A Nonstationary Standardized Precipitation Index (NSPI) Using Bayesian Splines

• DOI: 10.1175/jamc-d-21-0244.1
• Publication Date: 2022-03-30T16:38:46Z
• Authors: James H. Stagge; Kyungmin Sung
• Keywords: 13. Climate action; 0207 environmental engineering; 02 engineering and technology; 15. Life on land; 6. Clean water

Abstract The standardized precipitation index (SPI) measures meteorological drought relative to historical climatology by normalizing accumulated precipitation. Longer record lengths improve parameter estimates, but these longer records may include signals of anthropogenic climate change and multidecadal natural fluctuations. Historically, nonstationarity has either been ignored or incorporated into the SPI using a quasi-stationary reference period, such as WMO 30-yr period. This study introduces evaluates novel nonstationary model based on Bayesian splines, designed both estimates for stationary climates explicitly incorporate nonstationarity. Using synthetically generated precipitation, this directly compares proposed with existing approaches maximum likelihood estimation climates. not only reproduced performance models improved upon them in several key areas: reducing uncertainty noise, simultaneously modeling zero positive capturing nonlinear trends seasonal shifts across all parameters. Further, fully approach ensures parameters have including likelihood. notes that is too sensitive could be future iterations. concludes an application nine gauges range hydroclimate zones United States. Results experiment show stable reproduces patterns identified prior studies, while also indicating new findings, particularly shape Significance Statement We typically measure how bad comparing it record. With long-term changes other factors, however, typical today recent past. purpose build changing climate. Our results confirm accurate captures previously noted patterns—a drier western States, wetter eastern earlier summer weather, more extreme wet seasons. significant because can measurement identify drought.