Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Climate Data and Analytics for Maize Phenotype Predictability and Uncertainty Assessment
This dissertation aims to develop and implement a climate-analytics framework to improve maize yield predictability. The statistical genetic-by-environment (GxE) model is applied to identify how hydroclimate interacts with maize genetics' molecular markers through covariance matrix structure to improve the predictability of and propagate the uncertainty to maize yield simulations. A multi-year, multi-environment, and multi-dimensional database across the U.S. and Ontario in Canada is obtained from the Genomes to Field (G2F) initiative and preprocessed for the GxE simulations and the proposed analytical framework. The G2F environmental dataset (including temperature, dew point, relative humidity, solar radiation, rainfall, wind speed, wind direction, and wind gust time series during the maize growing season) is improved using deep learning analytics to fulfill the missing values. The improved environmental, genetic, and phenotypic data (OMICs) are integrated into the GxE-based prediction, and the skill enhancement is examined. The GxE performance applies three trial selection schemes (i.e., "random-based," "covariance-based," and "climate-based") to evidence how the hydroclimate variability contributes to the GxE model performance. The sensitivity of the GxE performance to the environmental covariance matrix constructed by hydroclimate drivers is analyzed by employing a global sensitivity analysis (GSA) method called PAWN. The quality and consistency control algorithms for consolidating a homogeneous and multi-dimensional database consisting of improved hydroclimate time series, OMICs, and metadata have been designed for crop phenotypes prediction following the FAIR principles. We can conclude that the proposed analytical frameworks improve GxE performance by enhancing the hydroclimate time series used by GxE simulations. The proposed GSA-GxE framework quantifies the sensitivity of GxE model performance to the covariant climate and molecular genetic markers, identifying the primary sources of uncertainty in phenotype predictability. A quality and consistency control preprocessing algorithms were developed for a multi-dimensional database consisting of hydroclimate time series, OMICs datasets, and metafile as the input of the GxE model. The homogeneous, integrated, and improved multi-dimensional database is released for other researchers interested in maize phenotypic predictability.
Water Resources Management|Agriculture
Sarzaeim, Parisa, "Climate Data and Analytics for Maize Phenotype Predictability and Uncertainty Assessment" (2022). ETD collection for University of Nebraska-Lincoln. AAI29166113.