Abstract
Vegetation plays an important role in regulating the climate system and terrestrial carbon cycle. In recent decades, numerous studies have focused on monitoring vegetation greenness changes at different scales with long-term remote-sensing data. Most previous studies adopted a monotonic linear regression approach (one-piece linear regression) to detect vegetation greenness trends at global and regional scales. However, the monotonic linear regression approach is limited in its ability to detect abrupt changes in vegetation greenness trends because non-monotonic phenomena may exist in these trends. This study applied a piecewise linear regression (PLR) method to detect the non-monotonic trends of vegetation greenness in the Tibetan Plateau from 1982 to 2006. The vegetation greenness was indicated by the normalized difference vegetation index (NDVI) data derived from Global Inventory Modelling and Mapping Studies (GIMMS) Advanced Very High Resolution Radiometer (AVHRR) data. The results implied that the PLR method could detect an abrupt change in the trend with a break point (i.e. the year when the vegetation greenness trend changed abruptly) mainly concentrated in 1989. We chose three typical regions to compare the PLR and monotonic linear regression approach for vegetation trend detection to verify the effectiveness of the PLR method. The PLR method can detect different trends during different periods, but the one-piece linear regression method can only detect monotonic trends. After comparing the results of the two methods for three climatic zones, we found that the greenness trend detected by PLR can be better explained by temperature and precipitation variations. Our results illustrate that the PLR method is superior to the one-piece linear regression method due to its ability to detect non-monotonic trends. It can, therefore, delineate vegetation greenness trends in detail and be applied to other, similar studies.
Acknowledgements
We thank the anonymous reviewers for their constructive comments on the manuscript.
Funding
This study was supported by the National Natural Science Foundation of China [grant number. 41271372], the Director Innovation Foundation of CEODE, CAS [grant number Y2ZZ19101B], and the National Natural Science Foundation of China [grant number 41330634].