Satellite observations of forest resilience to hurricanes along the northern Gulf of Mexico

Highlights

• Forest resilience to hurricanes along the northern Gulf of Mexico were evaluated.

• NDII is superior in monitoring the forest resilience to hurricanes.

• Wind speed is the leading factor that affects the forest resilience to hurricanes.

• Woody wetland is more resistant to hurricanes than evergreen forest.

• Drought conditions substantially extend the post-hurricane recovery of forests.

Abstract

As one of the most destructive natural disasters, hurricanes pose a great threat to forest ecosystems, particularly in the coastal areas. A better understanding of forest resilience to hurricane disturbances is essential for reducing hazard risks as well as sustaining forests in a time of increasing climate disasters. Although hurricane-induced forest damage has been extensively studied at both local and regional levels, the lack of large-scale assessment of post-hurricane recovery still limits our understanding of forest resilience to hurricane disturbances. In this study, we utilized four remotely sensed vegetation indices (VIs), including the normalized difference infrared index (NDII), enhanced vegetation index (EVI), leaf area index (LAI), and solar-induced chlorophyll fluorescence (SIF), to examine the forest resilience to hurricanes of different strengths by quantifying the resistance, net change, and recovery of the forest after hurricanes that made landfall along the northern Gulf of Mexico from 2001 to 2015. The results revealed that the NDII was superior in monitoring the large-scale forest resilience. SIF exhibited a performance similar to that of the EVI. Wind speed was found to be the leading factor affecting forest damage and post-hurricane recovery. The impacted forest canopy began to recover approximately one month after the landfall. Woody wetlands exhibited less VI reduction and shorter recovery time than evergreen forests for the same category of hurricanes. For regions dominated by evergreen forests, NDII values lower than the multi-year average were observed across all seasons during the year after being impacted by a major hurricane. The widespread drought of 2006/2007 has aggravated the VI decrease and substantially extended the recovery period after hurricanes Ivan and Katrina. Overall, our findings derived from satellite observations provide essential information for understanding forest resilience to hurricanes as well as implementing efficient post-hurricane forest restoration.

Introduction

Hurricanes, one of the most powerful and destructive climate disasters, pose great threats to both human and natural systems, especially along the coastal regions of the United States (Wang et al., 2010, Wang and D’Sa, 2010). The 2005 Atlantic hurricane season is considered to be a record-breaker in terms of the intensity and frequency of hurricanes as well as the catastrophic total damage of US$74 billion (Weinkle et al., 2018, Beven et al., 2008). Terrestrial ecosystems, especially forests, are deemed most vulnerable to frequent hurricane damage, particularly along the Gulf of Mexico (McNulty, 2002, Negrón-Juárez et al., 2014, Wang and D’Sa, 2010). The substantial impacts of hurricanes on forests include defoliation, the bending or breaking of branches, and the blowdown or even uprooting of entire trees (Boose et al., 1994, Boose et al., 2004). The intensity of these impacts usually depends on biotic and abiotic factors, such as wind strength, topography, forest type, forest density, and the distance of the forest from the hurricane track (Dahal et al., 2015; McNulty, 2002). Alterations of the composition and structure of coastal ecosystems resulting from hurricanes definitely affect the carbon and nitrogen cycles at the landscape and regional scales, owing to powerful winds, heavy rainfall, and subsequent flooding (McNulty, 2002, Chambers et al., 2007). From 1850 to 2000, hurricanes caused carbon release from the continental U.S. forests at the rate of 29 Tg/yr (Zeng et al., 2009). In 2005, Hurricane Katrina alone produced 105 Tg of carbon emissions, the magnitude of which is comparable to the total U.S. forest carbon sink (Chambers et al., 2007). In addition, the hurricane-driven flooding events cause an intrusion of salty water into coastal wetlands and inland watersheds, which can destroy marsh vegetation and increase release of greenhouse gases (Vidon et al., 2017, Wang and D’Sa, 2010).

The timely and accurate identification of post-hurricane damage is essential for the ability of land managers and government officials to take immediate protection and to guide the post-hurricane restoration (Stanturf et al., 2007, Wang et al., 2010, Wang and D’Sa, 2010). The existing estimation of post-hurricane impacts generally includes ground observation and remote sensing-based methods. Ground or aerial surveys can provide detailed information on forest damage along the track of hurricanes (Boutet and Weishampel, 2003, Chambers et al., 2007, Imbert, 2018). However, field observation-based studies are often constrained to small spatial scales due to the limited forest inventory plots and the resources consumed. The development of remote sensing technology has facilitated the quantification of post-hurricane damage at extended spatial and temporal scales (Dahal et al., 2015, Hu and Smith, 2018, Wang and D’Sa, 2010, Zeng et al., 2009). The Moderate Resolution Imaging Spectroradiometer (MODIS) products have frequently been used to assess of hurricane damage (Dahal et al., 2015, de Beurs et al., 2019, Negrón-Juárez et al., 2014, Potter, 2014, Wang et al., 2010, Wang and D’Sa, 2010). Vegetation indices, such as the normalized difference vegetation index (NDVI) (Ayala-Silva and Twumasi, 2004, Ramsey et al., 1998, Ramsey et al., 2001), enhanced vegetation index (EVI) (Rogan et al., 2011, Wang and D’Sa, 2010), tasseled cap water index (TCWI) (Mostafiz and Chang 2018), normalized difference infrared index (NDII) (Aosier et al., 2007, Dahal et al., 2015, Wang et al., 2010), are widely used to detect the forest canopy change by comparing the difference between pre- and post-hurricane status. In addition, Potter (2014) performed a global analysis concerning the damage to coastal ecosystem vegetation resulting from tropical storms between 2006 and 2012 based on a MODIS quarterly indicator of cover change. Most of these extensive studies, however, were narrowly focused on a single event or were based on a single indicator. In order to accurately quantify forest resilience to hurricane disturbances, as well as satellite-observational uncertainties, it is important to use multiple indicators and detect their performances on hurricanes of different intensities.

Compared with the large amount of research focusing on the assessment of post-hurricane forest damage, the trajectory of recovery has received less attention. Ground surveys of post-hurricane recovery are conducted by comparing the composition and physical parameters of trees pre- and post- hurricane (Burslem et al., 2000, Imbert, 2018). The lack of long-term post-disturbance data has made it difficult in monitoring the recruitment process of forests at a larger spatial scale (Bellingham et al., 1995, Burslem et al., 2000). Imbert (2018) found that mangroves in inner, tall-canopy stands need 23 years to recover from a hurricane disturbance, while the fringe and scrub stands require even more time. Satellite monitoring facilitates capturing the damage extent and the long-term pattern of forest recovery at large spatial scales. Wang and D’Sa (2010) assessed the utility of MODIS EVI in detecting the forest damage and monitoring the forest recovery after hurricanes Katrina, Rita, Ike, and Lili. de Beurs et al. (2019) developed a MODIS-based disturbance index to detect the damage from hurricane and drought on four major Caribbean islands since 2001. These efforts have improved our understanding concerning the post-hurricane recovery of forests, but it is still unclear how long a forest needs to recover from the damage induced by hurricanes, or whether different forest types have varied recovery speeds.

The frequent hurricanes in the Gulf of Mexico have repeatedly exposed the forests to strong winds, floods, heavy rain, landslides, and storm surges (Chapman et al., 2008, Wang and D’Sa, 2010). More than 250 hurricanes have struck the northern Gulf of Mexico since 1851, and nearly one-third of them were categorized as major hurricanes (Blake et al., 2005). The coastal areas of the southern U.S. will be exposed to a greater risk of hurricanes over the next 40 years (Stanturf et al., 2007). Therefore, in order to better understand the impacts of hurricanes on forests, we quantified the forest resilience to hurricanes that made landfall along the northern Gulf of Mexico from 2000 to 2015 through the utilization of four remotely sensed vegetation indices (VIs), including the NDII, EVI, LAI, and solar-induced chlorophyll fluorescence (SIF). Specifically, the objectives of this study were to: (1) quantify the resistance (how much a VI changes during a hurricane) and net change (how much the VI changed compared to the state before hurricane) of forest after each hurricane over the study period; (2) estimate the seasonal and interannual variations of VIs during the post-hurricane period; and (3) assess how long the damaged forests take to recover from a hurricane event and determine the potential influencing factors. By comparing the results among the four VIs, we also attempted to identify an optimal indicator for characterizing the post-hurricane damage and recovery trajectory of forests.