Geofizika, Vol. 39 No. 1, 2022.
Original scientific paper
https://doi.org/10.15233/gfz.2022.39.9
Prediction of bimodal monsoonal rainfall in the central dry zone of Myanmar using teleconnections with global sea surface temperatures
Hiroshi Yasuda
; Organization for Educational Support and International Affairs, Tottori University, Tottori, Japan
Ayele Almaw Fenta
; Arid Land Research Center, Tottori University, Tottori, Japan
Hidetoshi Miyazaki
; Global Environmental Forum, Tokyo, Japan
Shun Ishiyama
; National Museum of Ethnology, Osaka, Japan
Koji Inosako
; Faculty of Agriculture, Tottori University, Tottori, Japan
Aung Din
; Nature Lovers International, Yangon, Myanmar
Takayuki Kawai
; Graduate school of International Resource Sciences, Akita University, Akita, Japan
Abstract
In the central dry zone of Myanmar, the mean annual rainfall is less than 1000 mm. Although rainfed agriculture is commonly practiced there, the feasibility of rainfed farming is compromised by the large fluctuations of rainfall and the frequent occurrence of dry years. The monthly distribution of rainfall follows a bimodal pattern. The intensity of the monsoonal rainfall from May to October is characterized by two peaks, an early peak (May-June) and a late peak (August–October), separated by the inter-monsoon (July). The return times of dry and wet years make management of rainfed agriculture problematic. There is very little correlation between the early and late monsoonal rainfall (r=–0.257). However, monsoonal rainfall is teleconnected to sea surface temperatures (SSTs) in certain areas of the Pacific Ocean in real time. Furthermore, at lag times of 6–9 months, there are teleconnections between the early monsoonal, inter-monsoonal, and late monsoonal rainfall and SSTs in certain areas of the Indian Ocean and Atlantic Ocean. We used an Elman artificial neural network model to predict early monsoonal, inter-monsoonal, and late monsoonal rainfall based on teleconnections with SSTs in the Indian and Atlantic oceans 6–9 months before the rainfall occurred. The correlation coefficient between the predicted and observed rainfall exceeded 0.7 in all three cases.
Keywords
artificial neural network; bimodal rainfall; inter-monsoon; Myanmar central dry zone; rainfed agriculture
Hrčak ID:
279988
URI
Publication date:
4.7.2022.
Visits: 1.226 *