Skip to main content
Dryad logo

Decreasing effects of precipitation on grassland spring phenology in temperate China

Citation

Zhou, Xuancheng et al. (2021), Decreasing effects of precipitation on grassland spring phenology in temperate China, Dryad, Dataset, https://doi.org/10.5061/dryad.mkkwh70xn

Abstract

Vegetation phenology is highly sensitive to climate change. The timing of spring phenology in temperate grasslands is primarily regulated by temperature and precipitation. This study aims to study whether the primary factor regulating vegetation phenology changed under ongoing climate change and its underlying mechanisms. In this study, we extracted Start of Season (SOS) dates using five standard methods from satellite-derived Normalized Difference Vegetation Index (NDVI) data and determined the primary regulating factor for spring phenology using partial correlation analysis.

Methods

We used the satellite Normalized Difference Vegetation Index (NDVI) records by NASA’s GIMMS group from 1982-2015. This NDVI dataset has been produced at a spatial resolution of 8 km and a time resolution of 15 days.

In order to remove the disturbing artifacts, we defined that no spring phenological events occurred before air temperature surpassed 0 ℃ for 5 consecutive days (Cong et al., 2012). In addition, we excluded pixels where mean annual NDVI value was below 0.1. Croplands were removed from the data in order to avoid the anthropogenic influences in the data. We used in the Vegetation Map of the People’s Republic of China the same resolution as was used in the NDVI images and then removed the pixels marked as croplands. Extracting dates of phenological spring events took place in two steps.

First, we used one of five filter functions at a time to smooth the NDVI series, since in those series there are always some abnormal values due to atmospheric interference. Then we interpolated the daily values between the biweekly observations, because a biweekly resolution is too coarse to estimate dates of spring phenological events. Finally, we defined a threshold value of NDVI for identifying the date of the spring phenological events. To illustrate the methods used in extracting the SOS dates, we chose one pixel in a meadow steppe and used three filter methods as an example (Figure S2). Five methods, each characterized by its filter function and the corresponding threshold, were used for extracting the dates of start of season at each pixel from the NDVI: Gaussian, Spline, Polyfit, HANTS, and Timesat-SG. After we got the SOS from each method, we used the mean value of five different methods to represent the SOS date at each pixel.