David P. Berretta

Assessment of floodplain vulnerability during extreme Mississippi River flood 2011

Regional change in the variability and magnitude of flooding could be a major consequence of future global climate change. Extreme floods have the capacity to rapidly transform landscapes and expose landscape vulnerabilities through highly variable spatial patterns of inundation, erosion, and deposition. We use the historic activation of the Birds Point-New Madrid Floodway during the Mississippi and Ohio River Flooding of 2011 as a scientifically unique stress experiment to analyze indicators of floodplain vulnerability. We use pre- and postflood airborne Light Detection and Ranging data sets to locate erosional and depositional hotspots over the 540 km(2) agricultural Floodway. While riparian vegetation between the river and the main levee breach likely prevented widespread deposition, localized scour and deposition occurred near the levee breaches. Eroded gullies nearly 1 km in length were observed at a low ridge of a relict meander scar of the Mississippi River. Our flow modeling and spatial mapping analysis attributes this vulnerability to a combination of erodible soils, flow acceleration associated with legacy fluvial landforms, and a lack of woody vegetation to anchor soil and enhance flow resistance. Results from this study could guide future mitigation and adaptation measures in cases of extreme flooding.

On the Feasibility of Characterizing Soil Properties From AVIRIS Data

We evaluate the feasibility of quantifying surface soil properties over large areas and at a fine spatial resolution using high-resolution airborne imaging spectroscopy. Airborne Visible Infrared Imaging Spectrometer (AVIRIS) data collected by the National Aeronautics and Space Administration immediately after the large 2011 Mississippi River flood at the Birds Point New Madrid (BPNM, ≈ 700 km2) flood way in Missouri, USA, was used in a data mining lasso framework for mapping of soil textural properties such as percentages of sand, silt, clay, soil-organic matter, and many other soil chemicals constituents. The modeling results show that the approach is feasible and provide insights in the accuracy and uncertainty of the approach for both soil textural properties and chemical constituents. These models were further used for a pixel-by-pixel prediction of each the soil constituent, resulting in high-resolution (7.6 m) quantitative spatial maps in the entire floodway. These maps reveal coherent spatial correlations with historical meander patterns of Mississippi River and fine-scale features such as erosional gullies, represented by difference in constituent concentration, e.g., low soil organic matter, with the underlying topography immediately disturbed by the large flooding event. Further, we have argued and established that the independent validation results are better represented as a probability density function as compared with a single calibration-validation set. It is also found that modeled soil constituents are sensitive to NDVI and the calibration sample sizes, and the results improve with stricter (lower) NDVI thresholds and larger calibration sets.