Urbanization has become a global trend under the impact of population growth, socio-economic development, and globalization. However, the interactions between climate change and urban growth in the context of economic geography are unclear due to missing links in between the recent planning megacities. This study aims to conduct a multi-temporal change analysis of land use and land cover in New York City, City of London, and Beijing using a cellular automata-based Markov chain model collaborating with fuzzy set theory and multi-criteria evaluation to predict the city's future land use changes for 2030 and 2050 under the background of climate change. To determine future natural forcing impacts on land use in these megacities, the study highlighted the need for integrating spatiotemporal modeling analyses, such as Statistical Downscale Modeling (SDSM) driven by climate change, and geospatial intelligence techniques, such as remote sensing and geographical information system, in support of urban growth assessment. These SDSM findings along with current land use policies and socio-economic impact were included as either factors or constraints in a cellular automata-based Markov Chain model to simulate and predict land use changes in megacities for 2030 and 2050. Urban expansion is expected in these megacities given the assumption of stationarity in urban growth process, although climate change impacts the land use changes and management. More land use protection should be addressed in order to alleviate the impact of climate change.

This thesis sheds light on coupling potential flood risk and drainage infrastructure resilience of low-lying areas of a coastal urban watershed to flood hazards and subsequent multi-scale impacts of those hazards via detailed modeling frameworks. Physically based models along with statistical models are employed to highlight the complexity for characterizing flood risk while evaluating such risk under various levels of adaptive capacity from traditional flood management techniques to low impact development (LID), as a first step to conduct resilience assessment. Findings indicate that the coupling flood risk and infrastructure resilience is achievable by the careful formulation of flood risk associated with a resilience metric, which is a function of the hazard(s) considered, vulnerability and adaptive capacity. The results also give insights into improving existing methodologies for municipalities in flood management practices such as incorporating multi-criteria flood risk evaluation that includes resilience.

The United States Environmental Protection Agency (USEPA) regulates public water systems and has established limits for certain disinfection by products (DBPs) that have been linked to health effects, such as bladder cancer. The regulation of DBPs, specifically total trihalomethanes (TTHMs) and haloacetic acids (HAAs), have encouraged water treatment professionals to assess the type and amount of organic precursors in their supplies. Three of the more common water quality parameters that are monitored as DBP surrogates include dissolved organic carbon (DOC), ultraviolet absorbance (UV254), and specific ultraviolet absorbance (SUVA). Although DOC, UV254, and SUVA have been effectively correlated to DBP formation, efforts to correlate fluorescence excitation emission matrices (FEEM) to DBP formation remains limited within the drinking water community. In this research, a fluorescence regional integration (FRI) approach was used to compare FEEM with DOC, UV254, and SUVA as an alternative surrogate for characterizing TTHMs for groundwater sources located in south central Florida. To quantitatively evaluate FEEM, DOC, UV254, and SUVA as TTHM precursor surrogate parameters, a statistical correlation analysis was employed. Thirteen groundwater samples were collected from various Central Florida groundwater wells in Lake County, Polk County, and Palm Beach County, and analyzed for FEEM, DOC, UV254, and SUVA prior to determining the four-day TTHM concentration using a simulated distribution system dosing procedure. The FRI method was then used to quantify FEEM by dividing the three-dimensional matrix into five distinct regions, each representing a unique organic constituent. The volume under each region was determined and used for the correlation analysis. It was determined that a combinations of regions III and V of the FEEM possessed a strong linear correlation to four day TTHM content (R2 = 0.95) as compared to DOC (R2 = 0.906), UV254 (R2= 0.84), SUVA (R2 = 0.640), and the individual regions of the FEEM. However, DOC showed the strongest correlation when a second order polynomial regression was used (R2 = 0.937). Results for the individual regions of the FEEM revealed four day simulated TTHM correlation coefficients of 0.25, 0.62, 0.86, 0.74, and 0.88 for regions I through V respectively. These values indicated that a combination of regions III and V, which represent the fulvic and humic-like organic fractions detected by FEEM respectively, was the most accurate four day simulated TTHM precursor surrogate parameter based on the groundwater supplies tested. These results reveal that although DOC is still one of the strongest surrogate parameters to TTHM formation, fluorescence has also shown to also be a potentially strong surrogate for groundwaters. The implications of these results signify that fluorescence monitoring could be a viable method of measuring organic content in groundwaters once the technology further develops.

Public water systems add disinfectants in water treatment to inactivate microbial pathogens. Chlorine, when used as a disinfectant, reacts with natural organic matter in the water to form trihalomethane (THM) and haloacetic acid (HAA5) disinfection by-products (DBPs), which are suspected carcinogens. The Safe Drinking Water Act's Disinfectant and Disinfection By-Product (D/DBP) Rules were promulgated by the U.S. Environmental Protection Agency to regulate the amount of DBPs in water systems. Regulatory compliance is based on maximum contaminant levels (MCL), measured as a locational running annual average (LRAA), for total THM (TTHM) and HAA5 of 80 µg/L and 60 µg/L, respectively. Regulated DBPs, if consumed in excess of EPA's MCL standard over many years, may increase chronic health risks. In order to comply with the D/DBP Rules, the County of Maui Department of Water Supply (DWS) adopted two DBP control technologies. A GridBee® spray-aeration process was place into DWS's Lower Kula water system's Brooks ground storage tank in February of 2013. In March of 2015 the second DBP control technology, granular activated carbon (GAC), was integrated into DWS's Pi'iholo surface water treatment plant. To investigate the integration effectiveness of GAC and spray-aeration into a water system for DBP control, DBP data was gathered from the system between August of 2011 and August 2016, and analyzed relative to cost and performance. Prior to the spray aeration and GAC integration, it was found that TTHM levels at the LRAA compliance site ranged between 58.5 µg/L and 125 µg/L (at times exceeding the MCL). Additionally, HAA5 levels at the LRAA compliance site ranged between 21.2 and 52.0 µg/L. The concerted efforts of the GAC and GridBee® system was found to reduce LRAA TTHM and HAA5 concentrations to 38.5 µg/L and 20.5 µg/L, respectively, in the Lower Kula system. Hypothesis testing utilizing t-Tests confirmed that TTHMs levels were controlled by the spray aeration system and the GAC was responsible for controlling HAA5 formation. Although TTHM levels were reduced by 58 percent, and HAA5 levels by 48 percent, the estimated cumulative annual operation and maintenance (O&M) cost of the two systems was $1,036,000. In light of the cost analysis, total organic carbon (TOC)-based models for predicting LRAA TTHM and HAA5 levels were developed as equation (i) and (ii), respectively: (i) TTHM µg/L = (32.5 x (TOC ppm)) + 5.59, (ii) HAA5 µg/L = (8.37 x (TOC ppm)) + 12.4. The TTHM model yielded an R2 of 0.93, and the HAA5 model had an R2 of 0.52. F-Tests comparing predicted LRAA TTHM and HAA5 levels to actual LRAA TTHM and HAA5 levels determined no statistically-significant difference. With the knowledge of how the GAC and spray aerator controlled DBPs in the water system, a cost-effective and practical treatment operating parameter was developed. The parameter, Pi'iholo water plant filter effluent TOC content, can serve as an indicator that operators would use to alter DBP treatment process flow set points to achieve cost-effective treatment. Furthermore, the significant annual cost contribution by the GAC, coupled with HAA5 levels below DWS's MCLG, led to the recommendation of variable frequency drive (VFD) pumps for the GAC system. The addition of VFD pumps should reduce the frequency of carbon change outs while preserving adequate HAA5 control in the system.

Nanofiltration (NF) is a pressure driven membrane process employed in drinking water treatment that requires pretreatment for reliable operation. The objective of this research was to determine if NF membranes can proficiently operate with a decreased or eliminated dose of sulfuric acid pretreatment. When used as pretreatment, sulfuric acid prevents calcium carbonate scaling on NF membranes, yet is costly, hazardous, and imparts high sulfate concentrations to NF feed and concentrate streams. To conduct this research, a 0.324 million gallon per day (MGD) NF pilot plant was operated for 3,855 run-hours at a flux rate of 15 gallons per square foot-day. The NF pilot unit's process performance, productivity, and water quality were monitored while the sulfuric acid dose was gradually decreased, controlled by monitoring pH that ranged from pH 6.5 (80 mg/L sulfuric acid dose) to pH 7.0 (no sulfuric acid dose). NF pilot productivity, as measured by specific flux, was found to decline when sulfuric acid was eliminated by 2.33 percent, 9.61 percent, and 4.08 percent in the first stage, second stage, and total pilot system, respectively, with no distinguishable increase in pressure drop. Noticeable water quality trends include approximately 75 percent sulfate decrease in feed and concentrate streams, and 20 percent increase of calcium hardness and alkalinity in the permeate stream. After piloting, superimposed elemental imaging analysis revealed that the second stage, tail-end membrane surface was fouled with iron disulfide, calcium carbonate, clay, and natural organic matter. However, flux recovered to normal operating conditions after a membrane cleaning was performed. Results of the pilot study indicated that sulfuric acid could be eliminated from the full-scale NF pretreatment process; however, membrane cleaning frequencies could increase. If applied to the full-scale NF process, elimination of sulfuric acid pretreatment would reduce annual chemical costs by over $70,000.

Ultrafiltration (UF) is a low-pressure membrane process that yields higher permeate flux and saves significant operating costs compared to NF/RO. However, UF has not been applied as a primary method for landfill leachate treatment due to its large pore size. This research investigated the removal of heavy metals from landfill leachate using an UF membrane laminated with fiber mats produced from electrospinning of a polyelectrolyte complex. In this research, we modified the surface of the UF membrane with two polyelectrolytes including Polyacrylic Acid (PAA) and Polyallylamine Hydrochloride (PAH). The removal of heavy metals including Pb, Cd, and Cu from water using electrospun (ES) polyelectrolyte (PE) complex fibers of polyacrylic acid (PAA) and polyallylamine hydrochloride (PAH) was investigated. PAA/PAH fiber mats were fabricated under various electrospinning operating conditions to optimize fiber size and stability. The fiber mats exhibited approximately 63%, 42%, and 21% removals of Pb, Cd, and Cu, respectively in synthetic metal solutions at pH 3.4. Furthermore, approximately 70%, 98%, and 92% removals of Pb, Cd, and Cu, respectively were observed at a higher pH (7.4). Moreover, the removal of heavy metals from various synthetic feed solutions and landfill leachate by the PAA/PAH laminated UF membranes (PAA/PAH-UF) was studied. The PAA/PAH-UF membrane exhibited approximately 38%, 49%, and 85% higher removal of Pb, Cu, and Cd, respectively from laboratory-prepared metal ion solution (DI water) when compared to the unmodified UF membrane (UF). The PAA/PAH UF membrane exhibited approximately 18% and 15% higher removal of Pb and Cu, respectively in the leachate when compared to DI water. The PAA/PAH-UF membrane showed around 16% and 72% higher removal of Pb and Cd at the presence of NOM. Moreover, the UF membrane showed approximately 18%, 25%, and 30% more removal of Pb, Cd, and Cu at the presence of NOM, respectively.

This study investigated water reclamation from an impaired-quality water using forward osmosis (FO) membrane functionalized by nano zero valent iron (nZVI) loaded polyelectrolyte multilayer films. Stormwater runoff was selected as the impaired-quality water, which served as a feed solution (FS) and a NaCl salt solution at a concentration representing reverse osmosis (RO) concentrate was used a draw solution (DS). RO concentrate is another impaired-quality water that is discharged to the environmental with little or no treatment. A commercial cellulose triacetate (CTA) FO membrane was modified using poly allylamine hydrochloride (PAH) (a polycation) and poly acrylic acid (PAA) (a polyanion) following a dip coating method. In-situ synthesis of nZVI within the PAA/PAH layers was conducted through sodium borohydride reduction. The efficiency of the nZVI-PAA/PAH functionalized FO membranes in removing selected stormwater derived contaminants was evaluated. The virgin CTA membrane had a rough surface and repeated applications of coatings were required to ensure uniform layers of PAA/PAH. Following 4, 8, 12, and 14 'bilayer' (BL) coatings, it was deduced that 14 BL coatings resulted in the most uniform layers with maximum surface overage. The scanning electron microscopy (SEM) images demonstrated that the membrane coverage and uniformity of coating improved as more BL coatings were applied. While the unmodified membrane initially had a high flux (16 L/m2.h) compared to the modified membranes, there was a sharp decline in flux (approximately 1.5 L/m2.h) within the first 2 to 3 h of operation, likely due to a rapid accumulation of foulants on the membrane surface. The modified membranes, on the other hand, showed markedly less initial flux (2 L/m2.h), but the flux was maintained throughout the experimental period with only a slight decline. The flux decline that may be anticipated owing to the additional layers on the membrane was probably offset by the hydrophilicity rendered by the PAA/PAH functional groups and less foulant accumulation. The nZVI-loaded PAA/PAH coatings did not have any adverse impact on reverse salt flux (RSF). Any reduction in RSF due to the coatings might have been counteracted by the PAA/PAH induced dilution of the DS. While both NO3- and PO43- removals were already very high ( > 97%) when using the unmodified membrane, a slight increase in the removal of NO3- with increased BL numbers (8 and 14 BLs) was observed. The removal of selected heavy metals (Cd, Pb, and Cu) by the unmodified and modified membranes ranged from approximately 87% for Pb to almost complete (~99%) removal for Cd. Mass balance and energy dispersive x-ray spectroscopy (EDX) analyses confirmed that a higher number of coating resulted in a higher retention of heavy metals by the functionalized FO membranes. This could be attributed to the complexation of metal ions with carboxylate and amine groups from the PAA/PAH bilayers. Furthermore, the unmodified CTA FO membrane exhibited very high removal ( > 97%) of perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS). EDX mapping showed that the 14 BL-nZVI membrane adsorbed more PFOS/A compared to the virgin and the 8 BL-nZVI membrane.

Human exposure estimation to air pollution plays an important role in epidemiological studies which are designed to reveal correlations between human exposures to certain air pollutants and certain diseases, such as asthma, cardiovascular disease and reproductive diseases. Traditionally, when people's mobile data is hard to get, home location is used to estimate people's exposures assuming that people stay at home all the time. Whereas, people move and it is more accurate to estimate people's exposures including people's mobility. In our study, we showcased two methods to obtain people's mobile data: Google Maps location history (GMLH) data and Call Detailed Record (CDR) data. GMLH data was compared with Global Positioning System (GPS) data from four aspects: 1) spatial movement of the subject; 2) time the subject spent at different microenvironments; 3) time the subject spent on driving; 4) subject's time-weighted exposures to ambient particulate matter. The results showed that compared with GPS data, GMLH data capture well the subject's spatial mobility with resolution of 200m * 200m or larger and successfully captured the time the subject spent at different microenvironments and the time on driving. Also, with GMLH data we were able to accurately estimate the subject's time-weighted exposure to ambient PM pollution. CDR data was used to estimate subjects' mobile exposures for five chosen pollutants (CO, NO2, SO2, O3, and PM2.5). And the correlation between difference between static exposures and mobile exposures with mobility level is also investigated. My study revealed that there is no substantial difference between home based exposure (HBE) and CDR based exposure (CDRE) at population level. But at individual level, difference between HBE and CDRE increased with mobility increased. It was also found that HBE would likely under-estimate exposure to traffic-related pollutants (CO, NO2 and PM2.5) during afternoon rush-hour, but over-estimate exposures to ozone during mid-afternoon. As smartphone and Google Maps application are used widely, these two methods have huge potential on obtaining people's mobility data. My study also tested the relative accuracy and reliability of two brand commercial sensors (PurpleAir and Dylos). Results showed that PurpleAir has good relative accuracy and reliability, while Dylos has moderate relative accuracy and reliability.

In recent years, concentrations of nutrients such as nitrogen and phosphorus have increased in surface and groundwater resources, due in part to non-point source pollution associated with stormwater runoff. The elevated nutrient concentrations found in stormwater runoff have prompted the design of best management practices (BMP's) to mitigate the problem. The overall objective within this thesis is to analyze the performance of innovative surface BMPs and investigate connections between the BMPs and groundwater flows to freshwater springs within a karst environment. The performance of two stormwater BMPs, blanket filters and vertical reactors containing Bio-sorption Activated Media (BAM), are assessed in terms of hydraulic retention time. Capture efficiency is also evaluated for the blanket filters. Blanket filters captured, at minimum 68% of the stormwater runoff entering a stormwater basin in one year. Water content monitoring indicates that BAM is affected by the surrounding water table. The vertical reactors are more appropriate technologies for small contributing areas. Tracking a conservative tracer from an injection point within a stormwater basin to nearby Silver Springs reveals several unique flowpaths and velocities of groundwater. Subsurface velocities observed in the basin ranged from 0.1 m/d to 1.4 m/d, while velocities from the injection well to the spring vary from 2.3 m/d to 13.5 m/d. The fastest travel times observed in the spring may represent flowpaths that include macropore/conduit flow through karst features, while the slower peaks may be more representative of matrix flow. Interaction with karst features may reduce retention time of stormwater in aquifers, altering expected nutrient transformations. Understanding the variable pathways stormwater may take from the surface to spring discharge may assist environmental managers in preserving water quality in springs and other waterbodies in karst systems.

In late August 2016, a sinkhole spanning 45 feet (13.7 meters) in diameter opened at a phosphate fertilizer facility (Mosaic Company) near Mulberry, Florida, leaking an estimated 215 million gallons (813,000 cubic meters) of radionuclide-contaminated water 300 feet into the Floridan aquifer. An investigation to determine possible impacts to the environment and local community drinking water supplies was implemented that focused on two 1.5 million gallon per day (MGD) Tampa Bay Water (TBW) production wells and two Polk County Utilities (PCU) water treatment facilities. Water samples collected between June 2017 and January 2018 at the TBW and PCU sites were found to contain radionuclides below regulated levels. To evaluate the effectiveness of membrane treatment should the TBW and PCU drinking water wells be affected by the spill in the future, bench-scale, flat-sheet reverse osmosis (RO) and nanofiltration (NF) membrane process testing was performed using TBW and PCU wellfield sample aliquots. NF and RO were shown to be capable of removing at minimum of 86 and 92 percent, respectively, of the barium content that had been spiked into groundwater testing aliquots. Based on testing results, a conceptual opinion of probable capital cost for a membrane process ranged from $1.7 and $3.5 million for a 0.25 MGD and 2.0 MGD design capacity, respectively. Process operation and maintenance costs ranged between $0.99/Kgal and $0.26/Kgal for a 0.25 MGD and 2.0 MGD design capacity, respectively. The amortized total cost based on a 20-year period and 8 percent interest rate ranged between $1.88/Kgal for a 0.25 MGD and $0.49/Kgal for a 2.0 MGD design capacity plant. An estimate of unavailable water value due to a long-term well shut-down was approximated as $0.64/Kgal.