Browsing by Author "Bergin, Michael H"

Exposure to higher PM2.5 can lead to increased risks of mortality; however, the spatial concentrations of PM2.5 are not well characterized, even in megacities, due to the sparseness of regulatory air quality monitoring (AQM) stations. This motivates novel low-cost methods to estimate ground-level PM2.5 at a fine spatial resolution so that PM2.5 exposure in epidemiological research can be better quantified and local PM2.5 hotspots at a community-level can be automatically identified. Wireless low-cost particulate matter sensor network (WLPMSN) is among these novel low-cost methods that transform air quality monitoring by providing PM information at finer spatial and temporal resolutions; however, large-scale WLPMSN calibration and maintenance remain a challenge because the manual labor involved in initial calibration by collocation and routine recalibration is intensive, the transferability of the calibration models determined from initial collocation to new deployment sites is questionable, as calibration factors typically vary with urban heterogeneity of operating conditions and aerosol optical properties, and the stability of low-cost sensors can drift or degrade over time. This work presents a simultaneous Gaussian Process regression (GPR) and simple linear regression pipeline to calibrate and monitor dense WLPMSNs on the fly by leveraging all available reference monitors across an area without resorting to pre-deployment collocation calibration. We evaluated our method for Delhi, where the PM2.5 measurements of all 22 regulatory reference and 10 low-cost nodes were available for 59 days from January 1, 2018 to March 31, 2018 (PM2.5 averaged 138 ± 31 μg m-3 among 22 reference stations), using a leave-one-out cross-validation (CV) over the 22 reference nodes. We showed that our approach can achieve an overall 30 % prediction error (RMSE: 33 μg m-3) at a 24 h scale and is robust as underscored by the small variability in the GPR model parameters and in the model-produced calibration factors for the low-cost nodes among the 22-fold CV. Of the 22 reference stations, high-quality predictions were observed for those stations whose PM2.5 means were close to the Delhi-wide mean (i.e., 138 ± 31 μg m-3) and relatively poor predictions for those nodes whose means differed substantially from the Delhi-wide mean (particularly on the lower end). We also observed washed-out local variability in PM2.5 across the 10 low-cost sites after calibration using our approach, which stands in marked contrast to the true wide variability across the reference sites. These observations revealed that our proposed technique (and more generally the geostatistical technique) requires high spatial homogeneity in the pollutant concentrations to be fully effective. We further demonstrated that our algorithm performance is insensitive to training window size as the mean prediction error rate and the standard error of the mean (SEM) for the 22 reference stations remained consistent at ~30 % and ~3–4 % when an increment of 2 days’ data were included in the model training. The markedly low requirement of our algorithm for training data enables the models to always be nearly most updated in the field, thus realizing the algorithm’s full potential for dynamically surveilling large-scale WLPMSNs by detecting malfunctioning low-cost nodes and tracking the drift with little latency. Our algorithm presented similarly stable 26–34 % mean prediction errors and ~3–7 % SEMs over the sampling period when pre-trained on the current week’s data and predicting 1 week ahead, therefore suitable for online calibration. Simulations conducted using our algorithm suggest that in addition to dynamic calibration, the algorithm can also be adapted for automated monitoring of large-scale WLPMSNs. In these simulations, the algorithm was able to differentiate malfunctioning low-cost nodes (due to either hardware failure or under heavy influence of local sources) within a network by identifying aberrant model-generated calibration factors (i.e., slopes close to zero and intercepts close to the Delhi-wide mean of true PM2.5). The algorithm was also able to track the drift of low-cost nodes accurately within 4 % error for all the simulation scenarios. The simulation results showed that ~20 reference stations are optimum for our solution in Delhi and confirmed that low-cost nodes can extend the spatial precision of a network by decreasing the extent of pure interpolation among only reference stations. Our solution has substantial implications in reducing the amount of manual labor for the calibration and surveillance of extensive WLPMSNs, improving the spatial comprehensiveness of PM evaluation, and enhancing the accuracy of WLPMSNs. Satellite-based ground-level PM2.5 modeling is another such low-cost method. Satellite-retrieved aerosol products are in particular widely used to estimate the spatial distribution of ground-level PM2.5. However, these aerosol products can be subject to large uncertainties due to many approximations and assumptions made in multiple stages of their retrieval algorithms. Therefore, estimating ground-level PM2.5 directly from satellites (e.g., satellite images) by skipping the intermediate step of aerosol retrieval can potentially yield lower errors because it avoids retrieval error propagating into PM2.5 estimation and is desirable compared to current ground-level PM2.5 retrieval methods. Additionally, the spatial resolutions of estimated PM2.5 are usually constrained by those of the aerosol products and are currently largely at a comparatively coarse 1 km or greater resolution. Such coarse spatial resolutions are unable to support scientific studies that thrive on highly spatially-resolved PM2.5. These limitations have motivated us to devise a computer vision algorithm for estimating ground-level PM2.5 at a high spatiotemporal resolution by directly processing the global-coverage, daily, near real-time updated, 3 m/pixel resolution, three-band micro-satellite imagery of spatial coverages significantly smaller than 1 × 1 km (e.g., 200 × 200 m) available from Planet Labs. In this study, we employ a deep convolutional neural network (CNN) to process the imagery by extracting image features that characterize the day-to-day dynamic changes in the built environment and more importantly the image colors related to aerosol loading, and a random forest (RF) regressor to estimate PM2.5 based on the extracted image features along with meteorological conditions. We conducted the experiment on 35 AQM stations in Beijing over a period of ~3 years from 2017 to 2019. We trained our CNN-RF model on 10,400 available daily images of the AQM stations labeled with the corresponding ground-truth PM2.5 and evaluated the model performance on 2622 holdout images. Our model estimates ground-level PM2.5 accurately at a 200 m spatial resolution with a mean absolute error (MAE) as low as 10.1 μg m-3 (equivalent to 23.7% error) and Pearson and Spearman r scores up to 0.91 and 0.90, respectively. Our trained CNN from Beijing is then applied to Shanghai, a similar urban area. By quickly retraining only RF but not CNN on the new Shanghai imagery dataset, our model estimates Shanghai 10 AQM stations’ PM2.5 accurately with a MAE and both Pearson and Spearman r scores of 7.7 μg m-3 (18.6% error) and 0.85, respectively. The finest 200 m spatial resolution of ground-level PM2.5 estimates from our model in this study is higher than the vast majority of existing state-of-the-art satellite-based PM2.5 retrieval methods. And our 200 m model’s estimation performance is also at the high end of these state-of-the-art methods. Our results highlight the potential of augmenting existing spatial predictors of PM2.5 with high-resolution satellite imagery to enhance the spatial resolution of PM2.5 estimates for a wide range of applications, including pollutant emission hotspot determination, PM2.5 exposure assessment, and fusion of satellite remote sensing and low-cost air quality sensor network information. We later, however, found out that this CNN-RF sequential model, despite effectively capturing spatial variations, yields higher average PM2.5 prediction errors than its RF part alone using only meteorological conditions, most likely the result of CNN-RF sequential model being unable to fully use the information in satellite images in the presence of meteorological conditions. To break this bottleneck in PM2.5 prediction performance, we reformulated the previous CNN-RF sequential model into a RF-CNN joint model that adopts a residual learning ideology that forces the CNN part to most effectively exploit the information in satellite images that is only “orthogonal” to meteorology. The RF-CNN joint model achieved low normalized root mean square error for PM2.5 of within ~31% and normalized mean absolute error of within ~19% on the holdout samples in both Delhi and Beijing, better than the performances of both the CNN-RF sequential model and the RF part alone using only meteorological conditions. To date, few studies have used their simulated ambient PM2.5 to detect hotspots. Furthermore, even the hotspots studied in these very limited works are all “global” hotspots that have the absolute highest PM2.5 levels in the whole study region. Little is known about “local” hotspots that have the highest PM2.5 only relative to their neighbors at fine-scale community levels, even though the disparities in outdoor PM2.5 exposures and their associated risks of mortality between populations in local hotspots and coolspots within the same communities can be rather large. These limitations motivated us to concatenate a local contrast normalization (LCN) algorithm at the end of the RF-CNN joint model to automatically reveal local PM2.5 hotspots from the estimated PM2.5 maps. The RF-CNN-LCN pipeline reasonably predicts urban PM2.5 local hotspots and coolspots by capturing both the main intra-urban spatial trends in PM2.5 and the local variations in PM2.5 with urban landscape, with local hotspots relating to compact urban spatial structures while coolspots being open areas and green spaces. Based on 20 sampled representative neighborhoods in Delhi, our pipeline revealed that on average a significant 9.2 ± 4.0 μg m-3 long-term PM2.5 exposure difference existed between the local hotspots and coolspots within the same community, with Indian Gandhi International Airport area having the steepest increase of 20.3 μg m-3 from the coolest spot (the residential area immediately outside the airport) to the hottest spot (airport runway). This work provides a possible means of automatically identifying local PM2.5 hotspots at 300 m in heavily polluted megacities. It highlights the potential existence of substantial health inequalities in long-term outdoor PM2.5 exposures within even the same local neighborhoods between local hotspots and coolspots. Apart from posing serious health risks, deposition of dust and anthropogenic particulate matter (PM) on solar photovoltaics (PVs), known as soiling, can diminish solar energy production appreciably. As of 2018, the global cumulative PV capacity crossed 500 GW, of which at least 3–4% was estimated to be lost due to soiling, equivalent to ~4–6 billion USD revenue losses. In the context of a projected ~16-fold increase of global solar capacity to 8.5 TW by 2050, soiling will play an increasingly more important part in estimating and forecasting the performance and economics of solar PV installations. However, reliable soiling information is currently lacking because the existing soiling monitoring systems are expensive. This work presents a low-cost remote sensing algorithm that estimates utility-scale solar farms’ daily solar energy loss due to PV soiling by directly processing the daily (near real-time updated), 3 m/pixel resolution, and global coverage micro-satellite surface reflectance (SR) analytic product from the commercial satellite company Planet. We demonstrate that our approach can estimate daily soiling loss for a solar farm in Pune, India over three years that on average caused ~5.4% reduction in solar energy production. We further estimated that around 437 MWh solar energy was lost in total over the 3 years, equivalent to ~11799 USD, at this solar farm. Our approach’s average soiling estimation matches perfectly with the ~5.3% soiling loss reported by a previous published model for this solar farm site. Compared to other state-of-the-art PV soiling modeling approaches, the proposed unsupervised approach has the benefit of estimating PV soiling at a precisely solar farm level (as in contrast to coarse regional modeling for only large spatial grids in which a solar farm resides) and at an unprecedently high temporal resolution (i.e., 1 day) without resorting to solar farms’ proprietary solar energy generation data or knowledge about the specific components of deposited PM or these species’ dry deposition flux and other physical properties. Our approach allows solar farm owners to keep close track of the intensity of soiling at their sites and perform panel cleaning operations more strategically rather than based on a fixed schedule.

In highly polluted cities, many residents seek to reduce their personal exposure to air pollutants. However, over the course of a day, people spend time in a variety of micro-environments (e.g., different rooms in their home, at school or work, in vehicles, outdoors), making it challenging to prioritize actions to reduce exposure. In China, air purifiers are a common mitigation strategy. The impacts of air purifiers on indoor air quality and personal exposure can differ greatly in real-world settings depending on the participants’ behavior, the home environment, and the selected purifier. This work uses low-cost monitors to compare the impacts of air filtration on indoor air and personal exposure. Before deploying low-cost monitors, steps needed to be taken to ensure data quality: low-cost air monitors were evaluated against reference measurements in multiple field locations (i.e. Atlanta, Hyderabad, Beijing, and Shanghai), and various calibration methods were developed to reduce monitor error and bias. Results highlight the need to calibrate low-cost monitors under the same conditions in which they will be deployed, and illustrate methods to reduce error that will allow low-cost monitors to make powerful insights into addressing additional air quality issues in the future. After evaluation and calibration, low-cost air monitors were installed indoors (in study participants’ bedrooms) and outdoors. Additionally, monitors were worn by study participants to measure personal exposure. To evaluate the impacts of bedroom air filtration, participants in Beijing (N=7 adult participants) and Shanghai (N=43 child participants) had filtration devices set up in their bedrooms. During the study, all participants experienced both true and sham (i.e., placebo, with the filters removed) filtration. Personal exposure to O3 was significantly lower than ambient levels due to low indoor concentrations. However, few conclusions can be drawn about the impacts of filtration on O3 and the micro-environmental exposure to O3 since indoor concentrations were so often below the detection limit of our monitors (6 ppb). Measurements suggested that indoor PM2.5 was primarily of outdoor origin (≥75%). Purifiers significantly reduced bedroom PM2.5 concentrations by 70% and 78% during true filtration in Shanghai and Beijing, respectively. The reduction to personal exposure was less prominent, but still significant (Shanghai=24% and Beijing=36%). The largest time-weighted exposure to PM2.5 occurred in the bedroom during sham filtration since children spent the most time there. This suggests that the bedroom is the most important environment to tackle to reduce personal exposure. True filtration reduced the contribution of the bedroom micro-environment below that of the classroom and other rooms at home, highlighting that the classroom and other home environments should be the next-microenvironments to be address to reduce these children’s exposure. These results highlight the importance of reducing outdoor PM2.5 levels, but show that Chinese residents can reduce their PM2.5 exposure by installing an effective air purifier in the indoor environment where they spend the most time. Reductions in multiple micro-environments are likely required to further reduce exposure to PM2.5.

Low-cost particulate matter (PM) sensors are promising tools for supplementing existing air quality monitoring networks. However, the performance of the new generation of low-cost PM sensors under field conditions is generally not well understood. In this study, we characterized the performance capabilities of a relatively new low-cost PM sensor model (Plantower PMS3003) of measuring PM2.5 at 1 min, 1 h, 6 h, 12 h and 24 h integration times. We tested the PMS3003s in both low concentration suburban regions (Durham and Research Triangle Park (RTP), NC, US) with 1 h PM2.5 averaging 9 g m-3 (range: 0-62 μg m-3) and 10 g m-3 (range: 3-20 μg m-3) respectively and a high concentration urban location (Kanpur, India) with 1 h PM2.5 averaging 36 g m-3 (range: 0-127 μg m-3) and 116 g m-3 (range: 19-347 μg m-3) during monsoon and post-monsoon seasons, respectively. In Durham and Kanpur, the sensors were compared to a research-grade instrument (environmental -attenuation monitor (E-BAM)) to determine how these sensors perform across a range of PM2.5 concentrations and meteorological factors (e.g., temperature and relative humidity (RH)). In RTP, the sensors were compared to three Federal Equivalent Methods (FEMs) including two T640s and a SHARP to demonstrate the importance of the type of reference monitor selected for sensor calibration. The decrease of 1 h mean errors of the calibrated sensors using univariate linear models from Durham (207%) to Kanpur monsoon (46%) and to post-monsoon (35%) season showed PMS3003 performance generally improved as ambient PM2.5 increased. The precision of reference instruments is critical in evaluating sensor performance and -attenuation-based monitors are unfavorable for testing PM sensors at low concentrations as were underscored by 1) the gentler gradient at which the mean errors reduced over averaging time in RTP (from 27% for 1 h to 9% for 24 h) than in Durham (from 201% to 15%); 2) the lower errors in RTP than Kanpur post-monsoon season (from 35% to 11%); 3) the higher precision of the T640 (1 h average: ±0.5 g m-3) than the SHARP (24 h average: ±2 g m-3, better than the E-BAM). A major RH influence was found in RTP (1 h mean RH = 64%, spanning 27-93%) that can explain up to ~30% of the variance in 1 min to 6 h PMS3003 PM2.5 measurements. The RH corrections can reduce errors from ~22-27% to ~10%. We observed that PMS3003s appeared to exhibit a non-linear response when ambient PM2.5 levels exceeded ~125 g m-3 and found that the quadratic and two-segment piecewise linear fits might be more appropriate than the univariate linear model to capture this non-linearity and can further reduce mean errors by up to 11% and 8%, respectively. Overall, our results have important implications for how variability in ambient PM2.5 concentrations, reference monitor types, and meteorological factors can affect PMS3003 performance. These results can provide valuable guidance on future establishment of dense PM sensor networks with known and superior accuracy and precision.

It is well known that particulate matter (PM) has strong associations with various negative health endpoints. However, the precise mechanisms linking PM to these negative impacts are complex and not fully understood. The U.S. Environmental Protection Agency currently regulates PM on a mass concentration basis (μg of PM per m3 of air), which does not account for the differential toxicity of different particle species. More research is needed to improve understanding on how toxicity changes with different PM sources, and to answer: which environments have PM compositions that are particularly dangerous? The primary objective of this work is to characterize understudied aspects of particulate matter generated in environments that are relevant to human exposure (i.e., environments where people spend a large portion of their time). The exposure-relevant sites examined in this work investigate PM2.5 (particulate matter with diameters <2.5 microns) collected from inside cars during daily commutes in Atlanta, from urban India where roadside and residential trash burning is ubiquitously practiced, and from residential sites in rural and urban Guatemala. As mentioned, though the associations between negative health impacts and PM concentrations are striking, the toxicity pathways are not well understood. One proposed pathway of acute toxicity is related to an inhaled particle’s ability to generate reactive oxygen species (ROS) and exert oxidative stress on the lungs. In recent years, various assays have been developed to assess the ROS-generating capacity of particulate matter. Two of the most established assays used in air pollution research are the DTT (dithiotreitol) assay and the lung macrophage assay. These assays were used to make the first-ever measurements of oxidative potential of PM2.5 collected from in-vehicle commutes (in Atlanta) and from in-situ trash burning events (in Bangalore, India). In-vehicle results from ~2-hour morning commutes (n = 50) indicate that on-road DTT activity (median [IQR] = 0.68 [0.75] nmol min-1 m-3) is ~2 times higher than DTT activity measured from 23-hour roadside samples. Results highlight how gas-phase compounds make important contributions to DTT activity and that short-term exposures are associated with distinct changes in oxidative potential. This was echoed by results from trash-burning samples (n = 24), which suggest that ~1 minute of direct exposure to emissions from trash burning was equivalent in DTT activity to breathing in an entire day of ambient air in Bangalore. Though ambient samples (n = 6) show notable DTT activity (median [IQR] = 0.76 [0.03] nmol min-1 m-3), trash burning DTT activity was extremely high, averaging >1,000 nmol min-1 m-3. However, when considering DTT and macrophage results on a per-mass basis, ambient PM2.5 appears to be ~2 – 100 times more redox active than fresh trash-burning emissions, suggesting that many compounds found in fresh trash-burning emissions are not redox active; this may also indicate how atmospheric processing and aging can result in increased PM redox activity. Results highlight the importance of assessing PM with additional toxicity pathways since ROS activity alone is not sufficient to describe the many ways in which PM may impact health. Overall, results indicate that near trash-burning sources, exposure to redox-active PM can be extremely high. A follow-up project was launched in response to observing widely varying emissions from trash burning, which results from Bangalore show were vastly different even when comparing trash piles of similar size, composition, and burning conditions. This follow-up project was educational in nature as it was a collaborative effort between students at Duke University and the India Institute of Technology. To generate emissions in a more comparable way, we controlled for pile size, composition, and environmental variables (e.g., wind speed) that may affect burning conditions, and then we iteratively burned compiled mixtures of trash in a small-scale combustor. Burn piles (n = 28) were compiled to represent trash compositions observed and collected from six sites in Ahmedabad, India, where average pile composition was observed to be ~60% plastic by volume; plastic-only piles were also burned in the combustor. Plastic bottles were observed to generate the highest concentrations of PM2.5 and black carbon emissions, while plastic films emitted very low pollutant concentrations with PM2.5 close to background levels. Using low-cost sensors and thermocouples attached to the incinerator body proved to be an affordable way to make semi-quantitative assessments of controlled burns. We also demonstrate how low-cost sensors attached to a commercial UAV (unmanned aerial vehicle) could be useful for safely collecting pollutant data over a smoky municipal dumpsite. Trash burning is clearly a source of highly variable and spatially sporadic emissions. This follow-up project is valuable as it makes small but important steps toward finding affordable ways to measure and mitigate emissions. Lastly, a final project was pursued to assess air pollution and microbial concentrations from residential sites in Guatemala, including in a community where PM2.5 levels indoors were observed to be exceedingly high due to traditional cookstove use. Though many studies have measured PM in similar settings, existing research has not investigated microbial concentrations in the air at these settings (as the majority of existing research has focused on sites in high-income countries). Airborne viruses and bacteria were enumerated from filter samples using a staining microscopy technique. Air samples (n = 40) were collected at different times of day indoors and outdoors to provide insight on whether household or ambient sources dominate bioaerosol contributions. Results suggest that bioaerosols from indoor sources dominate in the mornings, while outdoor sources contribute more to bioaerosol concentrations in the afternoon. Links were observed between PM2.5 and microbial concentrations (Spearman’s rho, rs = 0.5; p < 0.001) but this correlation becomes insignificant when looking specifically at sites where cooking occurred; non-cooking sites continue to show significant correlation. Though the majority of viruses and bacteria are not pathogenic, recent research has indicated that even nonpathogenic and inactivated microbes may influence the oxidative potential of PM. Identifying important microbial sources in these high-PM environments is also necessary to create effect controls (for example, results show lowest microbial concentrations at two sampling environments that were well-sealed and where an air filter was present). In general, this work characterizes various aspects of PM2.5 in environments that many people encounter daily. Unless commutes can be shortened and traffic emissions reduced, or trash can be managed in ways other than burning, these will continue to be important factors in daily PM2.5 exposure.

Ground-level ozone values have been reported to be at unhealthy levels in many populated regions of China. An efficient and effective way for people to combat its harmful effects is seriously needed. However, there are only sparse measurements of ozone being made across China and more measurement sites are needed to understand health relevant concentrations both temporally and spatially. This is due in part to the fact that current monitoring approaches are costly and bulky. Ambient sensors with low cost, small size, and fast response time could potentially fill the current need. Our purpose in this study is to evaluate low cost, portable, and real-time sensors we hope to use in future studies. This study tested the effectiveness of the sensors in monitoring indoor and outdoor air quality, in particular ozone, in Beijing, China.

Seventeen sensors, which were to monitor indoor, outdoor, and personal ozone exposure in 7 homes, were collocated at Peking University (PKU) before and after running in selected residential homes. Pairwise comparisons were conducted using collocated sensor data with data from a standing reference sensor which was maintained by PKU to have best-fit regressions. Based on the best-fit regression when compiling both pre- and post- collocation periods, cleaned sensor data was calibrated and compared with reference data with R2 ranging from 0.63 to 0.97. Overall, the sensors are able to measure O3 within ±17 ppb. Average error of sensors after calibration is about -1 ppb. The geographical conditions and surroundings caused significant differences to the ambient ozone concentrations from sensors located across Beijing. Second, the sensors successfully showed that indoor ozone concentrations were likely to be lower when air filtration was on compared to the concentrations when filters were off. Moreover, the results confirmed higher levels of ozone outdoor concentrations compared to indoor. After calibration, average indoor O3 concentration ranged -2 ppb to 23 ppb, while outdoor O3 concentration was from 19 ppb to 47 ppb. The only personal sensor that worked well measured the average personal exposure as about 33 ppb. The peak of ozone outdoor concentration was usually 100 ppb higher than indoor concentration.

Several factors were considered that could affect the accuracy of the sensors including temperature (degree C), relative humidity (RH%), and concentrations of other chemicals in the ambient environment such as Nitrogen Oxides (NO2, NO). However, the impacts of those factors on the performance of sensors were not significant after calibration with little or no correlation between errors and those factors. Therefore, the calibrations based on the simple linear regressions between the sensors and the reference when collocated were valid on the data during the whole sampling period.

The efficiency of solar panels is significantly reduced by dust and air pollution deposition. This issue (known as soiling) is a roadblock to a renewable future and can only be mitigated by informed cleaning decisions. Globally, there is a lack of reliable soiling data. Furthermore, observed soiling losses are often poorly correlated with ambient air quality, making predictions and planning around mitigation difficult. To address the research and industry needs surrounding soiling, this work addressed the following aims: 1: Develop and validate low-cost soiling monitoring and sample collections systems 2: Deploy monitoring and sample collection systems as well as new analysis methods in a comprehensive study into the soiling processes of Western India 3: Develop an interactive platform that combines a soiling, photovoltaic, and economic model to determine optimized maintenance schedules for photovoltaic installations globally 4: Develop and assess a novel cleaning method using ultrasonic acoustic waves For the first aim, a sensor using a small clean (reference) and a dirty (test) photovoltaic panel was developed and tested in the field against a Campbell Scientific SMP100 Soiling Station. The sensor was shown to accurately quantify soiling to within ± 1.5 % soiling at an order-of-magnitude lower cost than available to the industry. A digital microscope system was then tested for the ability to estimate soiling from image analysis applied to microscopy images of deposited dust. The digital microscope sensor and mass loading estimation from a custom image analysis procedure showed a high correlation (R2 between mass loading and measured soiling loss). Calibration of the system allows for soiling estimations with an accuracy of ± 1.1 % soiling. Aim two first quantified soiling impacts throughout the year in Western India at a test site in Gandhinagar India. Significant seasonality was observed with high soiling rates in the dry season (0.45 ± 0.10 % day-1) and monsoon rains keeping soiling losses < 5 %. The threshold for removal from rainfall was found to vary from 0.5 mm to 8 mm per day and is likely impacted by meteorological conditions impacts particle adhesion prior to the rain event. A new method for analysis of deposited PM size distribution using light microscopy was developed. Size distribution of deposited PM displayed a bi-modal distribution with peaks in the 0.5-1 µm and the 16-30 µm diameter range. Monsoon rains shifted size distribution, cleaning off 89.8% of particles with a diameter > 10 µm in diameter. High humidity in the monsoon season created a buildup of more than two times (123%) the mass of particles smaller than 2.5 µm in diameter despite lower ambient concentrations – likely due to increased capture efficiency and lower particle rebound. This is also evident in effective particle deposition velocity being 5 to 10 times higher during periods of high (>80% humidity). The light microscopy method was also used to develop a novel method of determining soiling non-uniformity on the millimeter scale, deemed milli-scale non-uniformity (MSNU). Rainfall and high humidity leading to dew formation caused significantly higher MSNU on the surface. Building on the light microscopy method, a new method of estimating size-resolved soiling impacts was developed by applying Mie Theory to the particle size distribution. Despite making up less than 10% of the mass on the surface, particles smaller than 5 µm in diameter contribute to > 50% of the soiling impacts, highlighting the importance of small particles to energy losses. Scanning electron microscopy combined with energy dispersive x-ray spectroscopy was then performed on ~700 individual deposited particles to quantity particle composition. A majority of the deposited mass was found to be crustal mineral dust likely transported over long distances. A range of precipitation products highlight the chemical and physical reactions that take place on the panel surface over periods of wetting and drying. Cementation impacts were dominated by carbon-heavy precipitation masses, deemed carbonaceous caking masses (CCMs). It is likely CCMs are a concern for energy losses and cleaning difficulty in many regions of the world that experience dry periods with dew formation. Fungal growth was observed to be present and spore-producing after just 3-weeks – the smallest time period over found for fungal growth on solar panels. Building off the efforts of Bergin et al. 2017, an improved global model for estimating soiling losses throughout the year was developed. The model is shown to accurately estimate soiling losses over orders-of-magnitude at 16 locations across the world. This soiling loss model was combined with models of energy generation and economic optimization. Hosted in an interactive web platform, Solar Unsoiled INC. was formed and offers a first-of-its kind tool for the solar industry to prioritize, plan, budget, manage, and optimize panel cleanings. Finally, a transducer system using surface acoustic waves was developed and tested for deposited particle removal. The system is the first study that used surface acoustic waves (SAWs) to remove particles in the size range relevant to soiling. The system was shown to remove and average of 70% of deposited mass from a glass surface. These promising results indicate a potential for a new method to prevent and remove soiling buildup from solar panels.