EEWR Brown Bag Seminar with Graduate Students Stephanie Debats and Bo Guo
Speaker: Stephanie Debats and Bo Guo, Graduate Students
Series: EEWR Brown Bag Seminars
Location: Engineering Quad E225
Date/Time: Friday, April 5, 2013, 12:00 p.m. - 1:00 p.m.
Stephanie R. Debats
Title: A Modeling Framework for Understanding Landscape-Scale Crop Variability in Dryland Agriculture
The global food supply is under greater pressure from increasing climatic variability and global population growth. Smallholder farmers practicing rainfed agriculture in dryland ecosystems are particularly vulnerable to food insecurity. The objective of this research is to understand the drivers of crop yield variability in dryland agriculture, using Zambia as a case study. In this project, we consider the role of landscape heterogeneity in determining crop yields. Agricultural fields under different farming methods in Zambia are hand-digitized from Google Earth images. Estimates of crop yield are derived from the integration of time series of 250-meter resolution Normalized Difference Vegetation Index (NDVI) images collected by the Moderate Resolution Imaging Spectrometer (MODIS). Results demonstrate that farming methods, and their relative positions within a landscape, experience differing levels of success for various climatic conditions. The relative likelihood of these climatic events and the spatial distribution of agriculture will aid in determining overall food security in a community. Ongoing work includes the development of a statistical machine learning field mapping algorithm and a process-based crop model. The main crop modeling approaches, criticisms, and challenges are examined to inform the development of a modeling framework for studying landscape-scale crop variability.
Title: Inclusion of Vertical Dynamics in Vertically-integrated Models for CO2 Storage
Mathematical models of different complexity are needed to answer a range of questions for geological sequestration of carbon dioxide (CO2). One category of the simplified models is based on vertical integration, which reduces the three-dimensional problem to two dimensions. Usually, these models assume that brine and CO2 are in vertical equilibrium (VE). This type of model is useful and accurate when the VE assumption is satisfied, which occurs when the buoyant segregation of CO2 and brine is relatively fast. But the VE assumption is often inappropriate for geological formations that have low permeability or heterogeneities in the vertical such that the two fluids do not reach vertical equilibrium for a long period of time. For the problems for which VE models are not applicable, we propose a new multiscale method that still uses the vertically-integrated framework but also includes the vertical dynamics of CO2 and brine. This multiscale algorithm involves two scales: the coarse scale involves a set of vertically-integrated two-dimensional equations while the fine scale is a onedimensional equation representing the vertical dynamics of CO2 and brine in each coarse grid. These two scales are coupled sequentially: the vertical fine scale receives information from the horizontal coarse scale and is solved as a dynamic problem at each time-step, then passes information back to the coarse scale. This approach relaxes the VE assumption that is commonly used in the vertically-integrated models while still maintains much of their computational advantages. We expect that these kinds of multiscale models can simulate a wide range of problems including those with significant vertical dynamics.