Spatiotemporal Analysis of Air Pollution Patterns and Their Environmental Impacts Using Advanced Statistical Models

Panca P.H.; Flores B.C.; Acero R.C.; Coa N.S.J.; Vargas E.E.C.; Mamani G.Q.; Lima R.G. · World Journal of Environmental Biosciences (2026) · DOI: 10.51847/bt9odbzg9n

Resumen

Air pollution varies across space and time because emissions, atmospheric chemistry, meteorology, and land-use structure interact at multiple spatial and temporal scales. Its environmental impacts, including crop loss, ecosystem damage, and human health effects, depend on accurate exposure surfaces, yet routine monitoring networks provide sparse point observations rather than continuous fields. Conventional regression and simple interpolation often ignore joint spatial and temporal dependence. As a result, they can produce biased risk estimates, underestimated uncertainty, and weak predictions at unsampled locations or during extreme pollution episodes. This spatiotemporal analysis develops an advanced statistical framework to map air pollution patterns and quantify their environmental impacts. The focus is on PM2.5 and NO2, with extensions to PM10, SO2, O3, and CO where monitoring and satellite data permit. The proposed framework uses a Bayesian hierarchical model with Matérn spatial random effects and temporal autoregressive dependence, fitted to monitoring-station observations, satellite aerosol optical depth, chemical transport model outputs, meteorological fields, and land-use covariates. A second-stage exposure-response model links predicted pollution surfaces to crop-yield or health outcomes while propagating exposure uncertainty. Conceptually, the model produces high-resolution daily pollution surfaces, identifies persistent spatial clustering with Moran’s I greater than 0.6, and separates long-term trends from seasonal and event-driven variability. A 10 µg/m³ increase in PM2.5 is expected to correspond to a 5–12% reduction in wheat yield or increase in respiratory admissions, depending on the outcome model and regional susceptibility. Spatiotemporal modeling provides more precise air-pollution estimates and more credible environmental-impact measures than models that ignore dependence or uncertainty. These methods support targeted mitigation policies, early-warning systems, and spatially explicit environmental planning.

Instituciones: Universidad Nacional del Altiplano