Quantifying effects of compound dry-hot extremes on vegetation in Xinjiang (China) using a vine-copula conditional probability model

Highlights

• A VCCP model is developed for quantifying compound risks on vegetation in Xinjiang.

• Spatial quantitative response of vegetation to compound dry-hot event is identified.

• Vegetation loss reaches peak in August conditioned on compound extreme conditions.

• Results are helpful for decision maker to make extreme mitigation recommendations.

Abstract

Extreme events (e.g., drought and heatwave) occur frequently and intensively with climate change, where the combination of dry and hot events has catastrophic impacts on terrestrial ecosystems. It is challenged to quantitatively understand the vegetation vulnerability under compound dry-hot extremes. In this study, a vine-copula conditional probability (VCCP) model is proposed to quantify the impacts of dry-hot events on vegetation dynamics, where the dependence patterns of the Normalized Difference Vegetation Index (NDVI), standardized precipitation evapotranspiration index (SPEI), and standardized temperature index (STI) are modelled through vine copula functions. The VCCP model can evaluate the conditional probability of vegetation loss under multiple dry-hot events and reveal the temporal and spatial patterns of vegetation vulnerability of different land-use types. Then, the VCCP model is applied to Xinjiang province, where the ecological environment is fragile and soil erosion is serious. The dependence patterns among NDVI, SPEI and STI in summer season (June-August) during 1983-2015 are identified. The main findings are: (i) spatial and temporal responses of vegetation to drought and hot events present distinctively; (ii) under the extreme scenario, the average probability of vegetation loss below the 50th percentile in August reach 58.2%, followed by July (with 44.0%) and June (with 33.1%); (iii) the northern and southwestern regions of Xinjiang (especially for the grassland in the mountain areas) have the worst resistance to extreme dry-hot events in summer season. The findings can provide insights into the impacts of compound extremes on vegetation conditions and help decision makers take effective and efficient ecosystem management to mitigate climatic disasters.

Introduction

Climate change has aggravated the frequency of weather and climate extreme events (e.g., drought and heatwave) around the world (Trenberth et al., 2014; Touma et al., 2015; Le et al., 2020). Increasing extreme drought and high temperature would largely influence the vegetation growth conditions, such as declining soil moisture and increasing evapotranspiration. The weakened vegetation growth conditions would cause irreversible damage to the structures and functions of terrestrial ecosystems, such vegetation photosynthesis (Mohammat et al., 2013) and nutrient cycling (Cremonese et al., 2017; Berdugo et al., 2020). Consequently, vegetation coverage may tend to decline and bare ground would expend. Such influence may be further aggravated by compound dry-hot extreme events that are caused by the conditional dependence between drought and high temperature. Therefore, vegetation vulnerability under extreme dry-hot events, especially in semi-arid and arid regions, has aroused worldwide attentions. Quantifying the vegetation's response to compound dry-hot extreme events is of great significance for further understanding of the vegetation's response mechanism to climate extremes and formulating effective management for natural ecosystems to climate change.

Over the past decades, a number of studies related to impact assessments of drought and/or hot extremes on vegetation growth have been conducted based on deterministic and probabilistic approaches (Bento et al., 2018; Guo et al., 2018; Zhang and Zhang, 2019; Karimi et al., 2020). The impact of drought and hot extremes on vegetation is generally analyzed through integrating indices of drought (e.g., standardized precipitation index, SPI; standardized precipitation evapotranspiration index, SPEI), temperature (standardized temperature index, STI) and vegetation (standardized difference vegetation index, NDVI) (McKee et al., 1993; Vicente-Serrano et al., 2010; Zscheischler et al., 2014; Zellweger et al., 2019). For example, Xu et al. (2018) discussed the response of NDVI to SPEI under different climatic conditions and land cover types, results would enhance understanding the dominant impacts of different vegetation types to drought scales. Corona-Lozada et al. (2019) used the standard major axis regression to analyze the effects of extreme events on terrestrial ecosystems in the French Alps; results indicated that summer water balance played a key role in vegetation response to heatwaves in mountain grasslands. Jha et al. (2019) evaluated the conditional probabilities of vegetation drought impacted by precipitation, temperature and soil moisture in India with the copula method, results disclosed that the cropland vegetation types were most susceptible to drought events. Generally, most of the previous studies focus on the relationship between vegetation and extreme indicators (drought or hot extremes) individually, without jointly considering their interactive effects.

Recently, the copula method was increasingly employed to model dependencies between extreme events (Chen et al., 2019; Hao et al., 2020; Poonia et al., 2021) and to assess the likelihood of vegetation activity corresponding to extreme events, mainly by establishing joint distribution of bivariate variables (Fang et al., 2019a; Das et al., 2020; Jha et al., 2020). However, due to the heterogeneous dependence among different variable pairs, the multi-parameter copulas are often inflexible in the high-dimensional system modeling, which makes the copula unable to flexibly construct the dependent structures of compound events (Bevacqua et al., 2017). Encouragingly, the vine-copula approach, which decomposes the complex dependent structures into bivariate copulas through pair-copula constructions (PCCs), provides greater flexibility in modeling general high-dimensional systems compared to multi-parameter copulas. Therefore, a potential vine copula approach can better fill the knowledge gap in quantifying the vegetation response through high-dimensional models.

Xinjiang province, as an important region along the Belt and Road in China, is experiencing drought and hot events in recent years, which is the main driving for vegetation degradation. Recent research works have paid efforts in exploring the impacts of drought and hot extremes on soil erosion, land degradation and ecosystem damage for Xinjiang province (Zhang et al., 2015; Li et al., 2018; Yao et al., 2019). These studies are devoted to provide useful information for local ecosystem maintenance and improvement. For example, Hua et al. (2017) explored the relationships between the drought occurrences and vegetation activity in the growing seasons using correlation analysis method, results showed that the occurrence of drought events had significant impacts on vegetation activity during the growing season in semi-arid regions. Li et al. (2018) jointly studied the droughts and hot extremes in northwest China by analyzing the spatial trend of SPEI, results showed that uneven humid trend led to the change of the spatial distribution and trend of droughts and hot extremes in northwest China.

Generally, most of the previous studies mainly focused on the qualitative impact analysis of extreme events on vegetation ecosystems individually; however, they have difficulties in quantitative assessing the impacts of compound extremes (e.g., dry-hot extremes) on vegetation vulnerability. Through the proposal of copula method could model the dependencies between two variables (e.g., NDVI and STI), compound events with high-dimension interactions may beyond the functions of conventional copulas. Besides, as an important strategic region, few studies have been found devoting in model the complex relationships among vegetation and compound dry-hot extremes with vine-copula method. In this regard, it is becoming increasingly urgent to quantitatively assess the potential impacts of compound dry-hot extremes on vegetation growth in Xinjiang.

Therefore, the objective of this study is to develop a vine-copula conditional probability (VCCP) model to analyze the impacts of compound dry-hot extremes on vegetation dynamics. Based on the copula conditional probability method (i.e. pair-copula decompositions), a high-dimensional dependency relationship among the vegetation dynamic and compound extremes is constructed using SPEI, STI and NDVI, respectively. The proposed VCCP model can (i) build the dependence structure of SPEI, STI and NDVI by the pair-copula constructions, (ii) acquire probabilistic response of vegetation loss conditioned on the compound dry-hot extremes, and (iii) disclose the impacts of combined extreme events on different vegetation land-use types. The proposed model is applied to Xinjiang province, where the spatial patterns of correlations among SPEI, STI and NDVI are recognized in the summer season (i.e. June, July and August) and, then, joint distributions of the three indices are established based on the PCC theory. Results will help to enhance the understanding of the climatic extreme impacts on vegetation loss, thus supporting effective management for natural ecosystems to climate change.