transfer. Snow dune and wave growth likely increase heat fluxes through snow in most of the polar regions - sea ice, tundra, the Greenland and Antarctic ice sheets - by up to 94%

​Full text via ProQuest

Abstract

The humid tropics are undergoing land use and climate change at a rate which outpaces many other parts of the world. Despite this, the impacts of these environmental changes to many hydrological and biogeochemical processes remain poorly understood. To address outstanding knowledge gaps limiting our ability to anticipate how future changes will impact water resources in these regions, this dissertation investigated hydrological and biogeochemical processes across a gradient of tropical land covers in central Panama. By analyzing relationships between stream chemistry and physical hydrological processes in three catchments of varying land cover (mature forest, young secondary tropical forest, and cattle pasture), several outstanding questions in tropical hydrology and biogeochemistry were addressed.

A combination of geochemical mixing models was used to identify dominant hydrologic flowpaths in each catchment, and the hydrologic conditions under which they became active. Pasture land cover produced infiltration excess overland flow during large wet season storm events, resulting in massive exports of event water to streams and comparatively higher new- runoff efficiencies. Contrarily, lateral subsurface flow through macropores in the upper 30cm of the soil column was the dominant hydrologic flowpath in the forested catchments, producing lower new-water runoff efficiencies and allowing for greater vertical connectivity in the soil column. The activation of these flowpaths in each catchment were found to be driven by the exceedance of rainfall magnitude and intensity thresholds, which varied according to land cover. Differences in flowpaths between the catchments produced differences in seasonal runoff dynamics between them, with the forested catchments producing only a fraction of the total new-water driven runoff produced by pasture.

To study how hydrological and biological differences between the catchments impacts biogeochemical processes like weathering and nutrient cycling, concentrations-runoff relationships and exports of eight major solutes were quantified and compared between the catchments. Despite a large disparity in total runoff production between the different land covers, their export of bedrock and atmospherically derived solutes was largely the same. The catchments produced differing loads of biologically active solutes and nutrients, a function of both differences in their biota and in the hydrologic flowpaths connecting these solutes to the stream.