Volatile organic compounds (VOCs) are air pollutants that can react in the atmosphere to form ozone and other secondary products, which can be detrimental to human health. Despite a decrease in emissions in the Los Angeles (LA) area over the past few decades, a 2019 study showed that LA currently has the most air pollution related deaths of any US city. In this project, we aimed to see how the VOC emission profile in the LA Basin has changed in the past decade. Data from whole air samples collected on past SARP flights over the LA Basin were analyzed in the Rowland-Blake laboratory using multi-column gas chromatography. Principal component analysis (PCA) was used to reduce the dimensionality of the data, and a trained classifier with 10-fold cross validation was used to confirm that there were significant differences in each year’s data. Factor analysis was then used on the 2011, 2014, and 2017 datasets to characterize and identify primary pollution sources in those years. While oceanic, combustion, and natural gas factors were identified for each year, biomass burning and biogenic emissions were specific to 2011, hinting that the LA area may have shifted away from those sources later in the decade; however, further analysis would need to be done to support this theory. Nonetheless, this method is a valuable tool to monitor how emissions have changed and could be used in the future to help shape local air pollution policy regulation.

Dimethyl sulfide (DMS) is a compound primarily released into the atmosphere from phytoplankton in the ocean. Because of this and its relatively short photochemical lifetime (~1 day), it is typically found at low altitudes near the coast. NASA SARP airborne data from 2014-2019 indicates that there may be an inland source of DMS that could contribute to aerosol loading in the Imperial Valley. During the SARP flights, air samples were collected using Whole Air Sampling (WAS) canisters and were later analyzed at the UCI laboratory using multi-column gas chromatography to quantify concentrations of over 100 volatile organic compounds (VOCs). During this time period, samples taken in this area revealed enhanced concentrations of DMS. Enhancements in DMS were expected over the Salton Sea as a result of phytoplankton activity. However, a surprising finding was a greater enhancement of DMS north of the Salton Sea in a suburban area with moderate farming activity. Previous SARP studies using data obtained from 2014 estimated the impact of DMS in the Central Valley and the resulting effect on aerosol loading. These previous studies estimated that in the Central Valley DMS oxidation could account for to up to 3% of sulfate-based aerosols. Data for this study suggests the percentage of sulfate-based aerosols produced from DMS could be responsible for as much as 10% of the sulfate aerosol loading in the Imperial Valley. Our findings suggest that inland emissions in the Imperial Valley have a significant effect on DMS concentrations; therefore, I recommend dedicating flight plans to this area to better understand the impact on sulfate aerosol loading in the region.

Across eleven years of SARP flights, airborne whole air sampling data contains two instances above 10,000 feet where toluene concentrations reach sustained enhancements. Toluene is a toxic compound with adverse health effects and a variety of sources including industrial processes, fossil fuel emissions, and fires. The irregularity occurred once in 2011 during a spiral maneuver over Delano, CA, and once in 2015 over Palm Springs. Both plumes were measured within the 10,000-15,000 feet range. At a glance, the air masses appear to be similar in composition for a variety of compounds, namely toluene and dimethyl sulfide. However, by combining a variety of data visualization tactics to analyze the plumes, it becomes clear that the two plumes are not related by any common source or situation. Three-dimensional structure analysis reveals that while the 2011 plume arises from a single source, the 2015 air mass has multiple sources and three distinct sections of interest. In an effort to make this technique more robust, nitrate/alkane ratios were used to estimate ages for different sections of the plume, but the work was largely inconclusive. Using these toluene spikes as a case study, this project develops a strategy to explore plume structure and draws conclusions therefrom, which can be a valuable tool when studying atmospheric irregularities.

Airborne data from SARP 2019 Flights 1 and 2 over the California Central Valley and Edwards Air Force Base (AFB) respectively, showed enhanced concentrations of 1,2-dichloroethane (1,2-DCE) at altitudes between 20,000 and 40,000 feet. Additionally, SARP 2014 Flight T1, also over Edwards AFB, showed enhanced 1,2-DCE at similar altitudes. Enhanced concentrations of 1,2-DCE can pose hazardous health effects to humans and have the potential to release stratospheric chlorine that could contribute to the depletion of the ozone layer. Dichloromethane (DCM) and carbonyl sulfide (OCS) were also found to be enhanced on the same flights at high altitudes. DCM also has the potential to release stratospheric chlorine while OCS can contribute to aerosol formation. Other volatile organic compounds (VOCs) with much shorter lifetimes such as ethene, isoprene, and xylene were below the limit of detection in these areas. The absence of short-lived compounds and the high altitudes where these VOCs were observed suggests that the source of the 1,2-DCE, DCM, and OCS is long-range transport. Using the National Oceanic and Atmospheric Administration’s (NOAA) meteorological trajectory program, HYSPLIT, a four-day backwards trajectory was calculated for all three flights, indicating that the air plumes originated in Eastern Asia. The air masses were then aged using nitrate isomer ratios as well as ethyne to carbon monoxide (CO) ratios to determine if the age of the air plumes supported the HYSPLIT trajectories. Additionally, it was discovered that SARP 2018 Flight 4 overflew Edwards AFB in the same areas flown in 2014 and 2019; however, the 1,2-DCE concentration was not enhanced at high altitudes. Backwards trajectories and air mass aging were used to examine this flight and connect the results to the hypothesis that the air plumes in 2014 and 2019 were transported from Eastern Asia.

Airports are known emitters of air pollutants, and often have elevated levels of less prevalent compounds, such as xylene and toluene (Levy et al. 2008). Both xylene and toluene are toxic to humans, with their major anthropogenic sources being paints, solvents, and fuels. In this study, we utilized HYSPLIT trajectories and wind roses to understand wind patterns and meteorological conditions surrounding the Los Angeles International Airport (LAX) during the month that SARP flights occur. A previous study found that LAX affects downwind communities with elevated concentrations of PM 2.5 (Hudda et al. 2014), but did not focus on how volatile organic compounds (VOCs) are affecting communities downwind, such as Inglewood, South Los Angeles, and Huntington Park. Our study uses airborne data from six separate years, spanning 2011 to 2017, of SARP DC-8, P3-B, and Sherpa aircraft flights. While it is often difficult to estimate point source emissions from high-altitude flight data, our missed approaches into LAX provide low-altitude data surrounding the movement of xylene and toluene. Our study also uses a framework from the California Air Resources Board’s, known as CARB, Assembly Bill 617 (AB 617), which identifies disadvantaged communities with elevated levels of air pollutants in the Los Angeles Basin. We utilized isopentane to xylene/toluene ratios to trace the source of the pollutants to the airport. Our results find higher isopentane/VOC ratios further downwind from LAX, indicating gradually decreasing levels of xylene and toluene farther from LAX. This suggests that LAX may be affecting these communities with large continuous enhancements of VOCs, not limited to just xylene and toluene. Enhanced levels of VOCs in these communities may contribute to higher rates of health issues among residents – a major issue of environmental justice.

Bromoform (CHBr​3)​ is a natural volatile organic compound (VOC) outgassed from phytoplankton, macroalgae, and water treatment processes. This compound is the largest contributor of inorganic bromine, a known ozone destroyer, to the lower stratosphere (Dvortsov et al., 1999). In 2014, mean concentrations of CHBr​3​ increased by a factor of ~2.5-5 relative to other years in the 2009-19 SARP WAS record. Whole air samples, collected on June 24, 2014, by the NASA DC-8 for SARP 2014 Flight 2, exhibited a variable enhancement of CHBr​3 throughout the boundary layer of the LA basin with concentrations of ~10-15 ppt at Long Beach and Mt. Wilson and ~20-28 ppt at Ontario Airport (ONT). HySPLIT back-trajectory models were used to locate possible sources of each observed enhancement in the LA basin. Initiated at Long Beach, Mt. Wilson, and ONT, these models predicted boundary layer airmasses of marine origin to have travelled down the Californian coast and over the Channel Islands during the 48 hours preceding the flight. Trajectories corresponded with both a coastal California phytoplankton bloom, observed by the MODIS-Aqua satellite and located using NASA Earthdata, and kelp forests off the Channel Islands. Both of these features are sources of CHBr​3 and it is likely that the observed enhancements were a result of their combined outgassing. Additionally, monthly mean CHBr​3 concentration at Mt. Wilson and monthly mean sea surface temperature (SST) at the Santa Monica Southern California Coastal Ocean Observing System (SCCOOS) station were found to be positively correlated. Further research should be conducted regarding the contribution made by kelp to this enhancement as macroalgae are known to increase CHBr​3 production under temperature stress (Stemmler et al., 2015), and the enhancement was observed during the Californian marine heat wave known as the ‘warm blob’ (Peterson et al., 2015).

Whether you are buying a soda from a vending machine or cooling yourself in your car on a hot summer day, hydrofluorocarbons (HFCs) are all around us and a luxury many of us could not do without.  HFCs act as a refrigerant and were first produced in the mid-1980s as an alternative to chlorofluorocarbons (CFCs). CFCs are extremely long lived compounds that are known to catalytically deplete the ozone layer. Under the Montreal Protocol in the late 1980s, CFCs were effectively replaced by HFCs. This, however, was meant to act as a short-term solution, as HFCs have a global warming potential orders of magnitude greater than carbon dioxide (CO2), which can have drastic warming effects on a global scale. Using the FBProphet package in the Python programming language, and data from previous NASA SARP Whole Air Sampling groups, forecasts were made for several HFC concentrations in the Los Angeles region. Results displayed a concerning trend in HFC concentrations. Turning our attention to finding methods of recycling HFCs and exercising alternative refrigerants may be in our best interest. This project presents an accessible method of forecasting data and gives a comprehensive survey of the adverse implications this observed upward trend of HFCs could have in the close future.

Active fire temperature has important consequences in understanding burn severity, the role of fuels, and predicting fire spread, but because of the dangers presented by wildfire, in-situ temperature data are difficult to collect. Using the high-resolution data collected by the Airborne Visible/ Infrared Imaging Spectrometer (AVIRIS), fire temperature can be modeled accurately from the radiance within burning pixels. In SARP 2017, Keegan Quigley developed an active fire temperature retrieval model that was robust across multiple pixel scales. However, he was only able to apply it to the Zaca and Sherpa Fires. In this work, I used Quigley’s model on the Sherpa and Thomas Fire to validate his model and determine how portable it was across multiple fires. Like past models, this model assumes fire is a blackbody, but unlike other models it makes the assumption that temperature is continuous and interdependent within a pixel. Using Planck’s Law, 1201 maximum temperatures are forward modeled for a fire scene, corresponding to the possible maximum temperatures of 300-1500K within a pixel. The modeled fire is then adjusted accordingly to prevent false temperature measurements due to background radiance and saturation within the AVIRIS instrument. Measured fire spectra are then fit to these modeled fires by limiting Root Mean Squared Error (RMSE) between them. Results showed Quigley’s model to be valid and accurate compared to past models, as it retrieved fire temperature with minimal error for the Sherpa and Thomas Fire. The rate of temperature decay within a pixel was found to be a critical parameter that varies across fires and between scenes of fires. With careful attention to this variable the model is shown to be portable, and it should be put into use as the standard of fire temperature retrieval statistical models. Analysis of the Thomas Fire in this work showed the extreme conditions and the unusually strong Santa Ana winds that led to the fire spread resulted in high maximum temperatures, and presented evidence indicative of spot fires.

Wildfires in California have been tied to the natural ecosystem for decades, occurring naturally during the drier, summer and fall months with many plants adapted to fire, as evidenced by the germination of vegetation in response to the heat, smoke, and ash. Due to an increase in global temperatures and increasing vapor demand, California wildfires have seen a general upward trend in occurrence increasing frequency and changes in seasonality. During the years of 2008 and 2009, three wildfires occurred within a year of each other in the Santa Barbara area during different seasons; the Gap Fire in July 2008 (summer), the Tea Fire in November 2008 (winter), and the Jesusita Fire in May 2009 (spring). The timing and close geolocation of these fires provides the opportunity to explore post-fire regrowth and determine if there is a difference between seasonal impact on vegetation recovery in the affected areas. This was done by using Airborne Visible/ Infrared Imaging Spectrometer (AVIRIS) data acquired over Santa Barbara from 2004 and between 2009-2017. These data were used to construct Relativize differential Normalized Burn Ratios (RdNBRs); RdNBR from the Jesusita and Gap fires indicated steady vegetation recovery, but data from the Tea Fire showed a different and more varied trend. In addition, compared to a true color August 2004 image of the area of interest viewed through ENVI, a true color image of 2018 data was contrasted with spectral analysis to determine if the same quantities of general vegetation in the boundaries of the fires grew back to pre-fire levels or changed. The results can be used to determine the severity and impact of wildfires that occur out of season, especially with the uptick in wildfire prevalence.

Certified organic food must comply with stringent national standards for the production, handling and processing of products. Certification and monitoring of compliance for organic fields are expensive, time-intensive and arduous processes requiring on-site inspection. Identification of management practices via satellite imagery would lower the market barrier to entry and remove obstacles for efficient monitoring. Discriminating crop types is difficult because of varying biophysical traits, development stages, management practices, regional weather and topography, and the timing of plantings (Galvão et al. 2018). Consequently, spatial and temporal phenological differences can result in crop misclassification (Dudley et al., 2015). Even more challenging is spectrally separating organic and conventional fields. This study aims to distinguish between organic and conventional grape agricultural fields using Multiple Endmember Spectral Mixture Analysis (MESMA). Using a georeferenced crop-composition shapefile for Kern County, CA overlain on a June, 2018 Airborne Visible/ Infrared Imaging Spectrometer (AVIRIS) flight line, all values below a 0.3 NDVI were masked out. Then, regions of interest (ROIs) for organic and conventional grape crops were specified in ENVI. ROIs were sampled to create entries in the spectral library. To optimize endmember selection and improve performance in mixing models, we used the Iterative Endmember Selection (IES) process. The constrained mixing model derived in MESMA produced a pixel classification accuracy of 71% and a 0.58 kappa value. While producer’s accuracy between organic wine and conventional wine crops achieved 87.64% and 89.43% accuracy respectively, organic and conventional grape crops attained 55.60% and 64.40% accuracy. Degrading the image to the spectral resolution of the Sentinel-2A satellite (from 224 to 13 spectral bands) resulted in a kappa value of 0.57, suggesting that readily-available satellite imagery can be used to classify fields as organic or conventional at the same accuracy as hyperspectral imagery. Additional investigation as to how the spatial heterogeneity of fields, chemical treatments, and soil cover account for discernable spectral differences will help determine future uses of satellite imagery in mapping crop attributes.

Although it makes up less than one percent of the USA’s farmland, the Central Valley of California provides approximately 25% of the USA’s food table: however, agricultural practices are challenged by extensive areas of salinity and sodicity, particularly in the western San Joaquin Valley (Scudiero, 2015). Areas of substantial salinity and sodicity decrease plant growth and crop yields, making soil salinity a threat to the agricultural production in the region. We used Random Forest machine learning along with USDA soil electrical conductivity ground truth information to classify salinized soils using Airborne Visible/ Infrared Imaging Spectrometer (AVIRIS) imagery. Random Forest was applied to bare soil imagery, as determined by a Normalized Difference Vegetation Index (NDVI) using a threshold of 0.3, and used to evaluate spectral differences in soil salinity based on the USDA’s soil salinity classification system. The spectral differences were then used to map soil salinity within the San Joaquin region sampled by the AVIRIS image. The remote sensing of soil salinity could provide a rapid and cost-effective method for mapping regional soil salinity, while also being able to determine within field variability and abrupt changes between fields. Creating a regional soil salinity map can better inform producers and land owners on their land management practices.

The 2012-2016 California drought had adverse effects on vegetation leading to reduced agricultural yield and increases in the number of wildfires. The reduction of soil water content in shallow soils is well known, including the response of grasses and shrubs; however, less is known about deeper soils that are important for the health of trees. To understand the water content of these deeper soils, we built a geophysical model to describe the evolution of water content with depth in simulated soils. The transport of water in soils is naturally diffusive, not convective, and follows a modified diffusion equation ∂_t u=∇⋅[k∇u] where k is the hydraulic conductivity. However, because southern California’s grasslands are arid, the soils are unsaturated, leading to a hydraulic conductivity dependent on the water content itself, k = k(u). This causes the problem to be naturally nonlinear. The geophysical model was developed for simulated soil profiles for southern California grasslands, parameterized using properties of Sedgwick ranch in Santa Barbara, California. We selected soil properties to estimate unsaturated hydraulic conductivity, elevation to describe the domain of integration, meteorological data to characterize the influxes of water into the system due to precipitation, and plant cover derived from Airborne Visible/ Infrared Imaging Spectrometer (AVIRIS) imagery to describe outfluxes of water due to evapotranspiration.

Land surface temperature (LST) has been linked to higher rates of mortality and morbidity in human populations. City-scapes tend to have a warming effect, while vegetation has been found to have a net cooling effect. From 2012–2016, California experienced extreme drought, provoking strict water restrictions in multiple counties. This provides an interesting insight into the potential effects of socioeconomics on neighborhood-scale green vegetation coverage and consequent land surface temperature. Previous studies examining the effects of green vegetation, elevation, and coastal distance have shown fractional vegetation cover is the dominant factor in determining LST (Tiao 2019). In this project, we examine how vegetation cover and LST differ before and at peak drought conditions in relation to income in San Mateo County. We first applied multiple-endmember spectral mixture analysis (MESMA) to 2013 and 2016 Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) images in order to classify land surface types. MODIS/ASTER (MASTER) thermal imagery from the HyspIRI Airborne Campaign was then used to derive brightness temperature (as a measure of LST) for the two years. In order to understand the relationship between LST, green vegetation cover, and income, these data were analyzed using census tract maps from the American Community Survey (ACS) for the two years. Green vegetation fraction was then calculated for each census tract to determine vegetation coverage. We expect to find a negative relationship between LST and income, and a positive correlation between fractional green vegetation and income.

Photovoltaic solar panels are among the fastest growing and cheapest of clean energy sources. Over the past decade, the cost of producing a kilowatt hour of solar energy has dropped from over $350 to about $50 per kilowatt hour, making solar energy less than half the price of its dominant competitor, coal power (Lazard). What has been solar power’s primary drawback—the ability to store energy during nighttime or cloudy days—will be further mitigated as battery technology continues to advance. While solar panels continue to become more prevalent, the direct impact of low-albedo PV panels on earth’s radiation budget remains in question. Solar panels have relatively uniform spectral profiles, making them well suited for mapping with hyperspectral remote sensing. We used Airborne Visible/ Infrared Imaging Spectrometer – Next Generation (AVIRIS-NG) hyperspectral imagery from June 2014 flights over the Santa Barbara region to compile a library of spectral profiles for panels, along with shingle, red-tile, and commercial rooftops. Multiple Endmember Spectral Mixture Analysis (MESMA) was then used to map subpixel fractional coverage of solar panels. These panel fractions were then compared to corresponding land surface temperature (LST) derived from 2014 MASTER data. The relationship between the two could help define how the presence of panels is directly heating or cooling our environment.

Salinity is an important physical parameter governing a variety of biological processes. Accurate salinity measurements can inform on habitat distribution, species richness, and plankton biomass in dynamic estuarine environments. In estuary systems, where salinity can vary between 15 and 30 practical salinity units (PSU), a remote sensing method of salinity estimation would be useful for large-scale environmental monitoring. Here, we estimate salinity using hyperspectral imagery from the NASA Airborne Visible InfraRed Imaging Spectrometer (AVIRIS) and AVIRIS Next-Generation (AVIRIS-NG), in addition to multispectral imagery from Landsat-8. We use an algorithm developed in Housekeeper et al. (2020) to retrieve estimations of colored dissolved organic matter (CDOM) absorption directly from remote sensing data. The CDOM product is related to salinity using an empirical relationship developed in situ between salinity and CDOM absorption, using data collected in 2019. We apply this method to AVIRIS and AVIRIS-NG images of San Francisco Bay in 2016 to produce salinity maps. AVIRIS-derived salinity deviates on average by 7.2 PSU with the nearest in situ measurements, while AVIRIS-NG salinity deviates by 2.1 PSU. These deviations are attributed to year-to-year changes in salinity and differences in tidal phase between in situ and image data. Correcting for these error sources results in an average difference of .48 PSU from the nearest in situ measurements for the AVIRIS data, and .31 PSU for the AVIRIS-NG data. We also estimate salinity using a Landsat-8 image for which concurrent in situ data were available. Resulting salinity estimations differ from in situ values by an average of 1.54 PSU. Salinity error increased for sites further south in the bay; the average difference is .52 PSU for sites north of San Mateo bridge, and 3.06 PSU for sites south of the bridge. This difference is attributed to the optical complexity of water in the southernmost region of the bay, where high suspended sediment concentrations produce a spectrum beyond the foundational data of the CDOM algorithm. These results suggest that this method of salinity estimation is likely valid for waters with low to moderate suspended sediment concentrations; with the right spectral bands and available in situ data salinity can be estimated in San Francisco Bay from remote sensing data fairly accurately.

Coral reefs are foundational ecosystems that become highly stressed by occurrences such as destructive fishing practices or climate change. In the Philippines, coral reefs protect the islands from storms, while also providing food security and eco-tourism, helping to boost the economy.  There is a lack of previous work mapping changes in coral reef cover or coral health in the Philippines. Mapping coral reefs is challenging, but recent satellite data contains high enough resolution to perform the task using remote sensing. This project uses Sentinel-2 satellite data and the ESA’s Sentinel Application Platform (SNAP) to develop maps of coral reef changes in the Philippines within a two-year span. We analyzed changes in three separate islands’ coral cover with a mixture of supervised classification and radiometric normalization using the Sen2Coral software plugin. Consistency in our methods is demonstrated in the classification of Coron, Palawan from 2019 to 2020, where illegal practices such as dynamite fishing had been reported. We identified a possible shift in the coral-algae ecosystem state of Boracay as a response to the popular tourist destination being closed to visitors in 2018. Changes in coral cover are also seen off the coast of Jomalig and Patnanungan Islands in Quezon. Classification results from these islands are further confirmed by examining Google Earth and NASA Camp2Ex imagery. This project explores the use of remote sensing as a tool to monitor coral reefs over time. When standardized using freely available data and software, this will empower others to analyze coral health in their area.

Nearly 70% of all freshwater on Earth is in ice caps and glaciers, and 10-11% of land is covered in glacial ice. These glaciers contain important nutrients that life depends on–such as iron–which is released as the glaciers melt. Glacial melt carries an influx of nutrients, which can form a perfect environment for phytoplankton blooms when deposited into the coastal ocean. In the Prince William Sound in Alaska, the rate of glacial loss has increased dramatically over the past few decades; the Columbia Glacier in particular is one of the most rapidly changing glaciers in the world, having retreated more than 20 km since 1980. As sea temperature rises, glacial melt increases, which influences phytoplankton blooms and in turn alters the foundation of the aquatic food chain. Here, we investigate the relationship between glacial melt and phytoplankton dynamics in Prince William Sound using remote sensing. We use chlorophyll-a data from the MODIS Aqua satellite to construct a time series of phytoplankton biomass from 2014-2020, and imagery from Landsat 8 to quantify annual glacial melt.  Results show that an average early spring bloom of 4.31 +/- 2.06 mg/m3 occurs within the sound that is not impacted by glacier melt, but a later average autumn bloom of 9.01 +/- 5.64 mg/m3 also appears following significant melt indicated by an average glacial melt index of 10 +/- 8%. These results provide evidence of the effect glacial melt has on phytoplankton dynamics, which is particularly important to understand as climate change progresses.

Algal bloom events occur annually off the coast of California between the early spring and late summer. Warmer Sea Surface Temperatures (SSTs), low salinity, upwelling of nutrients (predominantly nitrogen and phosphorus) combine to create an environment conducive to algal growth. Although these blooms usually last between one or two weeks, a bloom in 2020 dominated by the dinoflagellate Lingulodinium polyedra lasted approximately six and a half weeks. Monitoring algal blooms with satellite remote sensing gives consistent coverage of a bloom’s progression and enables us to characterize the physical and chemical changes affecting the coastal environment.  We compare the April-May 2020 L. polyedra bloom off the Southern California Coast with a September 2013 bloom by constructing a time series of chlorophyll a and sea surface temperature using data from the Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua and Visible Infrared Radiometer Suite (VIIRS). A red tide algorithm (Su et al, 2016) is used to indicate specific areas of the red tide event. We also use wind speed from the Iowa Environmental Mesonet (IEM) and precipitation data from Surface meteorological dataset based on Portable Remote Imaging Spectrometer PRISM and North American Land Data Assimilation System (NLDAS-2) to identify contributing effects of the atmosphere and rainfall events. The time series revealed that the 2020 bloom experienced a sustained, high SST anomaly (~3°C), below average coastal wind speeds and a 5-fold anticipated increase in precipitation during the season. The 2013 bloom experienced similar conditions, but of lesser magnitude which contributed to its relatively short duration. L. polyedra blooms thrive in stratified conditions, which are amplified by higher SSTs, lower salinity, and low wind speeds. Quantifying these conditions highlights their relationships to algal bloom progression. These attributes should continue to be monitored as climate change models predict these factors to increase in magnitude.

California’s largest lake, the Salton Sea, has been steadily declining in water surface level since 2003 and increasing in salinity from approximately 43 ppm in 2004 to approximately 74 ppm in 2020. By studying how life adapts in these dramatically shifting conditions, the Salton Sea acts as an analog to other highly salinated environments, such as hypothesized Martian brines. Here, we examine how algal biomass in the Salton Sea has responded through time to increases in salinity and decreases in water surface level. We combined in situ chlorophyll and salinity data collected from 2003-2020 by the Bureau of Reclamation with chlorophyll, lake area, and temperature derived from Landsat 8 satellite imagery from 2013-2020 to examine how changes in the physical environment affected lake biology over time. Landsat 8 Level 2 (L2) data provided by the USGS paired with a blue-green chlorophyll algorithm from Ha, et al. 2017 was determined to be the best option after comparison with in situ data. Lake area was acquired using the Modified Normalized Difference Water Index (MNDWI) algorithm proposed by Wu et al., 2018. After comparing in situ, Landsat 8, and Airborne Visible / Infrared Imaging Spectrometer (AVIRIS) chlorophyll concentration averages to salinity, we found that there was no clear relationship. However, using Landsat 8 data, we did find a noticeable increase in the maximum annual average of chlorophyll by 1.1 μg/L per year from 2013 to 2020. Additional factors that may be offsetting the increasing salinity’s effect on chlorophyll may include an increase in total phosphorus from 0.085 mg/L to 0.462 mg/L during this time. Future research should investigate the impact of these other factors on chlorophyll as well as the application of AVIRIS to detect microbial community changes over time as measured by the pigments carotenoid and bacterioruberin as proxies for lake biology.

Wildfires are becoming more frequent and severe in dry, arid regions of California. Although the effects of wildfire on vegetation and air quality are well-studied, the effects on lakes and reservoirs are less documented. A unique opportunity to study the effects on lakes arose when Lake Casitas was surrounded on its western side by the Thomas Fire in late 2017, right after a major drought ended. We construct a time series of lake area and suspended particulate matter (SPM) using Landsat 8 imagery from 2014 – 2020 to observe how the lake responded to the fire. We hypothesize that lake area and SPM increased following the Thomas Fire because more runoff would enter the ecosystem after vegetation around the lake was burned. We observed an increase in lake area from 5.7 km2 before the fire to 6.3 km2 after the fire. An increase in SPM from 9 gm-3 to 11 gm-3 after the first precipitation event following the fire was observed near the inflow from one tributary feeding the lake. Other sites showed a decrease in SPM values from 8 gm-3 to 3 gm-3, which may be due to the implementation of sediment curtains by the Casitas Municipal Water District. Both lake area and SPM returned to initial patterns between six months and a year after the fire was put out. Future work should look for similar trends in other lakes and regions and study other parameters such as chlorophyll concentrations or how long-term erosion affects the land surrounding the lake.

The ocean and atmosphere are inherently related, meaning that phenomena in the ocean have an undeniable effect on our sky and vice versa. One such phenomena occurred in 2015-16 off the West Coast of North America: a marine heat wave of anomalously high sea surface temperatures (SST) reaching from 1- 4℃ above normal in the NE Pacific. Here, we explore 1) the relationship between cloud cover and the marine heatwave event to see if abnormally high SST leads to anomalous events in local cloud cover, and 2) The effect of cloud cover on satellite derived SST data to see if cloud cover increases error in these measurements. At two regions on the west coast of North America, cloud cover data from the MERRA-2 (Modern-Era Retrospective analysis for Research and Applications) model was compared to ship and buoy SST data from iQuam (in situ SST Quality Monitor), infrared SST measurements from MODIS (Moderate Resolution Imaging Spectroradiometer), and microwave SST values from the WindSat instrument aboard the Coriolis satellite. For each location, a time series was created from 2010-17. Climatology was removed to isolate the fluctuation of SST and cloud cover without the presence of seasonal variation, while SST from MODIS and WindSat was compared to in situ data to determine error. We found that cloud cover has a strong inverse relationship to SST (R2 = 0.606), which fluctuates seasonally. On a longer time scale, we find no statistically significant cloud cover deviation to match the warm anomaly in 2015-16. We also conclude that on average, SST from satellites exhibit the most error at cloud cover minima, having an average error of 1.67% for MODIS and 5.14% for WindSat. These results signify that SST measurements from satellites are imperfect and their error may be linked to cloud cover, while microwave data near coastlines may be negatively impacted by interference from land.

Brown carbon (BrC) is a subset of organic aerosols that absorbs sunlight at visible and ultraviolet (UV) wavelengths. It is estimated that BrC is responsible for 20% of the overall aerosol absorption contribution to radiative forcing, where the remaining 80% is attributed to black carbon (BC). BrC can be emitted directly by biomass burning or produced via gas-to-particle conversion of organic compounds. Estimating BrC absorption in the atmosphere is a challenging task due to the limited availability of direct measurements; this impedes a better understanding of the impacts of BrC on Earth’s radiation budget. Supervised machine learning provides a novel approach to estimate BrC absorption using other aerosol, gas, and ambient measurements. This study uses a random forest algorithm to estimate the absorption coefficient of BrC at 365 nm (bBrC) and the Absorption Angstrom Exponent (AAE) using in-situ airborne data collected in biomass burning plumes across the US during the 2019 Fire Influence on Regional to Global Environments and Air Quality (FIREX-AQ) campaign. The machine learning library SciKit Learn is used. We are able to estimate the bBrC with an accuracy of 92%, and AAE with an accuracy of 67%. The random forest algorithm shows that: (1) bBrC, an extensive property, is most influenced by chemical tracers traditionally considered proportional to the emission rate of biomass burning aerosols (namely, the concentrations of particulate magnesium and potassium); (2) AAE, an intensive property, is most influenced by measurements associated to mechanisms like photolytic aging and the secondary production of BrC (namely, the UV-visible photolytic frequency of propanal, and the concentration of isoprene hydroxy hydroperoxide). Machine learning provides a preliminary approach to consistently and accurately estimate BrC-related properties, as well as illuminate the mechanisms that affect BrC in the atmosphere, thus providing valuable insight to the research community.

Since December 2019, the COVID-19 pandemic has created a variety of societal problems around the world. The COVID-19 outbreak within the United States started in January 2020 and has already claimed approximately 98,000 lives by June 2020. In March of 2020, different states across the county issued lockdown policies as a precaution to stop the spread of COVID-19. The decrease in human mobility has resulted in changes in pollutant concentrations within densely-populated areas. This study analyzes weekly concentrations of particulate matter less than 2.5 μm (PM2.5), ozone (O3), and nitrogen dioxide (NO2) from January to June of 2020 in New York, Los Angeles, and Houston; to assess the degree to which each city’s respective stay-at-home order have caused deviations from the past. Meteorological factors like precipitation and wind direction are taken into consideration when evaluating the change in pollutant concentrations. Air quality data are obtained from the Environmental Protection Agency (EPA), the California Air Resources Board (CARB), and the Texas Commission of Environmental Quality (TCEQ). Meteorological data are obtained from the Automated Surface Observing Stations (ASOS) in each region. Results show below-average PM2.5 concentrations across New York City after stay-at-home orders were issued from March 22 to May 15th. However, in Houston, PM2.5 concentrations show an above-average trend into May 1st despite stay-at-home orders being issued on April 1st.

Marine environments emit trace gases and aerosols into the atmosphere that affect: (1) Earth’s radiative budget, (2) air quality and human health, (3) visibility, and (4) biogeochemical cycles. Sea spray aerosols affect Earth’s climate directly through their ability to scatter solar radiation, and indirectly by acting as cloud condensation nuclei. However, there is uncertainty in the extent to which they affect climate. Emissions from different types of marine bodies vary and can add to the uncertainty in climate models. Here we use airborne measurements collected in July 2019 during the NASA Student Airborne Research Program (SARP) campaign and show that measurements over the Salton Sea and the Pacific Ocean differ in the amount and composition of aerosols and trace gases. Trace gas concentrations over the Salton Sea are indicative of possible influences from agricultural runoff and urban emissions. Differences between the concentration of bromine and chlorine in the gas phase were seen between the two marine bodies, indicating differences in the chemistry. DMS emissions from the Salton Sea were greater while the Pacific Ocean emitted more BrO and BrCl. The data observed for the two locations highlighted possible variances in gas-to-particle conversions, and differences in composition and sizes of marine aerosols and condensation nuclei. Trends identified in this work will serve as a basis for further investigations into differences between aerosols and trace gases emitted from varying marine bodies.

The Salton Sea is a drying lake located in Southern California that experiences large phytoplankton blooms throughout the year due to increased nutrient loads carried in by agricultural runoff. Phytoplankton blooms produce dimethyl sulfide (DMS), which is then emitted into the atmosphere and oxidized into byproducts such as sulfur dioxide (SO2) and particulate sulfate (SO42-), negatively impacting air quality. The objective of this research project is to assess the extent to which emissions from phytoplankton blooms in the Salton Sea affect SO2 and sulfate concentrations in downwind areas. This research project combines air quality data from the Interagency Monitoring of Protected Visual Environments (IMPROVE), chlorophyll a data from Moderate Resolution Imaging Spectroradiometer (MODIS) onboard NASA’s Aqua satellite, and air trajectories from the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) software. Urban emission tracers such as carbon monoxide (CO) and nitrate (NO3–) were used to distinguish between Salton Sea emissions and urban emissions. Data were analyzed throughout 2018 and 2019 at four IMPROVE sites in the southwestern United States: Las Vegas, Lake Mead, Grand Canyon, and Phoenix. Las Vegas was the most promising location with enhanced SO2 and sulfate concentrations with HYSPLIT trajectories that indicated influence from Salton Sea. However, correlations between sulfate enhancement at Las Vegas and chlorophyll a concentration at Salton Sea were weak. Further research is needed to characterize the emissions from Salton Sea, and how chlorophyll a concentration affects these emissions.

The increasing rate of wildfires in the United States is raising concerns about health, air quality, and water quality for communities living within fire-prone regions. In the Northwestern United States, the number of weeks with ideal wildfire weather conditions are expected to increase sixfold by the mid-21st century under current climate change conditions. Wildfires release gases and aerosols into the atmosphere, which are vertically and horizontally transported, and ultimately deposited to the Earth’s surface at locations far from the original fire site via wet and dry deposition. Aerosols originating from fires are rich in species like potassium (K+), calcium (Ca2+), magnesium (Mg2+), nitrate (NO3–), and sulfate (SO42-). This research analyzes the wet and dry deposition of these species in the Northwestern United States. Wet deposition data are measured from the National Atmospheric Deposition Program’s (NADP) National Trends Network (NTN). Dry deposition data are calculated using in-situ airborne measurements of aerosol size distribution and chemical composition obtained during the 2019 Fire Influence on Regional to Global Environments and Air Quality (FIREX-AQ) campaign. Using potassium as a reliable indicator for fire influence on deposition, we are able to determine how aerosols may travel and deposit at regions far downwind of large fires. This project is relevant in determining the extent to which ecosystems and human health can be affected by the dry and wet deposition of gases and aerosols released by wildfires.

The Saharan Air Layer (SAL) is a layer of dust in the atmosphere extending across the Atlantic Ocean that occurs annually during the summer months. It is caused by the lofting of dust from the Saharan Desert, which is then transported across the Atlantic Ocean to the east coasts of North and South America. While the SAL occurs annually, measurements made by the Ozone Monitoring Instrument (OMI) onboard NASA’s Aura satellite, together with ground observations, showed that the 2020 SAL (1 June to 30 June) was both thicker and closer to the surface than typical dust events. The SAL can have a number of impacts on the transatlantic locations that it reaches including: (1) health effects due to a decrease in air quality, (2) reduced visibility, (3) transportation and deposition of nutrients and toxins, (4) impacts on the earth’s energy budget, and (5) possible effects on tropical storms. This study investigates the effect of the 2020 SAL in eight locations across the Atlantic, Caribbean, and southeastern United States using ground-based data from NASA’s Aerosol Robotic Network (AERONET). The AERONET values for aerosol optical depth (AOD), radiative forcing, radiative forcing efficiency, and single scattering albedo are used to determine the effect that the 2020 SAL had on these eight locations. We show that the presence of the SAL creates twice the cooling off the coast of Africa and up to three times more cooling in locations as far as Oklahoma. It was also found that, when normalized per aerosol amount, dust cools less than the background aerosols, but because dust is so abundant as a result of the SAL there is a net cooling effect. This research allows for a better understanding about how these annual dust storms may be a driver in additional cooling of the climate.

Particulate matter (PM) are solid particles suspended in the air, originating from both anthropogenic and natural sources. PM2.5 is the concentration of aerosol particles measuring 2.5 µm or smaller. PM2.5 can have adverse human health effects depending on composition and duration of exposure. This study focuses on how PM2.5 pollution levels have changed over time in 15 cities in the USA during the past two decades by analyzing annual and seasonal patterns. For each city, daily Air Quality Index (AQI) values for PM2.5 are obtained from the Environmental Protection Agency (EPA) from January 1, 1999 to June 30, 2020. These daily data are converted to PM2.5 mass concentrations and averaged into annual and seasonal time series. Linear regressions are applied to these time series to quantify the improvement in PM2.5 levels. Results reveal that over the past two decades, PM2.5 levels decrease for annual, winter, and summer averages in all 15 cities. The cities with the largest improvement in PM2.5 levels are Atlanta, Chicago, and Memphis, showing changes of -0.51, -0.464, and -0.44 µg/m3/year, respectively. These decreasing trends are likely due to increased awareness about environmental pollutants and tighter emission regulations. However, differences emerge when comparing seasonal trends; Chicago, Los Angeles, and Green Bay observed larger reductions in PM2.5 in the winter, whereas Atlanta, Cincinnati, and Memphis, observed larger reductions in the summer. These PM2.5 winter and summer trends varied depending on geographical location, population size, and emission trends of the city. Studying seasonal PM2.5 trends and providing inferences as to the sources and driving factors of the observed seasonal changes is imperative to understanding how human activity impacts the atmosphere.