A wavelet transform method to determine monsoon onset and retreat from precipitation time‐series

Scott Osprey,Lesley J. Gray,Jorge L. García‐Franco

doi:10.1002/joc.7130

Abstract

AbstractA new method to determine monsoon onset and retreat timings using wavelet transform methodology applied to precipitation time‐series at the pentad scale is described. The principal advantage of this method is its portability, since it can be easily adapted for any region and dataset. The application of the method is illustrated for the North American Monsoon and the Indian Monsoon using four different precipitation datasets and climate model output. The method is shown to be robust across all the datasets and both monsoon regions. The mean onset and retreat dates agree well with previous methods. Spatial distributions of the precipitation and circulation anomalies identified around the onset and retreat dates are also consistent with previous work and illustrate that this method may be used at the grid‐box scale, not just over large area‐averaged regions. The method is also used to characterize the strength and timing of the Midsummer drought (MSD) in southern Mexico and Central America. A two peak structure is found to be a robust structure in only in 33% of the years, with other years showing only one peak or no signs of a bimodal distribution. The two‐peak structure analysed at the grid‐box scale is shown to be a significant signal in several regions of Central America and southern Mexico. The methodology is also applied to climate model output from the Met Office Hadley Centre UKESM1 and HadGEM3 CMIP6 experiments. The modelled onset and retreat dates agree well with observations in the North American Monsoon but not in the Indian Monsoon. The start and end of the modelled MSD in southern Mexico and Central America is delayed by one pentad and has a stronger bimodal signal than observed.

Highlights

Monsoon regions are characterized by a strong contrast between the rainy and the dry seasons, with at least 70% of the total annual rainfall observed during summer in a monsoon (Zhou et al, 2016; Wang et al, 2017)
A current view is that monsoons can be defined based solely on the contribution of summer rainfall to the annual total highlighting the importance of rainfall to several societal sectors (Zhang and Wang, 2008; Wang et al, 2017)
The onset and retreat dates were determined for each year in each observed and model dataset for the Indian, North American and Midsummer drought (MSD) regions using the methods described in the previous section

Summary

Introduction

Monsoon regions are characterized by a strong contrast between the rainy and the dry seasons, with at least 70% of the total annual rainfall observed during summer in a monsoon (Zhou et al, 2016; Wang et al, 2017). Around the onset and retreat dates, changes in the predominant wind direction or strength cause abrupt changes in meteorological conditions such as rainfall, humidity and cloud cover (Zhou et al, 2016; Gadgil, 2018). A current view is that monsoons can be defined based solely on the contribution of summer rainfall to the annual total highlighting the importance of rainfall to several societal sectors (Zhang and Wang, 2008; Wang et al, 2017). The objective determination of onset and retreat dates is key for climate and weather research aiming to understand the short and long-term variability, trends and predictability of the regional monsoons (e.g., Kitoh and Uchiyama, 2006; Cook and Buckley, 2009; Htway and Matsumoto, 2011; LucasPicher et al, 2011; Nieto-Ferreira and Rickenbach, 2011)

Objectives

Results

Conclusion