Over the past decades, regional haze episodes have frequently occurred in eastern China, especially in the Yangtze River Delta (YRD). Satellite derived Aerosol Optical Depth (AOD) has been used to retrieve the spatial coverage of PM_2.5 concentrations. To improve the retrieval accuracy of the daily AOD-PM_2.5 model, various auxiliary variables like meteorological or geographical factors have been adopted into the Geographically Weighted Regression (GWR) model. However, these variables are always arbitrarily selected without deep consideration of their potentially varying temporal or spatial contributions in the model performance. In this manuscript, we put forward an automatic procedure to select proper auxiliary variables from meteorological and geographical factors and obtain their optimal combinations to construct four seasonal GWR models. We employ two different schemes to comprehensively test the performance of our proposed GWR models: (1) comparison with other regular GWR models by varying the number of auxiliary variables; and (2) comparison with observed ground-level PM_2.5 concentrations. The result shows that our GWR models of “AOD + 3” with three common meteorological variables generally perform better than all the other GWR models involved. Our models also show powerful prediction capabilities in PM_2.5 concentrations with only slight overfitting. The determination coefficients R² of our seasonal models are 0.8259 in spring, 0.7818 in summer, 0.8407 in autumn, and 0.7689 in winter. Also, the seasonal models in summer and autumn behave better than those in spring and winter. The comparison between seasonal and yearly models further validates the specific seasonal pattern of auxiliary variables of the GWR model in the YRD. We also stress the importance of key variables and propose a selection process in the AOD-PM_2.5 model. Our work validates the significance of proper auxiliary variables in modelling the AOD-PM_2.5 relationships and provides a good alternative in retrieving daily PM_2.5 concentrations from remote sensing images in the YRD.

A novel Symmetric Sparse Representation (SSR) method has been presented to solve the band selection problem in hyperspectral imagery (HSI) classification. The method assumes that the selected bands and the original HSI bands are sparsely represented by each other, i.e., symmetrically represented. The method formulates band selection into a famous problem of archetypal analysis and selects the representative bands by finding the archetypes in the minimal convex hull containing the HSI band points (i.e., one band corresponds to a band point in the high-dimensional feature space). Without any other parameter tuning work except the size of band subset, the SSR optimizes the band selection program using the block-coordinate descent scheme. Four state-of-the-art methods are utilized to make comparisons with the SSR on the Indian Pines and PaviaU HSI datasets. Experimental results illustrate that SSR outperforms all four methods in classification accuracies (i.e., Average Classification Accuracy (ACA) and Overall Classification Accuracy (OCA)) and three quantitative evaluation results (i.e., Average Information Entropy (AIE), Average Correlation Coefficient (ACC) and Average Relative Entropy (ARE)), whereas it takes the second shortest computational time. Therefore, the proposed SSR is a good alternative method for band selection of HSI classification in realistic applications.