Introduction

Hurricane Katrina made landfall in Mississippi near the outlet of the Pearl River on August 29, 2005, as a category 3 storm on the Saffir-Simpson scale (Knabb and others 2005). Loss of life and damage to property were catastrophic, as New Orleans was flooded, and many towns and cities along the Louisiana and Mississippi Gulf Coasts were destroyed or severely affected. Rural areas whose economy depends on agriculture and forest industry were devastated also. Accordingly, preliminary damage estimates obtained through aerial surveys of the affected region by the Mississippi Forestry Commission (MFC) exceeded $1 billion in damaged wood and timber stumpage. These estimates underscore the need for more continuous damage estimates that can be developed when remotely sensed, storm, and pre-existing thematic data are employed in the modeling process.

Moderate resolution remotely sensed data, from sources such as Landsat, have been used in modeling various forest parameters related to timber harvesting (Healey and others 2005), canopy closure (Butera 1986, Cohen and others 2001, Larsson 1993), and other forest attributes (Cohen and others 2001, Cohen and others 2003). Cohen and others (2001) modeled percent green canopy cover in a predominantly evergreen softwood region of the Pacific Northwest, similar to south Mississippi, with a coefficient of determination (R²) of 0.74 and a Root Mean Squared Error (RMSE) of 12 percent. Similarly, Healey and others (2005) used a series of transformations on independent variables as well as a natural logarithm transformation on percent cover, the dependent variable, in a series of simple linear regressions to determine adequate univariate models. The results of their work were promising with regard to using single Short Wave InfraRed (SWIR) bands, as well as the Normalized Difference Vegetation Index (NDVI) and Normalized Difference Moisture Index (NDMI) transformations described in the methods section of this work.

Damage assessments of past catastrophic tropical storms in the Southeast United States were not able to utilize the large number of image sources and processing techniques that are presently available. The use of Geographic Information Systems (GISs) and remote sensing techniques in these assessment activities were, for the most part, restricted to a minor role in the wake of Hurricane Hugo in South Carolina (Nix and others 1996) with an expanded role noted for Hurricane Andrew (Jacobs and Eggen-McIntosh 1993, Ramsey and others 1997, Ramsey and others 2001). With these two storms and the studies mentioned, a progression of technology and techniques can be noted. In Nix and others (1996), remote Hugo forest damage assessments were made in a GIS through aerial photo interpretation and digitization. Jacobs and Eggen-McIntosh (1993) also used visual image interpretation to perform assessments of Hurricane Andrew-induced damage with the imagery taking the form of airborne digital video frames. The two works led by Ramsey (Ramsey and others 1997, Ramsey and others 2001) show a final evolution to satellite acquired moderate (Landsat) and course (Advanced Very High Resolution Radiometer or AVHRR) resolution imagery, along with storm data, in identifying Andrew’s damage in a largely hardwood area in south Louisiana.