Qualifier Seminar by Mr Yaseen T. Mustafa
Dept. of Earth Observation Science
Bayesian Network Modelling for Forest Growth
A growing forest is carbon sink; in other words, it fixes more carbon through photosynthesis than the amount it releases via respiration. This fact has drawn attention to forests growth, where many model are constructed for measuring and observing the forests growth on the earth. Several models have been constructed and developed to predict and measure (scale) the growth in the forest, but they still inappropriate scale.
Also the satellite imagery have been used to monitor developments and changes of forests, which is considered the best and most accurate way in observation despite having some errors in the measures.
This work, therefore will address Bayesian network modelling for forest growth model; by using the time series satellite images data to achieve appropriate scale for the forest growth.
The main objective of this research is to construct a method to optimally combine the output of the physical model of the forest growth with remote sensing images (a series of satellite images). To achieve this goal, Bayesian network will be implemented in such away to combine the output from forest growth model and from satellite images in the first step.
After constructing Bayesian network model, it will be applied for a series of time. Also the accuracy indicator by which the forest growth represents actual conditions for the study area will be studied using the updating rules in Bayesian network. As well as the performance of the forest growth will be studied using merged information of different image sources of the satellite images data.
The study is taking into account the errors which may occur in reading satellite data (missing data) and this may be caused by environmental conditions such as cloud cover, cloud shadow and atmospheric pollutants such as smoke and haze, or it can be caused by sensor malfunction or orbital characteristics of the satellite.
Finally, we will put much effort on how our approach can applied in the real-situation and exactly when there are less available field data. Hence unavailable field data is one of the hard problems facing the real scientist work.
|Event starts:||Friday 24 October 2008 at 11:00|
|City where event takes place:||Enschede|
|Country where event takes place:||Netherlands|