Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Doctorat → Chine → 2021 → Some Novel Probabilistic and Machine Learning Techniques Based Regional and Ensemble Approaches for Drought Monitoring

Jiangxi University of Finance and Economics (2021)

Some Novel Probabilistic and Machine Learning Techniques Based Regional and Ensemble Approaches for Drought Monitoring

Muhammad Asif Khan

Titre : Some Novel Probabilistic and Machine Learning Techniques Based Regional and Ensemble Approaches for Drought Monitoring

Auteur : Muhammad Asif Khan

Grade : Doctoral Dissertation 2021

Université : Jiangxi University of Finance and Economics

Résumé partiel
Drought hazard is one of the main consequential of global warming and climate change.Unlike other natural disasters,drought is considered an interwoven natural disaster composed by a number of different factors,as for example agricultural,meteorological or hydrological.It is occurring recurrently in various climatic zones around the world.Therefore,accurate and continuous drought monitoring are essential for reliable drought mitigation policies.In order to monitor and characterize drought conditions,using Standardized Drought Indices(SDI)is recently the most frequently used practice.In past research,several drought monitoring indicators have been developed.Regardless of their scope and applicability,every indicator has certain amount of error regarding accurate determination of drought classes.Consequently,accurate drought monitoring is a challenging task in hydrology and water management research.In this regard,this study includes five major proposals,including ;the use of improved precipitation records,continuous monitoring under global warming scenario,spatial and temporal characterization of drought related hazards.For instance ;this thesis proposes SDI based on adding auxiliary data such as regional weights in order to make monthly rainfall records more accurate in relation to the dependency characteristics of temperature and rainfall records under regression and product estimation settings.Subsequently,we propose an innovative method of hydrological drought evaluation,a so-called Regionally Improved Weighted Standardized Drought Index(RIWSDI).One of the aims of this thesis is to improve drought monitoring system by providing a comprehensive data mining approach under Principle Component Analysis(PCA)and Kcomponent Gaussian mixture distribution.Consequently,we propose a new aggregative index named : Seasonal Mixture Standardized Drought Index(SMSDI).In preliminary analysis,aggregation of three multi-scalar Standardized Drought Indices(SDIs)is made for three meteorological gauge stations of Pakistan.For comparative assessment,individual SDI has been used to investigate the association and consistency with SMSDI.Outcome associated with this research shows that the SMSDI have significant correlation with individual SDIs.We conclude that instead of using individual indicator,the proposed aggregative approach enhances the scope and capacity of drought indicators for extracting reliable information related to future drought.This research provides a novel framework for the assessment of drought at the regional level under the prevailing global warming scenario-The Multi-Scalar Seasonally Amalgamated Regional Standardized Precipitation Evapotranspiration Index(MSARSPEI).In MSARSPEI,we employed Boruta algorithm for feature selection to ensemble monthly time series data of evaporation observed in various meteorological stations located in certain regions.In application,we have considered seven regions containing fifty meteorological stations of Pakistan.We conclude that the choice of MSARSPEI is more appropriate for regional assessment of drought hazard under a global warming scenario.

Mots clés : Drought ;Standardized Precipitation Index ;Drought Indices ;Regional Drought ;Auxiliary Information ;Standardized Precipitation Evapotranspiration Index ;Standardized Precipitation Temperature Index ;Climate Change ;Drought Monitoring ;Water Management Research ;Correlation;Global Warming ;Boruta Algorithm ;K Component Gaussian ;Markov Chain ;Transition Probability Matrix ;

Présentation (CNKI)

Page publiée le 12 novembre 2021