Evaluating the accuracy of and evaluating the potential errors in extracting vegetation phenology through remote sensing in China

ABSTRACT

Using remote sensing to study vegetation phenology faces a problem related to extraction methods. Phenology data from different methods vary greatly but there is a lack of widely recognized methods and evaluation approaches. In this study, based on the Leaf Area Index data from 2009 to 2015, we used two common fitting methods (Savitzky-Golay filter and Double Logistic function) and three phenological determining methods (Seasonal amplitude, Absolute value, and Seasonal Trend decomposition by Loess) to extract vegetation phenology in China. Using different thresholds, we obtained 18 extraction method combinations. Then the ground-based observations data from 31 phenology stations in relation to three key vegetation phenophases: Start Of growing Season (SOS), End Of growing Season (EOS) and Length Of growing Season (LOS) were used to evaluate the accuracy of these 18 method combinations. The results under five evaluation indicators showed that the suitable method combinations for SOS, EOS, and LOS were different. Compared with ground-based observations, SOS and EOS extracted by the suitable method combinations were delayed by 6.28 and 4.91 days, and the LOS was shorter. The potential difference of the suitable and unsuitable method combinations respectively reached −44.73, −35.79, and 37.38 days (for SOS, EOS, and LOS), clearly indicating the importance of selecting suitable method combination. The phenology dataset from Vegetation Index & Phenology (VIP) Lab. has also confirmed the reliability of our results. Furthermore, we explored the differences between remote sensing and ground-based phenology. Our study highlights the importance of using a suitable method combination to extract the vegetation phenology and provides a systematical assessment method for selecting a suitable method combination.

Acknowledgements

We greatly appreciate the editor and the anonymous reviewers for the valuable suggestions and comments. And we thank Leonie Seabrook, Ph.D., from Liwen Bianji, Edanz Group China, for editing the English text of a draft of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Natural Science Foundation Project of China [41671425, 41401504, 41401129], and Graduate education innovation and quality improvement program of Henan University [CX3040A0780203].