National
Forest
Inventory
homepage- Executive summary
- General information
- History
- Materials
- Methods
- Results
- Resources
- Expertise
- Products
- Publications
- Applications of results
Current events/projects- Use of SAR (radar) data
- Imaging spectrometry
- General information
[Finnish]METLA
METLA Research
IMAGING SPECTROMETRY IN FOREST INVENTORY
The purpose of this project is to study the methods to use imaging spectrometers (especially the Finnish airborne spectrometer AISA (Airborne Imaging Spectrometer for Applications) as a component in the multisource forest inventory. METLA has aqcuired an AISA instrument. The project is funded by METLA and the Technology Development Centre (TEKES). The project spans the years 1995 - 1996.
What Is An Imaging Spectrometer?
An imaging spectrometer resembles a multispectral scanner because it makes a spatial image of the area covered by the instrument field of view. The difference is that the imaging spectrometer makes a hyperspectral image, i.e., an image consisting of a very large number of narrow spectral channels.
The Tasks in the Project
This project is the first project in METLA using the AISA spectrometer. This means that, in addition to data analysis, all of the infrastructure necessary for operating the instrument has to be built during the project.The first task is to study the characteristics of the instrument since this has not yet been done with sufficient detail. The radiometric and geometric accuracy and the stability of the instrument have to be determined as a basis for method development.
The effects of the atmosphere on the measured light signal have to be determined. The atmosphere and the light source (sun) are strong components in the measured spectra. This can be seen from a graph that shows a Lowtran7 simulation of the signal received from a target with constant spectral reflectance.
The effects of the atmosphere can't be completely removed from the signal but we have to find ways how to compensate the atmospheric variation.
The next step will be to test flights to collect material for algorithm development. The collected images will be examined both spectrally and spatially to see how different phenomena relevant in forest inventory manifest themselves in the images. This will form the basis for method development.
The algorithm development will concentrate on finding methods to estimate the forest parameters used in the NFI. In addition to this, methods for smaller scale inventories will be investigated.
The last step in this project will be to assess the impact of the developed methods in forest inventories of different scale.
The Current Status and Future Plans
After some practical problems the first test flight was performed on July 11, 1995. The flight was successful and provided about 1.6 gigabytes of raw data. The data has been radiometrically corrected but geometric processing has not yet been done. Another test flight was performed on August 18, 1995. Also this flight was succesful and about 5 gigabytes of new raw data was collected.The algorithm development will begin when the time-critical data collection tasks permit and the geometric preprocessing has been done. The first tasks will be to find interesting channel combinations based on supervised and nonsupervised pattern recognition methods. Another important task will be to find methods to utilize the good spatial resolution with reasonable computational load.
VKan, December 2000