Application of Artificial Neural
Network and Climate Indices to Drought
Forecasting in South-Central Vietnam

Nguyen, Luong Bang; Le, Manh-Hung

doi:10.15244/pjoes/105972

Cited by 5 publications

(2 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Artificial intelligence, particularly in the form of artificial neural networks, has shown promise for drought forecasting. Luong Bang Nguyen and J. Lee demonstrated the effectiveness of this technology for predicting drought indices and rainfall, respectively [26,27]. The use of climate indices as input variables in these models further enhances their accuracy.…”

Section: Introductionmentioning

confidence: 99%

Application of Machine Learning to Forecast Drought Index for the Mekong Delta

Duong,

Phong,

et al. 2024

Preprint

View full text Add to dashboard Cite

Droughts have a substantial effect on water resources, agriculture, and ecosystems on a worldwide scale. In the Mekong Delta of Vietnam, droughts exacerbated by climate change are significantly endangering the region's agricultural sustainability and output. Conventional forecasting techniques frequently do not capture the intricate dynamics of meteorological phenomena associated to drought effectively, prompting the exploration of more advanced methodologies. This work utilises artificial intelligence, par-ticularly machine learning methods like Gradient Boosting and Extreme Gradient Boosting (XGBoost), to enhance drought pre-diction in the Mekong Delta. The study utilises data from 11 meteorological stations spanning from 1990 to 2022 to create and evaluate Machine Learning models based on several climatic factors. We utilise Gradient Boosting and XGBoost algorithms to estimate the Standardised Precipitation-Evapotranspiration Index (SPEI) and evaluate their effectiveness in comparison to con-ventional forecasting techniques. The results show that Machine Learning, particularly XGBoost, surpasses traditional approaches in predicting SPEI accuracy at various time scales. XGBoost demonstrates skill in understanding the complex relationships between climatic factors, with R² values falling between 0.90 and 0.94 for 1-month forecasts. The progress highlights the potential of Machine Learning in improving drought management and adaptation tactics, proposing the incorporation of Machine Learning forecasting models into decision-making processes to enhance drought resistance in susceptible areas.

show abstract

Section: Introductionmentioning

confidence: 99%

Application of Machine Learning to Forecast Drought Index for the Mekong Delta

Duong,

Phong,

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Drought predictability is a current challenge for hydrometeorological sciences and it has also been assessed for some regions of Vietnam, such as the Khanhhoa Province Vietnam [46] or South-Central Vietnam [47]. Statistical techniques have also been used to evaluate the influence of different infilling techniques on the spatial distribution of the Standardized Precipitation Evapotranspiration Index (SPEI) in the South-Central Region of Vietnam [48].…”

Section: Introductionmentioning

confidence: 99%

Trends and Extremes of Drought Episodes in Vietnam Sub-Regions during 1980–2017 at Different Timescales

Stojanović

Liberato

Sorí

et al. 2020

Water

View full text Add to dashboard Cite

This study investigated the temporal occurrence of dry conditions in the seven climatic sub-regions of Vietnam during the 1980–2017 period. This assessment was performed using the Standardized Precipitation Evapotranspiration Index (SPEI) and the Standardized Precipitation Index (SPI) at 1 to 24 months timescales. Results show that the main periods of extreme drought occurred simultaneously throughout the country in 1992–1993 and 2003–2004, except for 2015–2016, when it was not identified in the southern region. In addition, a slight temporal lag was identified latitudinally (north–south) at the beginning of dry conditions, revealing the largest difference between the northern and southern regions. A positive trend in the time series of both indices (SPEI and SPI) prevailed in all sub-regions, with the SPEI minus SPI difference always being negative, suggesting the importance of temperature and evapotranspiration for this trend. Further detailed analyses were then performed using SPEI at 1-month and 12-months timescales for all climate sub-regions, as well as the main indicators to characterize duration and severity. Results show that the number of drought episodes did not vary much between regions, but they did vary in duration and severity at the annual scale. Moreover, changes in the soil root zone are largely associated with dry and wet conditions not only from season to season, but also in longer accumulation periods and more strongly in the northern regions of Vietnam. Indeed, a study of the most severe drought episodes also revealed the occurrence of negative anomalies of the root-soil moisture in the subsequent four or more months. Dynamic atmospheric conditions associated with the peak of most severe drought episodes show the crucial role of subsidence of dry air in the middle and high atmosphere, which prevents convection in the lower troposphere. Finally, the linkages between drought conditions in Vietnam and large-scale atmospheric and oceanic teleconnection patterns were revealed to be quite different among northern and southern sub-regions. During the positive phase of El Niño–Southern Oscillation (ENSO), drought episodes at different timescales were identified in the southern climate sub-regions, while the negative phase was associated with drought conditions in the northern regions.

show abstract