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

Accueil du site → Doctorat → Inde → 2020 → A Data Driven Approach of Fault Tolerant Reference Evapotranspiration Prediction for Irrigation Planning

Vellore Institute of Technology (VIT) 2020

A Data Driven Approach of Fault Tolerant Reference Evapotranspiration Prediction for Irrigation Planning

Abraham Sudharson Ponraj

Titre : A Data Driven Approach of Fault Tolerant Reference Evapotranspiration Prediction for Irrigation Planning

Auteur : Abraham Sudharson Ponraj

Université de soutenance : Vellore Institute of Technology (VIT) Grade : Doctor of Philosophy (PhD) 2020

Résumé
In India, semi-arid regions constitute close to 75% of the agricultural area. Availability of water in these regions is a major concern for increasing crop production. An ever-increasing water demand due to population increase and industrial development brings in a need for efficient management of the agricultural water resources, which is facing an alarming depletion rate. Reference evapotranspiration (ETo) plays a vital role in solving the issues like soil water balance, irrigation system and water supply in the agro-ecosystem by providing a sustainable water management in these water starved regions. Hence proper irrigation planning by matching ETo with active crop growth requirement leads to an improved water usage efficiency and thereby improving the crop yield.The ETo can be calculated by many empirical and non-empirical equations which depends on large amount of weather parameters.The air temperature, relative humidity, wind speed and solar radiation are the primary influencers of ETo.This research work contributes by creating a fault tolerant model for predicting the reference evapotranspiration ETo using the daily weather newlinedata like the air temperature minimum and maximum, relative humidity, wind speed and solarradiation.A model based on Gradient Boost Regression (GBR) was developed to predict ETo accurately,and their results were compared with multivariate linear regression and random forest models to evaluate its performance. An iterative imputation technique was developed to accommodate the missing data to enhance the gradient boost regression model for predicting ETo. In the newlinemodelling of ETo prediction, the Borrego Springs daily weather data were used, and it was further tested with Tamil Nadu Agriculture University (TNAU) Coimbatore weather data. The performance of the models with and without the proposed imputation techniques was evaluated.Further, the influence of soil temperature in predicting ETo was investigated

Présentation et version intégrale (Shodhganga)

Page publiée le 16 mai 2021