Evaluation and comparison of growing season metrics in arid and semi-arid areas of northern China under climate change

Abstract

The differences of three key growing season metrics, namely, the start of growing season (SOS), the end of growing season (EOS) and the length of growing season (LOS), derived from ground observations, satellite-derived normalized difference vegetation index (NDVI) data and air temperature were compared, and the spatial distribution and temporal trend of NDVI-based growing season metrics were analyzed in the arid and semi-arid areas of northern China. The results show that the growing season metrics obtained by three methods were quite different. The temperature-based SOS dates were earlier than those observed at the four phenological sites, while the NDVI-based SOS dates were the latest. The EOS (LOS) derived from NDVI and temperature were later (longer) than the observed values. At 240 meteorological stations, temperature-based SOS and EOS dates were generally earlier than the NDVI-based results, leading to longer temperature-based LOS than that derived from NDVI. From 1982 to 2015, the NDVI-based SOS was advanced by 2.3 days per decade, and the NDVI-based EOS was delayed by 9.5 days per decade, causing a prolonged LOS of 11.8 days per decade. Earlier SOS, later EOS and prolonged LOS were significantly appeared at 44.3%, 40.6% and 52.6% of the vegetated area respectively. The significant advanced SOS was concentrated in the central Inner Mongolia, northern Hebei, northern Shanxi and northwestern Xinjiang, while the significant delayed EOS was mainly distributed in the majority of western and northern Xinjiang and northern Inner Mongolia, resulting in the significant prolonged LOS in most of the study area. Each method has its distinct advantages and disadvantages. It is suggested to use high spatial and temporal resolution satellite images and appropriate methods to establish phenological models for different land cover types and to take into account the powerful vegetation indices and key climatic factors such as daytime temperature and precipitation and their interactions, which are of great significance for large-scale and accurate phenological monitoring and assessment.