Evaluation of non-stationarity in summer precipitation and the response of vegetation over the typical steppe in Inner Mongolia

The typical steppe in Inner Mongolia is an important component of the Eurasian steppes. It plays a dominant role in preventing desertification and against sandstorms, but highly sensitive and vulnerable to climate change. Based on long-term observed precipitation data and remotely sensed Normalized Difference Vegetation Index (NDVI) images, the non-stationary behavior of summer precipitation and its linkage with vegetation change were investigated, by means of incorporating time-varying and physical based explanatory covariates in non-stationary modeling. Results indicated that time-dependent models exhibited good performance to reproduce the temporal variations of eco-hydrological variables. The non-stationarity of summer precipitation was prominently visible for the majority of sites during the period from 1957 to 2017, with the mean behavior described as a linear or nonlinear time-varying pattern. In general, the steppe has experienced a decreasing trend in summer precipitation, but whether the decline tends to maintain or weaken or strengthen depends on the spatial location of the site studied. Differences appeared in the changes of vegetation in summer from 1998 to 2017 in different sub-regions. Evidences for the presence of stationary evolution was found in most sub-regions in the middle part, together with a linear increase in the westernmost sub-regions while a non-linear decrease in the easternmost sub-regions. Covariate analyses further highlighted the role of precipitation variabilities in the modeling of the NDVI-related vegetation dynamics over the steppe. The potential relations of summer precipitation to vegetation growth were characterized as both linear and non-linear positive forms. In particular, precipitation extremes could be responsible for the occurrences of exceptional cases in vegetation condition. The fluctuations in summer precipitation have crucial significance for future predictions of vegetation succession. Findings from this study would lead to additional insights to understanding the effect of climate change on grassland ecosystem processes.