ITC analysis of satellite images

• The individual tree crown approach applied to IKONOS images of a coniferous plantation area
• Individual tree crown delineation on Ikonos and QuickBird imagery: the Cockburn Island study.
• Specialized forest inventory from an IKONOS image
• ITC species classification and regrouping with an IKONOS image

---

Specialized forest inventory from an IKONOS image

Abstract

The intensification of forest management implies the acquisition of information with ever increasing precision. Fortunately, research done using high spatial resolution (10-100 cm/pixel) aerial images allow us to think in terms of semi-automatic individual tree crown species identification and stand delineation on an operational basis in the no-so-distant future. The recent availability of the IKONOS satellite (1 m/pixel) also opens new horizons for forest inventory. Within this framework, a research project was carried out to evaluate the potential of white pine detection from IKONOS images and to compare the efficiency of the ITC (Individual Tree Crown) suite on pan-sharpened and on separate panchromatic/multispectral IKONOS images.

The panchromatic (1m/pixel), multispectral (4m/pixel), and pan-sharpened (1m/pixel) versions of an IKONOS scene of a region in the Quebec North-West (46°N, 77°W) were acquired and analysed with the ITC suite. After smoothing the image and creating masks of the non-forested areas , the individual tree crowns were delineated by following the valleys of shade between them. Signatures were created for white pine and a few other species and then, crowns were classified individually by a maximum likelihood process. In spite of haze over a significant part of the scene, the white pine inventory appears remarkably precise when compared with the interpretation of infrared aerial photographs and with ground plots. The use of the pan-sharpened image does not appear to hinder crown classification and that of separate panchromatic and multispectral images, although more costly, brings additional precision to stem counts.

Acknowledgements

This work was done in collaboration with CLC-Camint (Gatineau) and Industries Davidson Inc. and was funded in part by the Programme de mise en valeur des ressources du milieu forestier - Volet 1 of the Quebec Department of Natural Resources.

Reference:

Gougeon, F.; Labrecque, P.; Guérin, M.; Leckie; A.; Dawson, A. 2001. Détection du pin blanc dans l'Outaouais à partir d'images satellitaires  à haute résolution IKONOS. In Proc. 23rd Canadian Symposium on Remote Sensing / 10e Congrès  de l'Association québécoise de télédétection (CD-ROM), Sainte-Foy, Québec, Canada, August 21-24, 2001.

Figure 1. Polygons from the current Quebec government forest inventory superimposed on a section of the pan sharpened IKONOS image (near infrared colour style). One can easily recognize the stands made mostly of deciduous trees in pink and those of conifers in darker shades. One also sees lakes with some cloud reflections and their often associated swamp zones (brown tones). Note that the alignment of these polygons with the image is better in some areas than other.

Figure 2. Trees from the IKONOS image section of Figure 1, as delineated and classified by the ITC Suite. Classes are the following: white pine (in red), other conifer (in blue), yellow birch (in yellow), other deciduous (in green), and regeneration (in light green). Swamps (orange) and lakes (light blue) come from the base map, with minor repositioning.

Classification results:

Table 2. Accuracy by vegetation classes for the multispectral and pan-sharpened datasets with and without haze

	Multispectral without haze	Multispectral with haze	Pan sharpened without haze	Pan sharpened with haze
White pine	70%	70%	64%	59%
Other conifers	78%	73%	77%	75%
Yellow birch	42%	62%	40%	69%
Other deciduous	63%	93%	69%	60%
Regeneration	82%	79%	86%	59%

---