Monitoring the impact of aerosol contamination on the drought-induced decline of gross primary productivity

• Published in: International journal of applied earth observation and geoinformation Vol. 36; pp. 30 - 40
• Main Authors: Zhang, Yao, Li, Weizhong, Zhu, Qiuan, Chen, Huai, Fang, Xiuqin, Zhang, Tinglong, Zhao, Pengxiang, Peng, Changhui
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.04.2015
• Subjects: Aerosol; Atmospheric contamination; Cloud; Fire disturbance; Quality flag; Vegetation response; Atmospheric contamination; Quality flag; Vegetation response; Aerosol; Fire disturbance; Cloud
• ISSN: 1569-8432; 1872-826X
• DOI: 10.1016/j.jag.2014.11.006

•Previous estimation of the GPP decline in southwestern China in 2010 is unreliable due to aerosol contamination.•Temporal filling method does not work well if too little uncorrupted scenes available.•Remotely sensed data should be cautiously used in drought monitoring if the aerosol exists. Southwestern China experienced a period of severe drought from September 2009 to May 2010. It led to widespread decline in the enhanced vegetation index (EVI) and gross primary productivity (GPP) in the springtime of 2010 (March to May). However, this study observed a spatial inconsistency between drought-impacted vegetation decline and the precipitation deficit. Significant aerosol loads that correspond to inconsistent areas were also observed during the drought period. After analyzing both MODIS GPP/NPP Collection 5 (C5) and the newer Collection 5.5 (C55) data, a large area observed to be experiencing GPP decline in the eastern part of the study area proved to be unreliable. Based on EVI data, after atmospherically contaminated data were screened from analysis, approximately 20% of the study area exhibited browning whereas 33% displayed no change or greening and the remaining area (approximately 47%) lacked sufficient data to document changing conditions. Correlation analysis showed that fire occurrences, aerosol optical depth, and precipitation anomalies during the two drought periods (from September to February and from March to May) all contributed to a decrease in GPP. C55 data remains vulnerable to aerosol contamination due to a much higher correlation coefficient with aerosol optical depth compared to C5 data. In the future, users of remotely sensed data should be cautious of and take into account impacts related to atmospheric contamination, even during drought periods.