UC Berkeley

Fog has broad impacts on transportation safety, agricultural production, drought resilience, and climate. The frequency of wintertime radiation fog in valleys throughout the world has been changing over the past century. This dissertation focuses on understanding the drivers of fog trends observed in California’s Central Valley and Italy’s Po Valley, specifically investigating the competing effects of warming climate versus air pollution controls in the twentieth and early twenty-first century.

Beginning in the Central Valley, this dissertation finds that dense fog frequency (visibility < 400 meters) increased 85% from 1930-1970, then declined 76% in the last 36 winters. Throughout these changes, fog frequency exhibited a consistent north-south trend, with maxima in southern latitudes. I analyzed seven decades of meteorological data and five decades of air pollution data to determine the most likely drivers changing fog, including temperature, dew point depression, precipitation, wind speed, and NOx (oxides of nitrogen) concentration. Climate variables, most critically dew point depression (DPD), strongly influence the short-term (daily to annual) variability in fog frequency; however, the frequency of optimal conditions for fog formation show no observable long-term trend from 1980 to 2016. NOx concentration, which is a limiting precursor to the ammonium nitrate aerosol that dominates wintertime particulate matter in the valley, has an increasing north-south concentration gradient, consistent with the gradient in fog frequency. NOx declined continuously over this period, also consistent with the long-term temporal and spatial trends in fog. As development in the Central Valley increased direct particle and other pollutant emissions from 1930-1970, fog frequency increased. Following the Clean Air Act, particle emissions quickly declined, and NOx emissions declined steadily, reducing the cloud condensation nuclei (CCN) available for fog formation. I conclude that while the short-term fog variability is dominantly driven by climate fluctuations, the longer-term temporal and spatial changes in fog have been driven by changes in air pollution. For conditions close to the dew point, a decrease in fog of 5 days per year per 10 ppb NOx decrease occurred across the Central Valley.

To further understand the multivariate, nonlinear contributors to fog formation, this dissertation used generalized additive models to identify the relative significance of climate and air pollution variables affecting visibility, an indicator of dense fog, and compare the drivers changing fog in the Central Valley to Italy’s Po Valley, which saw a 50% decline in fog frequency since 1980. Effective regulatory strategies in the Po Valley have also resulted in stark reductions in inorganic pollution emissions, thus reducing the CCN available for fog formation. Over 56-65% of the variance in visibility is consistently explained by variability in DPD, pollution concentration, wind speed, and precipitation. Variability in DPD, which incorporates both water availability and temperature, has the most pronounced influence on daily time scales, but shows no substantial long-term trends in time over the observation period. The variability in NOx concentration explains 33-70% of the variance in visibility (depending on the site) when investigating days with average DPD < 3.5℃. This suggests that fog frequency is specifically sensitive to fluctuations in CCN number concentration when meteorological conditions are favorable to fog formation (e.g. when DPD is low). While DPD is a primary driver of daily variability, the significant influence of NOx concentration on the visibility response suggests that rapid pollution declines in both valleys have had an important impact on the diminished fog season since 1980. This demonstrates that the regulatory measures that mitigate pollution concentration have valuable benefits, not only on the health outcomes of those potentially exposed, but also in reducing the dangerous dense fog frequency that was anthropogenically enhanced with industrialization.

To further understand the safety implications of reductions in air pollution and fog events, I analyzed a 20-year record of fog-related accidents in the Central Valley (1996-2016). Decades of multicar pile-ups along its highways made the region widely-known for the frequency and severity of its fog-related accidents. Yet, the Central Valley saw a 65% decline in fog-related accidents over 20 winters, the variance of which is best explained by the sharply declining trend in seasonal fog hours over the same period. Annual frequency of fog hours as summarized for each fog season explain an average of ~80% of the annual variability in fog-related accidents in the counties of highest roadway volume, showing that the declining trend in fog is a strong determinant in the declining trend in accidents. The subsequent improvement in visibility results in annual fog-related injuries falling by 72%, with the valley seeing an average of 550 fewer injuries from fog accidents in 2015-2016 than in 1996-1997.

The human safety and commercial benefits to a reduction in fog-accidents and the resulting roadway delays is well documented. The declining trend in dense fog in this region has had a pronounced impact on the declining frequency of fog-related accidents. This dissertation implicates regional air pollution concentration as a critical driver in the long-term trend of fog frequency. The strong link between the historical number of Central Valley fog events and trends in pollution concentration provides a measure of how regulations that led to decreases in aerosol concentration, and thereby wintertime fog frequency, also influenced the declining trend in fog-related accidents.