Spatial variability and time series analysis of aerosol optical properties over Indio-gangetic Plain

Atmospheric aerosols play an important role in determining Earth's radiation budget directly by scattering and absorbing incoming solar radiation and indirectly by modifying the cloud microphysical processes. To understand the effect of aerosols on climate, it is essential to know their optical properties as well as spatial and temporal variability. To estimate the spatial variability of aerosols, MODIS aerosol product was acquired from the NASA Goddard Earth Science Distributed Active Archive Center (DAAC) for a period of six years. The satellite data for aerosol optical depth (AOD) was validated with the data retrieved from the ground based multi-wavelength Prede Sun/sky radiometer (POM-02) over Delhi, India. The correlation coefficient was found to be 0.5 indicating medium to high correlation between MODIS and SKYRAD data. Positive fractional bias showed moderately over-predicting MODIS AOD values as compared to Skyradiometer AOD. The spatial distribution pattern of AOD was merged with the landuse/landcover map of Delhi and adjacent areas. The resulting image indicated that the higher AOD (>0.7) was observed in the built-up areas with construction activities, mostly covering the central, south and east Delhi. Also a large seasonal variation was observed in AOD, Angstrom exponent (AE) and Single Scattering Albedo (SSA) throughout the study period. Further in this study, the seasonal trend in the past ten years of AOD data was utilized in developing a stochastic model using Box- Jenkins Autoregressive Integrated Moving Average (ARIMA) modeling approach for predicting the future AOD values. Based on the evaluation criteria (considering normalized Bayesian Information Criterion or BIC), it was found that ARIMA (1,0,0)x(0,1,2)₁₂ model is adequate for simulating the AOD parameter. Forecasts based on the selected model were carried out for the year 2014. It was observed from the results that some aspects of the past continue to influence the future values of AOD, indicating that there is a long memory for AOD over the study region. This stochastic modeling process could be further applied to other climatic modeling purposes and future trend analysis.