Impacts of terrain on land surface phenology derived from Harmonized Landsat 8 and Sentinel-2 in the Tianshan Mountains, China

Chao Ding, Yao Li, Qiaoyun Xie, Hao Li, Bingwei Zhang

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)


Knowledge of terrain impacts on land surface phenology (LSP) is crucial for understanding the responses of mountainous ecosystems to environmental changes. While the effects of terrain factors on LSP spatial patterns have been observed to vary across regions due to their different climate and terrain conditions, the specific effects across different elevations are still largely unclear, especially in regions with diverse hydrothermal conditions, such as the Tianshan Mountains located in arid and semiarid region. Here, we investigated the spatial relationships between LSP metrics and terrain factors (i.e. elevation and aspect) in the Tianshan Mountains in Xinjiang, China. Our analysis utilized surface reflectance at a 30 m spatial resolution from the Harmonized Landsat 8 and Sentinel-2 dataset for 2021 and 2022. We focused on two LSP metrics, vegetation greenup (GU20) and maturity (GU90), which were estimated using 20% and 90% thresholds of seasonal amplitude of the enhanced vegetation index (EVI) time series, respectively. We modeled the spatial relationships using ordinary least square (OLS) linear regression for the entire study region and then applied geographically weighted regression (GWR) with a 2.5 km bandwidth to explore local variations. Our results suggest that, at a large scale, elevation played a primary role in controlling the spatial variations in both LSP metrics, overshadowing the role of aspect. However, when examined at a local scale using GWR, aspect emerged as an important factor, with south-facing aspects associated with earlier dates of GU20 and GU90 for most regions. Furthermore, we found that the influences of terrain on the LSP metrics varied across elevations. The explanatory power of terrain was stronger at middle elevations (approximately 2000–3000 m) than at lower (<2000 m) and higher (>3000 m) elevations. In addition, the sensitivities of the LSP metrics to elevation and aspect demonstrated varying patterns at elevations above 2000 m. Our findings highlight the diverse environmental controls on LSP across elevations, with a particular emphasis on the phenological sensitivities to aspect-induced local climatic differences.

Original languageEnglish
Article number2242621
JournalGIScience and Remote Sensing
Issue number1
Publication statusPublished - 2023

Cite this