Return Home

Examining the Post-Fire Regrowth from the Bastrop Fire Using LandSat-5 TM imagery and OBIA Classification

Introduction

Wildfires are natural phenomena that affect humans and ecosystems. The overall impacts vary locally and globally, depending on the severity and extent of the fire. Over the decades, the frequency and severity of wildfire outbreaks have increased in the United States (Miller et al. 2009). These increases are closely correlated to rising temperatures and precipitation activity (Westerling et al. 2006, Holden et al., 2007). Monitoring wildfires and their post-fire impacts can help humans coexist with wildfire as well as aid in the post-fire recovery of the burned area.

A majority of wildfires in the United States are caused by humans, while some only occur naturally (NPS 2018). Factors such as weather, topography, and fuel availability influence wildfire behavior (Idaho Firewise, 2020). One concern of wildfire outbreaks is determining where the fire ignites. Ecologically, there can be positive benefits, such as stimulating seedling germination, resulting in increased vegetation growth (Bell et al. 1993). However, there are also short-term negative impacts, such as losses in forest structures and above-ground biomass. In the long-term, wildfires can alter ecosystems significantly and cause global effects. For example, post-burn soils can decrease nutrient availability and show an increase in erosion. These changes can thereby radiate negative effects throughout the entire ecosystem over time (Francos et al. 2018, Huagassen et al. 2003).

Proximity to humans and property is another wildfire concern. Irreversible damage can be inflicted on both private and public properties. Wildfires can also cause the potential displacement of fleeing populations. For a local government and homeowner, the associated costs with suppressing and extinguishing the fire, along with repairing the damages, can be expensive. To mitigate and minimize wildfire impacts locally, Moritz et al. suggest extensive wildfire research be applied to fire management decisions (Moritz et al. 2014).

Wildfire research has examined different temporal scales: before, during, and post-fire. Governments have used Geographic Information Systems (GIS) to determine which areas have a high fire risk (Chuvieco et al. 2009). Remote sensing is used to detect active fires and monitor the present activity. Sensors such as MODIS and Hyperion are useful for these purposes because of their ability to detect visible, mid-infrared, and thermal infrared wavelengths (Lentile et al. 2006).

Post-fire research is incorporated into many applications. Due to the extent of wildfires, monitoring, and assessing the affected area can be difficult. One method to evaluate post-fire effects is to employ in-situ surveying of the burned area. Surveying is an effective way to obtain accurate field measurements directly. Another technique is to indirectly survey the area by using remote sensing (Lentile et al. 2006). There are many multispectral sensors and spectral indices to monitor post-fire disturbance (Chen et al. 2014, Lentile et al. 2006). Commonly used indices are the normalized burn ratio (NBR), differenced normalized burn ratio (dNBR), and normalized difference vegetation index (NDVI). NBR, dNBR, and NDVI are metrics used to assess the fire severity and damages to vegetation (Lentile et al. 2006). The NBR is based on the short-wave infrared and near-infrared bands. It emphasizes burned areas post-burn (Lentile et al.). The dNBR takes the difference of the NBR pre-fire and post-fire to reveal where the burned occur and the severity of the fire. There are limitations in using these metrics, such as the inability to distinguish vegetation types and quantify impacts on vegetation. This can be especially useful in an affected area with a large proportion of forest cover.

Another method in remote sensing used to gain more detailed post-fire impacts is Light detection and ranging (lidar). Lidar is used to extract information such as volumes of the biomass burned, the extent of wildfire in the understory, and classifying burned objects (Nelson 2020). However, lidar can be costly and not a realistic or timely option for local governments looking to perform immediate post-burn analysis.

An alternative approach is to classify the object burned is using object-based image analysis (OBIA). OBIA has been effective at quantifying tree cover and can even identify select tree species with high-resolution imagery (Baena et al. 2017, Moskal et al. 2011). Additionally, it can incorporate multiple data types to accomplish classification. Specifically, it can include spectral properties and qualities such as texture, context, and edge detection. In contrast to lidar, OBIA can achieve classification without the high costs associated with lidar.

For these reasons, this study will focus on the post-fire assessment of the Bastrop complex fire occurring in 2011 using an object-based image analysis (OBIA) approach. The Bastrop complex fire devastated Bastrop county, an area with high forest land cover, in 2011. It started on September 04, 2011, and was officially declared over October 10, 2011, burning for at least one month. The objectives of this study are to quantify the post-fire disturbance on forest canopy cover (forest cover) by examining tree canopy cover using OBIA and comparing the change in the post-fire forest cover in 2012 to pre-fire forest cover in 2010. Lastly, forest recovery will be examined by comparing 2012 to 2018 tree canopy.

Return Home