Skip to main content
eScholarship
Open Access Publications from the University of California

Using Imaging Spectrometry and Thermal Imagery to Study Agriculture During Severe Drought in California’s Central Valley

  • Author(s): Shivers, Sarah Wells
  • Advisor(s): Roberts, Dar A
  • et al.
No data is associated with this publication.
Abstract

In California, predicted climate warming increases the likelihood of extreme droughts. As irrigated agriculture accounts for roughly 80% of the state’s managed water supply, agriculture simultaneously shows high vulnerability to a warming climate while also offering the greatest opportunity to mitigate the impact of future droughts through adaptation strategies. This dissertation took advantage of hyperspectral Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and thermal MODIS/ASTER Airborne Simulator (MASTER) imagery collected by the HyspIRI Airborne Campaign over three years of a multi-year drought to measure agricultural response to changes in water availability and highlight the potential of remote sensing to aid in future agricultural management during water-limited times. In Chapter 2, I used the Random Forest classifier to classify crops into categories of similar water usages, assess the accuracy at which hyperspectral imaging could be of use for agricultural mapping, and evaluate changes in crop plantings with drought. The results showed overall field-level accuracies of 94.4% with AVIRIS, as opposed to 90.4% with Landsat OLI and 91.7% with Sentinel, indicating that hyperspectral imagery has the potential to identify crops by water-use group at a single time step at higher accuracies than multispectral sensors. With these classifications, I quantified shifts in crop plantings over drought and found a prioritization of high-value, perennial crops over lower-valued annual plantings. In Chapter 3, I used paired AVIRIS and MASTER imagery to assess crop stress in fruit and nut orchards using an approach that accounted for thermal complexity within the landscape, without the need for ancillary environmental measurements. I used surface characteristics obtained by AVIRIS data to model expected pixel temperatures. These temperatures were compared to measured temperatures as a way of assessing crop stress. Crop species showed significantly different temperature distributions independent of fractional cover (F(10, 33135) = 735, p < 0.001 for 2013, with similar results in 2014 and 2015). Crop residuals were found to increase as the drought progressed with average residuals of 0.14°C in 2013, 0.97°C in 2014, and 1.1°C in 2015. In addition, crops with the highest LST residuals had the largest reductions in yield during the study period. In Chapter 4, I analyzed spatiotemporal patterns of water vapor as they varied across a diverse and dynamic agricultural scene. I tested a set of hypotheses to better understand surface-atmosphere interactions and their ability to be evaluated through AVIRIS-derived column water vapor estimates. Results showed a positive correlation between crop water use and the frequency with which that crop showed directional alignment between wind and water vapor (r=0.42). We also found patterns of water vapor across the region that support advection of moisture across the scene. Results conclude that accumulation of water vapor above fields in these scenes is observable with water vapor imagery while advection at the field level is obscured by variable winds, differences in field structure and composition, and smaller-scale patterns of vapor. Overall, this dissertation enhances scientific and social ability to monitor and assist in food and water resource management with remote sensing during drought.

Main Content

This item is under embargo until October 26, 2019.