Publications

Zang, Z; Guo, YS; Jiang, YZ; Zuo, C; Li, D; Shi, WZ; Yan, X (2021). Tree-based ensemble deep learning model for spatiotemporal surface ozone (O-3) prediction and interpretation. INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION, 103, 102516.

Abstract
Tree-based machine learning and deep learning approaches are widely applied in ozone (O-3) retrieval, but they cannot achieve high accuracy and interpretability simultaneously. To overcome this limitation, a tree-based ensemble deep learning model, named semi-SILDM, was proposed for O-3 prediction at both national (5 km) and urban scales (250 m) in China. The Moderate Resolution Imaging Spectroradiometer (MODIS) Top of Atmosphere (TOA) measurements were first investigated through significant linear and nonlinear relationships with surface O-3. To examine the actual predictive ability of the semi-SIDLM, time-based validation was employed to divide data chronologically by year into training (2018), validation (2017), and test data (2019). The semi-SIDLM predicted O-3 in 2019 showed a coefficient of determination (R-2) of 0.71 (0.69) and a Root Mean Square Error (RMSE) of 21.88 (26.59) mu g/m(3) at the national (urban) scale in China. In addition to its high accuracy, the semi-SIDLM has interpretability for retrieval results, which indicates the strong influence of the Fangshan and Tongzhou districts on the principle O-3 Beijing urban area; the temporal characteristics reveal the higher contributions of May-July to O-3 pollution compared to other months. The proposed model of this study will benefit further studies on O-3 monitoring and deepen the understanding of its spatiotemporal characteristics.

DOI:
10.1016/j.jag.2021.102516

ISSN:
1569-8432