Use of airborne PM10 concentrations at air quality monitoring sites in Imperial County, California, as an indication of geographical influences on lung health during drought periods: a time-series analysis

Background: 2012–16 marked the worst drought in California, USA, in more than a century. Droughts are expected to recur in broader areas as a result of anthropogenic climate change. The Salton Sea, a landlocked depression without natural river inflow, is almost entirely dependent on agricultural irrigation of Imperial Valley's 475 000 acres of farmland, although its water level has been diminishing because of decreased precipitation and river flow. Exposure of the lakebed and ground heating in an already arid climate might cause an increase in airborne PM10 (particulate matter less than 10 μm in diameter), reactive nitrogen, and ground-level ozone, exacerbating asthma symptoms. During the most severe drought years, admissions of children to the emergency department for asthma-related symptoms in Imperial County were among the highest in California and double the number for the entire state. However, previous analyses have not taken into account environmental mediators with more specific geographical granularity than the county level. Methods: In this analysis, we used 2006–16 data from the US Drought Monitor and the five air quality monitoring stations (AQMS) within Imperial County, CA, USA. To reconcile sampling differences between 7-day drought severity reports and 6-day PM10 measurements, a date windowing function was applied to match each PM10 measurement to the drought severity value at the most closely related point in absolute time. Upon visually examining the data, lower concentrations of PM10 appeared to drive the correlation between AQMS. Therefore, in two time series—all years and non-drought years only—all concentrations of PM10 that were lower than 75 ppm were removed, and the remaining concentrations were used to measure correlations between AQMS. Findings: PM10 concentrations across all years were poorly correlated between AQMS (r=0·41–0·80; p<0·05). When removing PM10 concentrations lower than 75 ppm and examining only the remaining values, the correlation coefficients for these values decreased (r=0·16–0·53; p<0·05). When removing drought years (2012 onwards), these values were slightly more correlated in comparisons between some AQMS than across all observed years (r=0·46–0·77; p