The majority of potentially active volcanoes worldwide are not well monitored. Eruptions and surficial mass wasting activity can go entirely undetected. Low-frequency acoustic waves known as infrasound can provide valuable information for remotely detecting, locating, and modeling these hazardous volcanic processes. Infrasound is produced by sources coupled to the atmosphere, with signals often able to propagate further than seismicity in the solid earth, and without need for line of sight (e.g., cloud cover).The overall aim and contribution of this dissertation is toward improving volcanic event detection and localization workflows at local (< 15 km), regional (15–250 km), and remote (> 250 km) distances from the source, using a variety of sensor network configurations. These topics include effective use of dense regional infrasound networks, the impact of pre- and post-recording data noise-reduction, and the role of local infrasound monitoring in surficial mass wasting observations. Together, three principal projects demonstrate some of the benefits and limitations of various signal processing and detection techniques, site selections, and station hardware designs. Dense infrasound sensor deployments on a regional scale, such as the EarthScope Transportable Array (TA) in Alaska, can fill the observational gap between local and global scale volcano monitoring. The rolling TA uses ~80 km spaced single-sensor stations. In comparison, several small infrasound arrays (< 1 km), each with multiple sensors, already populate the area. Volcanic explosion detection and location using a backprojection (delay and stack) scheme achieves relatively limited success solely using TA-type sensors. Backprojection with the small arrays offers greater performance in terms of the number of events detected, and location accuracy. This is due to their shorter source-station distances, lower interstation spacing, higher azimuthal coverage, and more efficient wind-noise reduction hardware. Traditional array processing with the small arrays typically offers higher event detection rates than backprojection. Improvements can be made to the aforementioned detection and location schemes by using signal processing techniques to first isolate volcanic events from background noise sources. Such sources include incoherent wind noise, and pervasive ambient infrasound, primarily microbaroms (ocean wave-wave interaction source). Microbarom reduction typically improves array processing and enhances detection of weak events. Wind-noise reduction does not affect such detection rates, indicating sufficient turbulence can make coherent signal information unrecoverable. This approach does, however, improve the signal-to-noise ratios of isolated or stacked waveforms. At local infrasound monitoring scales relevant to relatively low-amplitude mass wasting signal detection, an additional ambient infrasound noise source with critical influence on detection capability arises from waterfalls, and other fluvial turbulence (river noise). At structurally unstable locations such as Mount Adams, Washington, large mass wasting events (e.g., lahars) are expected to propagate down the same fluvial channels responsible for this clutter. In addition to wind and microbaroms, this temporally and spatially varying fluvial infrasound can significantly affect detection thresholds. Wind noise precluded detection of several witnessed small debris flows below the summit. Conversely, such flows were not visually observed for a range of signals that are characteristic of mass wasting. A well-constrained glacial avalanche provides an exception, permitting assessment of detection and location schemes, and other techniques.

Nanomaterials, materials with one or more of their dimensions on the nanoscale, have emerged as an important field in the development of next-generation sensing systems. Their high surface-to-volume ratio makes them useful for sensing, but also makes them sensitive to processing defects and inherent material defects. To develop and optimize these systems, it is thus necessary to characterize these defects to understand their origin and how to work around them. Scanning probe microscopy (SPM) techniques like atomic force microscopy (AFM) and scanning tunneling microscopy (STM) are important characterization methods which can measure nanoscale topography and electronic structure. These methods are appealing in nanomaterial systems because they are non-damaging and provide local, high-resolution data, and so are capable of detecting nanoscale features such as single defect sites. There are difficulties, however, in the interpretation of SPM data. For instance, AFM-based methods are prone to experimental artifacts due to long-range interactions, such as capacitive crosstalk in Kelvin probe force microscopy (KPFM), and artifacts due to the finite size of the probe tip, such as incorrect surface tracking at steep topographical features. Mechanical characterization (via force spectroscopy) of nanomaterials with significant nanoscale variations, such as tethered lipid bilayer membranes (tLBMs), is also difficult since variations in the bulk system’s mechanical behavior must be distinguished from local fluctuations. Additionally, interpretation of STM data is non-trivial due to local variations in electron density in addition to topographical variations.

In this thesis we overcome some limitations of SPM methods by supplementing them with additional surface analytical methods as well as computational methods, and we characterize several nanomaterial systems. Current-carrying vapor-liquid-solid Si nanowires (useful for interdigitated-electrode-based sensors) are characterized using finite-element-method (FEM)-supplemented KPFM to retrieve useful information about processing defects, contact resistance, and the primary charge carriers. Next, a tLBM system’s stiffness and the stiffness’ dependence on tethering molecule concentration is measured using statistical analysis of thousands of AFM force spectra, demonstrating a biosensor-compatible system with a controllable bulk rigidity. Finally, we utilize surface analytical techniques to inform the development of a novel three-dimensional graphene system for sensing applications.

The atomic force microscope (AFM) is a widespread tool for the study of surfaces, as it allows for the unobtrusive measurement of nanoscale topography. AFM is also versatile, with variants that allow for the measurement of other quantities such as the local surface potential and Young’s modulus. A hurdle to the use of AFM systems is that interpretation of force probe data is nontrivial due to non-idealities of experimental systems including noise and artifacts. This thesis covers two computational techniques to overcome the limitations of AFM systems. First, finite element method simulations are used in conjunction with KPFM studies of vapor-liquid-solid grown silicon nanowires. The nanowires are studied with a variable electric bias applied across them, analogous to a field-effect transistor in operation. Then, batch analysis is applied to extract mechanical information from large amounts of AFM force spectroscopy data from tethered lipid bilayer membrane systems, which are important model systems for studying the mechanics of cell membranes.

Cognitive practice effects (PEs) are often ignored and likely delay detection of mild cognitive impairment (MCI). The replacement-participants method of PE adjustment subtracts the scores of demographically-matched, test naïve, replacement participants from returnees’ follow-up scores. The aims of this project were to determine whether this PE adjustment: 1) results in earlier MCI detection; 2) improves diagnostic stability, accuracy, and validity based on concordance with Alzheimer’s disease biomarkers; and 3) reduces costs of clinical trials.

The method was adapted for studies that did not recruit replacement participants by identifying “pseudo-replacements”, a subset of baseline participants that are demographically-matched to returnees at follow-up. Alzheimer’s Disease Neuroimaging Initiative data were extracted. Study 1 (baseline cognitively unimpaired [CU], N=722) and study 2 (baseline MCI, N=329) calculated PEs at 1-year follow-up. Study 3 added data from a 2-year follow-up (N=809). Biomarkers include CSF tau and beta-amyloid. Cost estimates were based on recent grant budgets and effects of PE adjustment on required sample size in a large clinical trial. Primary analyses included McNemar χ 2 tests, logistic regressions, and generalized estimating equations.Study 1: PE-adjusted follow-up scores led to significantly greater incident MCI as compared to PE-unadjusted scores (124 vs. 104; +19%). Significantly more MCI participants were biomarker positive when PE-adjusted scores were utilized (173 vs. 152; +14%). Cost estimates showed that replacements could save a large Alzheimer’s disease clinical trial over $5,000,000. Study 2: PE adjustment significantly reduced reversion to CU as compared to PE-unadjusted scores (57 vs. 80; -29%). Study 3: Significant PEs were found at both follow-ups for baseline MCI and CU samples. When adjusting for PEs, participants remained stable over time or declined, as expected in this older sample. Several PEs increased at the second follow-up (range: +9% to +133%). Adjusting for PEs with the replacement method leads to earlier and more accurate MCI diagnoses, reduces reversion rates, and is cost effective for large-scale studies. Unlike other methods, it also unmasks PEs even when scores worsen over time. The results indicate that PEs warrant greater attention and also suggest the potential value of developing PE norms.