Academic journal article The Geographical Bulletin

Improved Monitoring of Forest Disturbance and Succession Using an Optimized Satellite Image Index

Academic journal article The Geographical Bulletin

Improved Monitoring of Forest Disturbance and Succession Using an Optimized Satellite Image Index

Article excerpt


Identifying and mapping ecosystem disturbances is an important task made more efficient through the use satellite imagery, especially at regional to continental scales. Recently, Healey et al. (2005) proposed a new image-processing technique, which they call the Disturbance Index (DI), that highlights forested areas that have been recently disturbed. The DI has been successfully used to detect disturbance caused by insects (Deel et al. 2012; DeRose et al. 2011; Has et al. 2009; Wulder et al. 2006), logging and windfall (Dyukarev et al. 2011; Rick et al. 2010), armed conflict (Gorsevski et al. 2012), and to map continental-scale disturbances regardless of the cause (Masek et al. 2008). The DI has also been used to assess the effects of disturbance on the nitrogen cycle (Dyukarev et al. 2011; McNeil et al. 2007) and to detect disturbance caused by the construction of new Walmart locations (Potere et al. 2008).

The DI is simple to calculate and interpret, but possibly its most useful advantage is that it can be fitted to specific disturbances and locations using weighting coefficients (Healey et al. 2005). In this way, disturbances of interest can be highlighted while minimizing other differences that may be present in the image, such as those caused by atmospheric or climate change (Thayn 2013). This has not been assessed over time, as the forest recovers from the disturbance. In the current paper, we address the following key questions: (1) Do the weighting coefficients help the DI remain sensitive to disturbances even as recovery progresses, and (2) Does the weighted DI help reduce confusion between disturbed sites and areas that are naturally less verdant? We also began to assess qualitatively whether changes in the derived weights might reflect changes in biophysical conditions as succession progresses, although additional research and field work are required to fully assess this relationship.

Images collected by satellite remote sensing systems record relative measurements of solar energy reflectance, which can be used reliably as surrogates for vegetation condition. Kauth and Thomas (1976) and Crist and Kauth (1986) introduced a linear transform of satellite imagery, called the Tasseled Cap Transform (TCT), that converts satellite imagery into estimates of relative brightness (overall reflectivity), greenness (the amount of photosynthesis occurring), and wetness (including moisture held in soil or in foliage). Healey et al. (2005) recognized that brightness increases and greenness and wetness decrease in forested pixels that have been disturbed, and suggested that disturbed areas could be highlighted by subtracting the sum of the greenness and wetness values from the brightness value: DI = B - (G + W). DeRose et al. (2011) used this Disturbance Index (DI) to identify insect-caused mortality of Engelmann Spruce trees {Picea engelmannii) and noted that map accuracy increased with the severity of the disturbance.

Healey et al. (2005) suggested that weighting the TCT components (brightness, greenness, and wetness) before calculating the DI might improve its ability to detect forest disturbances. Weighting the TCT values would fit the DI to local conditions and accentuate the disturbance of interest while dampening other differences that may be present in the forest canopy or in the satellite data. Thayn (2013) used a weighted DI, which he called an optimized DI (DIopt), to map insect defoliation in deciduous forests in northern Wisconsin: DIopt = wiB - (w2G + W3W), where wi, W2, and W3 are weighting coefficients which are used to emphasize the TCT components that are more useful for detecting disturbance. Thayn found that disturbance detection was consistently better with the weighted DI than with the unweighted DI (2013).

In this paper, we report on a study of the recovery of the 1988 fires in Yellowstone National Park. The accuracy of disturbance detection was assessed quantitatively by comparing the results to a burn scar outline prepared and provided by the Yellowstone Spatial Analysis Laboratory (YSAL, 2013). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.