Efficient Forecasting of Precipitation Using LSTM

Muhammad Salman Pathan, Mayank Jain, Yee Hui Lee, T. Alskaif, Soumyabrata Dev

Research output: Chapter in Book/Report/Conference proceedingConference paperAcademicpeer-review

9 Citations (Scopus)

Abstract

Precipitation is one of those many critical elements of the hydrological cycle that has a direct impact on human life in many aspects. An accurate and early detection of a future precipitation event can help in preventing human and financial losses. Therefore, it is vital to design a framework that can predict the precipitation with a significant accuracy. Accordingly, in this paper we have proposed a Long-Short-Term-Memory (LSTM) based forecast model which can predict the precipitation values efficiently using the historical values. The design of forecast model is made simple to avoid the heavy training time. Furthermore, a transformation technique was applied on the precipitation dataset to make the data more normal distribution-like. The proposed model was evaluated in terms of Root Mean Square Error (RMSE). The proposed model achieves a 6.186 RMSE and outperforms the traditional persistence and average forecast models.
Original languageEnglish
Title of host publicationProgress in Electromagnetic Research Symposium (PIERS)
EditorsJ. Au Kong, W. Cho Chew, S. He
PublisherIEEE
ISBN (Electronic)9781665409889
ISBN (Print)9781728172477
DOIs
Publication statusPublished - 2021
Event2021 Photonics & Electromagnetics Research Symposium (PIERS) - Hangzhou, China
Duration: 21 Nov 202125 Nov 2021

Publication series

Name Progress in Electromagnetic Research Symposium (PIERS)
ISSN (Print)1070-4698
ISSN (Electronic)1559-9450

Conference/symposium

Conference/symposium2021 Photonics & Electromagnetics Research Symposium (PIERS)
Country/TerritoryChina
CityHangzhou
Period21/11/2125/11/21

Fingerprint

Dive into the research topics of 'Efficient Forecasting of Precipitation Using LSTM'. Together they form a unique fingerprint.

Cite this