Comparison between canonical vine copulas and a meta-Gaussian model for forecasting agricultural drought over China

Abstract

Abstract. Agricultural drought mainly stems from reduced soil moisture and precipitation, and it causes adverse impacts on the growth of crops and vegetation, thereby affecting agricultural production and food security. In order to develop drought mitigation measures, reliable agricultural drought forecasting is essential. In this study, we developed an agricultural drought forecasting model based on canonical vine copulas in three dimensions (3C-vine model) in which antecedent meteorological drought and agricultural drought persistence were utilized as predictors. Furthermore, a meta-Gaussian (MG) model was selected as a reference to evaluate the forecast skill. The agricultural drought in China in August of 2018 was selected as a typical case study, and the spatial patterns of 1- to 3-month lead forecasts of agricultural drought utilizing the 3C-vine model resembled the corresponding observations, indicating the good predictive ability of the model. The performance metrics – the Nash–Sutcliffe efficiency (NSE), the coefficient of determination (R2), and the root-mean-square error (RMSE) – showed that the 3C-vine model outperformed the MG model with respect to forecasting agricultural drought in August for diverse lead times. Moreover, the 3C-vine model exhibited excellent forecast skill with respect to capturing the extreme agricultural drought over different selected typical regions. This study may help to guide drought early warning, drought mitigation, and water resource scheduling.

Predicting Hydrological Drought With Bayesian Model Averaging Ensemble Vine Copula (BMAViC) Model

“…As shown in Figure 5b and Figure S2b in Supporting Information , with the increased lead times, the skills of the MG model and C‐vine model (i.e., the best C‐vine model selected from the six different C‐vine structures via the smallest AIC) for the hydrological drought prediction tended to deteriorate at each selected hydrological station. Intuitively, the MG model had a poor performance for the extreme drought prediction and possessed a delay effect during some periods (Hao et al., 2016; Wu et al., 2022), especially on the 2–3‐month lead SSFI predictions. For instance, compared with the BMAViC model, a pronounced time delay between the 2‐ and 3‐month lead SSFI predictions via the MG model and SSFI observations was witnessed during the 198901–199006 and 199106–199302 periods (the pink shadow rectangles in Figures 5a and 5b), where these SSFI predictions based on the MG model visibly deviated from observations.…”

“…In view of the meta‐Gaussian (MG) model performing the powerful ability in drought prediction in comparison with the persistence‐based or random forecast models (Hao et al., 2016; Wu, Su, Zhang et al., 2021; Wu et al., 2022), we compared the BMAViC model with the MG model for predicting the hydrological drought with the 1‐ to 3‐month lead times (i.e., l = 1‐ to 3‐month). More details about the MG model can be found in Hao et al.…”

“…Compared with physically‐based or dynamical‐based approaches that require much field data (not easily accessible) as input, statistically‐based methods, analyzing physical mechanisms and causal interactions among associated variables, can model statistical relationships from historical observations (Liu et al., 2021; Wu, Su, Singh, et al., 2021; Wu et al., 2022; Yan et al., 2017). Numerous statistical methods, for example, machine learning, persistence‐based, and multiple linear regression, have been extensively utilized for hydrologic predictions (AghaKouchak, 2015; Belayneh et al., 2014; Shamshirband et al., 2020; Wu, Su, Singh, et al., 2021; Xu et al., 2021).…”

“…Accordingly, each vine copula with specific ordering variables may correspond to a unique prediction (Liu et al., 2021). The best vine copula form determined among different vine structures may discard some valuable prediction information (Bevacqua et al., 2017; Liu et al., 2021; Wang et al., 2019; Wu, Su, Singh, et al., 2021; Wu et al., 2022). Therefore, it seems appropriate to investigate as to how to couple these diverse structures of vine copula to facilitate drought prediction.…”

Status and prospects for drought forecasting: opportunities in artificial intelligence and hybrid physical–statistical forecasting

“…To address these premises, statistical drought models use a plethora of predictors, representing the underlying processes [83] or relationship between drought-related variables or drought features (e.g. [84][85][86][87][88][89]). While earlier studies were mostly reliant on the persistence properties of drought indicators [32,33,90], recent studies merge information from initial land conditions and climate/weather data (i.e.…”