With a changing climate there is an increasing risk of more frequent and severe disturbance events causing damage to our forests. These disturbances can be events such as storms, insect attacks or forest fires. They affect the carbon balance as well as the economy. Hence, it is important to develop methods to efficiently monitor these disturbances. Lund Earth Observation group is focusing on methods to monitor damage caused by storms and insects. These events are generally different in nature since storms lead to abrupt changes in the landscape. The amount of damage can be estimated with change detection methods where images prior to, and after a storm, are compared. Insect attacks on the other hand are dynamic events with a gradual defoliation lasting for longer periods. An area may be infested by insects for several years with both the amount of defoliation and the spatial distribution changing from year to year. It is possible to apply change detection methods also for insect attacks, both to find damaged areas and to estimate the amount of damage. However, time series analysis of frequently obtained satellite images is more suitable since it enable monitoring of the dynamics in an attack.
Storms lead to sudden impact on forests; hence, a suitable approach to estimate damage caused by storms can be to use change detection techniques. By studying satellite images obtained prior to and after a storm it is possible to find areas that have changed. These changed areas are likely damaged by the storm, and can be used to estimate and map the damage caused.
The gradual development of insect attacks makes monitoring methods based on time series analysis suitable (see figure below). Satellite images with high temporal resolution, such as MODIS, make it is possible to estimate the state of the vegetation in a forest with weekly or bi-weekly intervals. The development of the vegetation can be studied during entire growing seasons for several years to decide the normal seasonal state of a forest. By comparing the normal state to present development, deviations can be identified. Disturbance monitoring can then be based on a set of rules, defining when a deviation is large enough to be considered a disturbance. To decide on these rules, and how to decide what the normal development is are some of the challenges with disturbance monitoring based on time series analysis. However, if these rules are well designed, it is possible to monitor both how defoliation is increasing during an attack and how the attack is spreading in space. This makes it possible to estimate the amount of damage, as well as studying the development of an insect attack. The high temporal resolution also makes it possible to detect disturbances in an early stage to develop early warning systems. We have used this methodology to study attacks of a new invasive insect in S. Sweden (Olsson et al. 2012). If you have Google Earth installed, then click on the links below to see an example of change detection:
- Forest damage in Scania, south of Sweden, 2009
- Forest damage in Scania, south of Sweden, 2010
Olsson, P. O., Jönsson, A. M., Eklundh, L., 2012: A new invasive insect in Sweden - Physokermes inopinatus: Tracing forest damage with satellite based remote sensing. Forest Ecology and Management, 285, 29-37.