Nonstationary bayesian modeling of precipitation extremes in the Beijing-Tianjin-Hebei Region, China

Highlights

• Nonstationary frequency analysis can better capture the temporal evolution of precipitation extremes

• The best covariates for nonstationary modeling of precipitation extremes are dominated by the physical-based covariates

• Higher return period indicates higher uncertainty in expected return levels both for stationary and nonstationary models

Abstract

This paper investigates the nonstationarity of precipitation extremes by incorporating time-varying and physical-based explanatory covariates, using daily precipitation data across the Beijing-Tianjin-Hebei (BTH) region, China. We perform the stationary and nonstationary generalized extreme value (GEV) models based on the Bayesian framework to estimate the expected return levels of precipitation extremes with the 90% credible intervals. Results reveal that the nonstationarity of precipitation extremes is not prominently visible for the majority of sites in BTH. However, the nonstationary GEV models exhibit better performance to capture the variations of precipitation extremes by comparison to the stationary models based on four evaluation criteria. Further, this work attempts to determine the best covariate to illustrate the possible effects of environmental changes on the frequency analysis. Results indicate that the El Nino-Southern Oscillation (ENSO) is the top of the best covariates, followed by the East Asian summer monsoon, North Atlantic Oscillation (NAO) and local temperature anomaly. Moreover, the best covariates are dominated by the physical-based covariates, and the best models with nonlinear functions of covariates are found in the majority of sites. Finally, the best-fitted models are used to estimate the design values of return levels in precipitation extremes. Results illustrate that the differences between the stationary modeling and nonstationary modeling in the median condition of covariates are not significant for most of the sites. But the discrepancies will be enhanced if the covariates locate in a high (95-percentile) or low (5-percentile) value. Our findings suggest that the nonstationary modeling of precipitation extremes might prove more useful and reliable, especially in the uncommon conditions of physical-based covariates.

Introduction

The variations in extreme precipitation have been receiving more and more attention as a part of the changing trends in the global climate (Alexander et al., 2006; Easterling et al., 2000). The intensity and frequency of extreme precipitation have been varying with positive or negative trends in different regions (Aerenson et al., 2018; Dong et al., 2011). Simultaneously, the spatiotemporal distribution of extreme precipitation varies between regions in terms of their magnitude, intensity, and temporal distribution within a given time frame (Donat et al., 2016). As a consequence, there is a need to fully investigate the observed and projected changes of regional extreme precipitation, which is crucial for policymakers developing regional prevention schemes for precipitation-related natural disasters (Li et al., 2018; Zhao et al., 2019).

Frequency analysis based on extreme value theory is widely used to establish a relationship between extreme values and return periods. Generally, under the assumption of stationary, the return period analysis provides crucial information that is used in a wide range of applications, such as urban and rural development, public infrastructure, watershed management (Cheng et al., 2014). Traditionally, it is assumed that the probabilistic characteristics of hydro-meteorological processes will not change over time, and that future water resources planning can be designed with records (Sarhadi et al., 2016). However, climate change and human activities can generate trends in hydro-meteorological series, which may become nonstationary. Milly et al. (2008) asserted that nonstationary probabilistic models should be identified and applied because anthropogenic climate change is affecting the extremes of hydrologic variables. Afterward, various studies on nonstationary modeling were conducted to predict future events under changing environmental conditions, taking into account time-varying and physical-based covariates, such as time, temperature and climate indices (Cheng and AghaKouchak, 2014; Lu et al., 2019; Mondal and Daniel, 2019; Ragno et al., 2019; Sun et al., 2017; Um et al., 2017; Xiong et al., 2018).

Recently, changes in total and extreme precipitation over China have been examined (Chen et al., 2019; Xiao et al., 2017; You et al., 2011; Zhai et al., 2005), as well as the nonstationary analysis of precipitation extremes (Gao et al., 2016; Gu et al., 2019). In China, the climate is strongly influenced by the eastern Asian monsoon (Ding and Chan, 2005). Certainly, it is also affected by large-scale atmospheric circulation, such as El Niño Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), North Atlantic Oscillation (NAO), Indian Ocean Dipole (IOD). Considering that there are distinct discrepancies for each region in China with different climatic and geographical conditions, there is a need to fully investigate the changes in precipitation extremes and possible influence factors for different regions in China.

The Beijing-Tianjin-Hebei (BTH) region, as one of the metropolitan areas of China, suffers from various precipitation-related natural hazards (like extreme precipitation, urban flooding, flash flooding, and mudslide). Although previous studies have demonstrated that most of the precipitation extremes in this region have a decreasing trend during the past five decades (Mei et al., 2018; Song et al., 2019a; Zhao et al., 2019), few studies have focused on the frequency analysis of precipitation extremes under a changing environment, which may be important for the public safety and engineering design. Zhang et al. (2015a) investigated the nonstationarity of precipitation extremes over the BTH region with the climate indices as the external covariates based on 12 meteorological stations. However, only the use of climate indices as covariates may not be appropriate for the nonstationary modeling of precipitation extremes. To our best knowledge, the full understanding of the nonstationarity of precipitation extremes in BTH and their expected return levels under stationary and nonstationary conditions have not been conducted.

The objectives of this study are: (1) to know about the nonstationary behavior of precipitation extremes based on block maxima series; (2) to estimate the expected return levels of precipitation extremes under stationary and nonstationary conditions; (3) to determine the best covariates and models for frequency analysis of precipitation extremes; and (4) to quantify the uncertainties in return levels from stationary and nonstationary models. This study provides a protocol in terms of nonstationary precipitation behavior in regions of climatic demarcation between semi-arid and semi-humid climates with the decreasing trends in total precipitation and precipitation extremes. Besides, the results will be important for the understanding of precipitation extremes in a changing environment at a regional scale.