*2.3. Data Sources*

## 2.3.1. MOD09A1 Data

The main vegetation types in the TRHR are alpine steppe and alpine meadow. These do not have high amounts of vegetation coverage, with the highest values of NDVI being less than 0.6. Therefore, there are no areas where vegetation is so saturated that the NDVI cannot be accurately expressed. Hence, NDVI was selected in this paper for use in analyzing plant phenological characteristics. Based on a previous study, this paper selected MOD09A1 data products because they have a high temporal resolution, which can provide us with better detailed information related to vegetation growth. The NDVI time-series

data came from the Google Earth Engine (https://developers.google.com/earth-engine/ datasets/catalog/MODIS\_006\_MOD09A1 (accessed on 20 July 2020)), with a temporal resolution of 8 days and a spatial resolution of 500 × 500 m. In order to improve the calculated phenological results accurately, we used the following rules for data processing. (1) NDVI was calculated based on the MOD09A1 band using Equations (1) and (2). To eliminate the influence of bare soil, sparse vegetation, and evergreen forest, the pixels of NDVI data in this study had to meet the following requirements: (a) the average value of NDVI should be more than 0.2 in April–October; (b) the maximum value of annual NDVI shall exceed 0.30; (c) the annual maximum value shall occur from July to September; and (d) the average value of NDVI in winter shall be less than 0.4. (3) The NDVI data with a temporal resolution of 1 day was obtained by using HANTS to fit the data, which were processed in steps 1 and 2 (Figure 3a).

**Figure 3.** (**a**) Fitting effect of the harmonic analysis of time series; (**b**) definition of the normalized difference vegetation index (NDVI), threshold of the start of the growing season (SOS), and end of the growing season (EOS). The green line is the NDVI time-series change curve of 8 days after smooth treatment. The vertical axis of the left side is the NDVI value, and the vertical axis of the right side is the change rate of NDVI. For comparison, the change rate of NDVI has been zoomed in integer times, the ratio of absolute change value is 1000, and the ratio of relative change rate is 100.

### 2.3.2. Phenological Observation Data

Vegetation phenological observation data (2001–2013) of the THRH used in this study were extracted from the ten-day datasets on crop growth and farmland soil moisture in China, which were obtained from the Chinese Meteorological Administration (http: //data.cma.cn/ (accessed on 12 October 2020)). We selected the phenological stations according to the principle that the vegetation type around each station is grassland. Finally, we selected five phenological observation stations (Figure 2c).
