PROGRESS IN GEOGRAPHY ›› 2019, Vol. 38 ›› Issue (5): 718-730.doi: 10.18306/dlkxjz.2019.05.009

• Articles • Previous Articles     Next Articles

Detecting Granger effect of vegetation response to climatic factors on the Tibetan Plateau

Yuke ZHOU()   

  1. Key Laboratory of Ecosystem Network Observation and Modeling, Institute of Geographic Sciences and Nature Resources Research, CAS, Beijing 100101, China
  • Received:2018-09-11 Revised:2019-02-25 Online:2019-05-28 Published:2019-05-28
  • Supported by:
    National Natural Science Foundation of China, No. 41601478 and 31700417; National Key Research and Development Program of China, No. 2016YFC0500103; STS Program of Chinese Academy of Sciences, No. KFJ-SW-STS-167; Open Fund of the State Key Laboratory of Resources and Environmental Information Systems in 2016

Abstract:

Due to the complex plateau climate and unique geographical environment, the vegetation responds strongly to climatic shifts on the Tibetan Plateau. Therefore, it is of great significance to discuss the causality between vegetation and climate changes. Using the meteorological dataset including average temperature and precipitation and the GIMMS (Global Inventory Modeling and Mapping Studies) NDVI3g remote sensing data from 1982 to 2012 to analyze the causal relationship between NDVI and its influencing factors at the monthly and seasonal scales by the Granger causality test on the pixel level, this study examined the response of plateau vegetation (mainly grassland) to average temperature and precipitation change and causality. The results show that: 1) The stationarity proportion of vegetation NDVI and average temperature (99.13%), NDVI and precipitation (98.68%) at the monthly scale was higher than at the seasonal scale (64.01% and 71.97% respectively). 2) Lagging effects of average temperature on NDVI and precipitation on NDVI were around 12-13 months at the monthly scale and mainly 3, 4, and 6 quarters at the seasonal scale on the Tibetan Plateau. The three vegetation types—desert steppe, typical steppe, and meadow steppe—showed high similarities at the monthly scale, while greater heterogeneity was observed at the seasonal scale. 3) For 98.95% of the area covered by vegetation on the Tibetan Plateau, it is believed that average temperature change was the Granger cause of NDVI change, while for 89.05% of the region (except for the southeast), NDVI change was supposed to be the Granger cause of average temperature change at the monthly scale. At the seasonal scale, average temperature change was considered the Granger cause of NDVI change in 92.03% of the regsion (except for the central part of the Tibetan Plateau). Nevertheless, in the eastern and western parts of the plateau (about 50.55% of the region), NDVI change was interpreted as the Granger cause of average temperature change. 4) In the northeast and northwest of the region (about 98.95% of the area) precipitation change was believed to be the Granger cause of NDVI change, while in 94.86% of the region (except for a few areas in the southeast) NDVI change was supposed to be the Granger cause of precipitation change at the monthly scale. At the seasonal scale, precipitation change was considered the Granger cause of NDVI change in the southeastern part of the plateau (61.43% of the area). Nevertheless, in the central and eastern parts of the region (about 48.98% of the area), NDVI change was interpreted as the Granger cause of precipitation change. Overall, climatic factors on the Tibetan Plateau have an interactive relationship with vegetation and each can cause a Grainger effect to the other, with climatic factors having stronger Grainger effect on vegetation than the other way round. The Granger effect region on the Tibetan Plateau at the monthly scale is larger than the Granger effect region at the seasonal scale.

Key words: Tibetan Plateau, stationarity test, Granger causality, multi-scale analysis, lag order, NDVI, climate change