Department of Atmospheric and Environmental Sciences Abstracts

Presenter(s): Layla Miller, Liz Mcdonald, Quintin Ashley, Brian Chiang, David Green, Katherine Jakubiak, Bernadette Leyba, Emely Martinez-Pichardo, Tyler Montgomery, Marcella Rodgers, Adriana Saint-Hilaire

Showcase Advisor: Justin Minder

Abstract: We are conducting a weather balloon launching in the ETEC yard. We will proceed as we usually do, connecting our radiosonde via satellite to the computer, attaching the parachute, setting up the helium tank, and finally filling up the weather balloon. We will be narrating everything to our audience as well. We are doing this for a program that involves 16 universities across North America, including UAlbany. Student teams from each institution are launching weather balloons during active storm events to collect data that can help improve understanding of atmospheric rivers and other high-impact weather systems.

Presenter(s): Matthew Case, Quintin Ashley, Kaitlin Bou, Brian Chiang, David Green, Gabriel Kramp, Micaela Maywald, Michael Miller, Chantel Nelson, Marcella Rodgers, Adriana Saint-Hilaire

Showcase Advisor: Ross Lazear

Abstract: As part of ATM 350 (Meteorological Data Analysis and Visualization), students create graphics from a reanalysis dataset for a chosen major weather event in the past two decades.  Each student will create one graphic for the Department of Atmospheric and Environmental Sciences' "Science on a Sphere", displayed on a loop during Showcase Day in ETEC-148.

Presenter(s): Autumn Johnnson

Showcase Advisor: Sujata Murty

Abstract: The Red Sea is a unique semi-enclosed marginal sea with high salinity due to high evaporation rates and limited freshwater input. The North Atlantic Oscillation (NAO) and the Atlantic Multidecadal Oscillation (AMO) are important climate drivers in the Red Sea, influencing ocean circulation and regional climate patterns. Yet, observational data prior to the satellite era are sparse and less reliable, limiting our ability to assess interannual to multi-decadal variability. Corals are useful climate archives to help fill in this gap as they grow continuously for centuries, incorporating trace elements into their aragonitic skeletons that vary in relation to changing environmental and climate conditions. Here, we present 180-year long, monthly-resolution records of coral Sr/Ca and δ18O values from the northern Red Sea to examine seasonal to multi-decadal sea surface temperature (SST) variability. Preliminary results show non-stationary spatio-temporal relationships between the NAO, AMO and the coral Sr/Ca and δ18O proxies.

Presenter(s): Matthew Sinnenberg

Showcase Advisor: Oliver Timm

Abstract: Hawaii experiences extreme precipitation that varies spatially and by season. Observational trends in extreme events are uncertain, necessitating a focused assessment of regional changes in the surrounding atmosphere. Here, I present an analysis based on ERA5 data to assess systematic changes in the atmospheric conditions associated with extreme precipitation. The fraction of attributable risk (FAR), a parameter that compares the relative risk of event occurrence, is employed to analyze rates for two periods: 1980-1999 and 2000-2019.

The FAR analysis displays statistically significant positive trends in temperature extremes during both seasons. Extremes in upward motions exhibit negative FAR values, while low level moisture, column-integrated water vapor, and precipitation show negative FAR values in the wet season, and positive values in the dry season. The lack of proportionality between the extreme precipitation FAR values and those of associated thermodynamic variables indicates an influence of other factors for the formation of extreme precipitation.

Presenter(s): Katelyn Jacques

Showcase Advisor: Aubrey Hillman

Abstract: Long Pond, located in Grafton Lakes State Park, has recently been accumulating large quantities of algae. This accumulation has raised concerns about whether or not algal blooms are common in the lake’s past, or if they are a newer development. This study investigates the history of algal blooms in Long Pond through analyzing sediment cores for biogenic silica (BSi), which looks at the presence of diatoms in the sediment. We hypothesized that if there is a larger amount of biogenic silica present in the sediment, then there were algal blooms at the time that the sediment was deposited. To test this, a wet-chemistry, alkaline extraction of biogenic silica was performed on the sediment that was sampled. We observed a relationship between biogenic silica, chlorophyll a, and metals present in the soil.

Presenter(s): Gabriella Fanti

Showcase Advisor: Sujata Murty

Abstract: The Red Sea plays a key role in thermohaline circulation due to its strong temperature gradients. However, sparse data sets limit our understanding of past climate. The chemical composition of coral skeletal material is an important tool for reconstructing past changes in climate. Specifically, the strontium-calcium ratio (Sr/Ca) found in the skeletal material of Porites sp. corals serves as an important proxy of temperature changes. Here we present coral Sr/Ca from the last few decades to reconstruct past changes in temperature. Instrumental sea surface temperature data is compared to the Sr/Ca data to determine the relationship between coral Sr/Ca and temperature. We find a robust relationship between coral Sr/Ca and satellite-derived sea surface temperatures. This record will allow for reconstruction of past climate and circulation patterns and give insight on how interannual and multidecadal climate mechanisms impact the central Red Sea’s sea surface temperature over recent decades.

Presenter(s): Patrick Miller

Showcase Advisor: Jeffrey Freedman

Abstract: The New York Bight (NYB) area is susceptible to different extreme weather events, which can induce disruptions to offshore wind operations and maintenance (O&M). Convective events such as thunderstorms can bring the risk of cloud-to-ground (CG) lightning, which can strike turbines. Multiple offshore wind farms are planned for operations in the next decade, so this work will summarize the potential lightning risk to these wind farms.

Using data from the National Lightning Detection Network (NLDN; Orville, 1983), this work creates a detailed lightning climatology for proposed offshore wind farm development, with focus on the New York Bight (NYB), encompassing regions adjacent to the New York and New Jersey coasts. This work also examines the potential effects of existing wind farms located on the Tug Hill Plateau in New York on CG strike frequency, by analyzing NLDN data from before and after the deployment of the turbines in the mid 2000’s.

Presenter(s): Robert Ford

Showcase Advisor: Brian Rose

Abstract: In certain parts of the Southern Ocean (SO), the density of seawater at the surface can increase enough that large-scale vertical mixing, called deep convection, occurs. This deep convection brings heat to the surface and melts large areas of sea ice (polynyas), contributing to an overall warmer Antarctic climate. In a climate model under constant preindustrial levels of CO2 with a high-resolution (~10 km) ocean, we show that there are regular multidecadal oscillations in deep convection. Previous studies focused on precipitation or wind variability to explain the oscillations. We propose instead a mechanism involving changes in the ocean circulation and salinity advection over the region where deep convection occurs. In the same climate model with a lower-resolution ocean (~100 km), these oscillations do not exist. While this raises questions about how realistic the oscillations are, the lack of a long observational record of the SO makes comparison difficult.

Presenter(s): Bella Condo

Showcase Advisor: Zheng Wu

Abstract: Artificial Intelligence (AI) weather forecasting models have been shown to outperform traditional physics-based models in terms of global-mean metrics. However, it remains unclear how these models perform specifically across the continental US (CONUS) and how their predictive skill varies by weather states (normal vs. extreme), seasons, and geographic locations within CONUS.  This study systematically evaluates and compares the performance of two AI forecasting models, AIFS and AIFS-CRPS, in predicting CONUS, using ERA5. Evaluating 2-m temperature and total precipitation across the nine climate regions within CONUS, we assess performance using root-mean-square-error (RMSE), spatial correlation coefficient (SCC), and ensemble spread. The results show that while the RMSE and ensemble spread generally increase with forecasting days, significant regional differences exist. Furthermore, forecast skill is lower during extreme atmospheric states compared to normal conditions. This study provides a comprehensive picture of how forecast skill varies across conditions and geographic regions in CONUS.

Presenter(s): Mike Miller

Showcase Advisor: Zheng Wu

Abstract: While AI-based models are rapidly transforming the field of weather prediction, it remains unclear about their performance in predicting extreme weather and their ability to learn the correct dynamics. In this study, we evaluate how well the AI models predict the high-impact January 23-27, 2026 snowstorm and the related dynamics. By gathering model output for varying lead times, we assess surface variables and process-oriented diagnostics (POD) of the storm. The results show that the prediction errors increase with lead times for surface variables. We further examined what dynamical process the AI models incorrectly predicted and how this affected the surface predictions of the event. While the AI models generally capture the dynamics of the snowstorm, they may have performed better in certain regions that the storm impacted than others. The results provide a comprehensive dynamical analysis for AI models and help improve our confidence in AI predictions for extreme events.

Presenter(s): Matt Case

Showcase Advisor: Zheng Wu

Abstract: Global AI weather forecasting models have been shown to outperform traditional weather models. However, it remains unclear how these models predict regional severe weather and whether they learn and capture the correct dynamics. The goal of this study is to evaluate the performance and dynamical consistency of two AI models, AIFS and AIFS-CRPS, in predicting the high-impact atmospheric river event that affected the U.S West Coast in late December 2025. By comparing with the ERA5 reanalysis, we assess how well the models predict precipitation and the associated dynamical processes in California at different lead times. Preliminary results indicate that notable differences emerge in key dynamic features, such as moisture transport, vorticity, and upper-level jet, with increasing lead times. However, the precipitation forecast errors are consistent with errors in the predicted dynamical processes. These results provide a comprehensive assessment of the AI models and help improve confidence in AI predictions.

Presenter(s): Matthew Lynne

Showcase Advisor: Aiguo Dai

Abstract: Under global warming, snow has become less frequent across the middle latitudes as mean cold-season temperatures have increased. Yet, it is unclear how historical shifts in extratropical cyclones have impacted changes in mid-latitude snowfall alongside warming temperatures. An impact-based extratropical cyclone tracking algorithm is applied over North America during October-April 1950/51-2023/24 and all precipitation regions associated with these cyclones are identified. Total precipitation associated with extratropical cyclones increased across much of central and eastern North America, yet much of this increased precipitation fell as rain, rather than snow.  

Extratropical cyclone associated snowstorm data were also used to train a machine learning model to predict the number of power outages caused by winter storms. Winter storm driven power outages have slightly increased in recent decades, with total precipitation and wind direction being the most important variables for power outage predictions.

Presenter(s): L Stephens

Showcase Advisor: Aubrey Hillman

Abstract: The climate across the Yucatan Peninsula is variable, and periods of drought have been recorded in Central America, coincident with the shift from the Classic to Postclassic Mayan Periods. It is not well understood how Maya populations adapted to the changing climate in the Yucatan during this time, especially in the peninsula’s southeastern region. Honey Camp Lagoon is located in this region and was continuously occupied during the Classic to Postclassic periods of Maya history. Through paleolimnology, we reconstructed climate variations during this period can see evidence of stable and, or wetter conditions that disagree with previous studies in Central America. To better understand why these conditions occurred, we are focusing on carbon and nitrogen isotopes present in core samples taken from Honey Camp to reconstruct the carbon cycle. Preliminary results suggest an open basin lagoon, supporting ideas that there was once a human mediated canal.

Presenter(s): Josilyn Bouffard

Showcase Advisor: Aubrey Hillman

Abstract: The Indian Summer Monsoon (ISM) serves as the main source of precipitation to India. Previous paleoclimate studies on the ISM have focused on marine or discontinuous terrestrial records. However, understanding terrestrial water availability is most important for assessing drought or flood periods which heavily impact local agriculture and food availability. Loktak Lake in Manipur, India offers an opportunity to assess terrestrial ISM strength back to Marine Isotope Stage 3 (MIS 3). Loktak has a shallow depth and a large surface area, making it ideally sensitive to moisture changes. We evaluate lake level and monsoon intensity using biogenic silica (BSi) and organic matter percentage. In addition, we use grain size to assess sediment source and delivery. Preliminary results from the 30,000-year record confirm the importance of insolation as a driver of ISM variability, but forthcoming higher resolution study will examine the role of other climate forcings.

Presenter(s): Ian Plummer

Showcase Advisor: Sujata Murty

Abstract: Multidecadal sea surface temperature (SST) and sea surface salinity (SSS) variability in the Red Sea influences marine ecosystems, hydrologic variability, and circulation. But, understanding of multidecadal SST and SSS variability in the Red Sea is limited by short and sparse observational data. Paired Sr/Ca and δ¹⁸O data from massive coral colonies provide centuries-long, monthly resolved records of SST and SSS that extend beyond the limited observational records. Here we present a 346-year Sr/Ca and δ¹⁸O record from a Porites sp. colony in the central Red Sea and combine it with a previously published northern Red Sea record to investigate basin-scale variability. Both sites exhibit coherent multidecadal variability in SST and relative SSS, which are anticorrelated with the Atlantic Multidecadal Oscillation but positively correlated with the Indian Ocean. These results highlight the Indian Ocean’s crucial role in modulating Red Sea surface conditions on multidecadal timescales.

Presenter(s): Sumar Hart

Showcase Advisor: Aubrey Hillman

Abstract: Studies in Central America have shown that periods of drought and climate shifts coincide with the end of the Late Classic Mayan period. The degree to which climate influenced Mayan populations in northern Belize during the transition from the Late Classic to Postclassic period is not well understood. While climate shifts throughout the northern Yucatán and Guatemala are more extensively studied, paleoclimate data in the southeastern part of the peninsula is limited. Here we analyze a 1500-year lake sediment record taken from Honey Camp Lagoon, located in northern Belize. We use geochemical and oxygen isotope data to reconstruct the environmental changes that occurred from hydroclimate shifts and human activity. Preliminary results suggest major changes in nutrient input to the lake coinciding with the beginning of the Postclassic period. Understanding the environmental response to these perturbations will strengthen our comprehension of ecosystem tipping points and assess consequences of future anthropogenic warming.

Presenter(s): Alex Blackmer

Showcase Advisor: Zheng Wu

Abstract: Communities across the Horn of Africa (HoA) rely on subsistence-based pastoral agriculture, which is greatly impacted by precipitation extremes. To support early warning systems, enhancing skill and transparency of precipitation predictions at subseasonal timescales is critical. This work explores the predictability of seasonal precipitation for the short rains over the HoA using interpretable machine learning (ML) models. We employ and compare predictions by several relatively simple ML models at 1-3-month leads. The results show that random forests persistently outperform other models. Further analysis shows that sea surface temperatures over the equatorial Pacific are generally the most important predictors at all leads. By applying the trained models to different datasets, the stability of the prediction skill across data sources indicates that the models have learned robust physical relationships and are resilient to dataset-specific noises. This study highlights the potential of simple non-linear models in S2S predictions, especially with limited sample size.

Presenter(s): Kathryn Rooney

Showcase Advisor: Sujata Murty

Abstract: The Red Sea exhibits a strong north-south temperature gradient and high evaporation, driving thermohaline circulation. In the north, cold winters increase surface water density, forming intermediate water that flows south into the Indian Ocean, impacting hydroclimate and ecosystems. Satellite SST data spans only the last 40 years, limiting understanding of circulation changes and climate drivers like ENSO and NAO. Massive corals provide long-term archives by recording seawater trace elements and isotopes. We analyzed a 70-year record of Sr/Ca and δ¹⁸O from a central Red Sea Porites coral to examine SST and salinity variability. Monthly coral Sr/Ca strongly correlates with NOAA OISST SST (r² = 0.93, p<0.01), confirming its reliability as a SST proxy. Using this calibration, we reconstruct seasonal to multi-decadal SST changes and assess ENSO and NAO influences, offering insight into Red Sea circulation and ocean-climate interactions across the 20th century.

Presenter(s): Daniel Harkin

Showcase Advisor: Justin Minder

Abstract: Winter storms frequently produce high-impact precipitation across the northeastern United States, yet accurately forecasting precipitation intensity and placement remains challenging for numerical weather prediction models. This study evaluates precipitation forecasts across the Northeastern US from the High-Resolution Rapid Refresh (HRRR) model during synoptic winter storms using the National Centers for Environmental Prediction (NCEP) Stage IV quantitative precipitation estimates as observations. For each case, a 24-hour verification period is defined from the maximum rolling 24-hour Stage IV accumulation. Stage IV QPE and HRRR modelled QPF are analyzed within storm-focused subdomains. Model performance is assessed through comparisons of precipitation intensity, spatial coverage of heavy precipitation, temporal forecast biases, and the Fractional Skill Score across multiple neighborhood radii.

Presenter(s): Elizabeth Smith

Showcase Advisor: Kristen Corbosiero

Abstract: Forecasting tropical cyclone (TC) intensification remains challenging due to the complex kinematic and thermodynamic processes that control storm evolution. This study examines the structural and environmental changes associated with the intensification of Hurricane Edouard 2014 by analyzing multiple observational datasets during its development. The analysis is divided into four categories: (1) environmental moistening, evaluated using dropsonde and aircraft flight-level data; (2) vortex structure, examining changes in the storm’s circulation and organization; (3) convective structure, assessed through radar and satellite observations; and (4) lightning bursts, using lightning detection data in the storm’s outer and inner rainbands. By comparing the timing of these processes, this study aims to determine whether the alignment of these factors corresponds with the early onset of intensification in Hurricane Edouard.

Presenter(s): L Stephens

Showcase Advisor: Aubrey Hillman

Abstract: The climate across the Yucatan Peninsula is variable, and periods of drought have been recorded in Central America, coincident with the shift from the Classic to Postclassic Mayan Periods. It is not well understood how Maya populations adapted to the changing climate in the Yucatan during this time, especially in the peninsula’s southeastern region. Honey Camp Lagoon is located in this region and was continuously occupied during the Classic to Postclassic periods of Maya history. Through paleolimnology, we reconstructed climate variations during this period can see evidence of stable and, or wetter conditions that disagree with previous studies in Central America. To better understand why these conditions occurred, we are focusing on carbon and nitrogen isotopes present in core samples taken from Honey Camp to reconstruct the carbon cycle. Preliminary results suggest an open basin lagoon, supporting ideas that there was once a human mediated canal.

Presenter(s): Harrison Miller

Showcase Advisor: Jake Mulholland

Abstract: Dry lightning poses potential for wildfire ignition because lightning strikes can occur with little to no accompanying precipitation, allowing fuels to remain dry and receptive to ignition. Improving situational awareness of environments favorable for dry lightning can support more effective forecasting and response to wildfire ignition potential. This study investigates the environmental conditions associated with dry lightning events and evaluates how effective past operational Fire Weather outlooks have performed with such events. Cloud-to-Ground lightning data from the National Lightning Detection Network were combined with Stage-IV precipitation analyses to classify strikes as ‘dry’ or ‘wet'. This resultant dataset, in addition to reanalysis data, was used to examine the associated environmental conditions that favored dry lightning occurrences. These relationships were then used to develop a post-processing tool via machine learning techniques, specifically forest-based decision models, to identify regions where dry lightning was most likely to occur based solely on environmental conditions.

Presenter(s): Kailyn Oreglio

Showcase Advisor: Aubrey Hillman

Abstract: Linos Lake is located in northern Belize and is in an area that has been occupied by people for thousands of years, notably during the Mayan Civilization (ca 2000 BCE to 900 CE). We hypothesize that land use change and agriculture within the Linos watershed impacted the elemental components within the lake, specifically potassium and phosphorus. The motivation behind this study is to observe how agriculture affects the surrounding bodies of water since it can have an impact on aquatic life. During periods of increased agricultural activity, we expect to find higher concentrations of potassium and phosphorus in the sediment cores from the lake. To test this, we measured the elemental concentrations of sediment cores from three different locations in Linos using an ICP-OES. Through this, we have identified the impact of human activity on the lake.

Presenter(s): Quintin Ashley

Showcase Advisor: Jake Mulholland

Abstract: Low-level vertical wind shear plays a crucial role in squall line dynamics by generating horizontal vorticity in conjunction with the horizontal vorticity created by the vertical circulation of the cold pool. Interactions between environmental low-level shear and a cold pool’s gust front strongly influence the size and tilt of squall line updrafts, which in turn, dictates the intensity of squall lines through modulating entrainment-driven dilution. A matrix of nine idealized numerical model simulations of squall lines with varying low-level shear and CAPE was used to isolate these separate effects on the intensity and maintenance of squall line updrafts. The results of these simulations are consistent with the classical “RKW Theory” and suggest that environments with stronger low-level shear and higher CAPE produce squall lines with wider, stronger, and more sustained convective updrafts than environments with weaker low-level shear and lower CAPE.

Presenter(s): Bella Condo

Showcase Advisor: Jake Mulholland

Abstract: Organized severe convection is difficult to forecast because the conditions conducive to it vary widely and occur only a few times each year. To better understand these events, this project applies self-organizing maps (SOMs), an unsupervised machine learning technique, to group Southern Great Plains environments with similar characteristics. Severe weather days were defined as outlooks where the Storm Prediction Center (SPC) issued an Enhanced or higher risk from 2014 to October 2025. ECMWF Reanalysis 5th Generation (ERA5) geopotential heights on severe weather days were used to create the SOMs. Using a 3×2 SOM configuration, these cases cluster into six distinct synoptic-scale circulation regimes. Preliminary results indicate that higher SPC risk categories occur within specific SOM nodes rather than mapping to a single large-scale pattern. Identifying these synoptic patterns provides a clearer picture of atmospheric setups capable of producing extreme severe weather.