The application of polynomial analyses to detect global vegetation dynamics during 1982-2012

Use of the normalized difference vegetation index (NDVI) to build long-term vegetation trends is one of the most effective techniques for identifying global environmental change. Trend identification can be achieved by ordinary least squares (OLS) analysis or the Theil-Sen (TS) procedure with a Mann-Kendall (MK) significance test, and these linear regression approaches have been widely used. However, vegetation changes are not linear, and thus the response of vegetation to global climate change may follow non-linear trends. In this article, a polynomial trend-fitting method, which uses stepwise regression and expands on previous research, is presented. With an improved fitting ability, this procedure may reveal trends that were concealed by linear fitting methods. Globally, the traditional TS-MK method reveals significant greening trends for 37.27% of vegetated land, and significant browning trends for 7.98%. Using the polynomial analysis, 34.62% of pixels were fitted by high-order trends. The significant greening trends covered up to 30% of cultivated land, thus indicating that cultivated vegetation may be increasing faster than natural vegetation. Significant vegetation browning mostly occurred in sparse vegetation areas, which suggests that vegetation growth may be more sensitive to climate change in arid regions. Our results show that use of polynomial analysis can help further elucidate global NDVI trends.