! 2004 
FORESTLAND CLASSIFICATION USING AN ECOSYSTEM APPROACH 
C. C. Cheng *. Y. K. Chen *. S. F. Wang “. J. FL Jan “ 
“Division of Forest Management, Taiwan Forestry Research Institute, Taiwan -cechengáóserv.tfri.gov.tw 
" Department of Forestry, National Taiwan University, Taiwan — vkchen/tntu.edu.tw 
* Department of Geography, National Changhua University of Education, Taiwan - sfwang/ce.ncue.edu.tw 
" Department of Land Economics, National Chengchi University, Taiwan — jfjan@pehome.com.tw 
Commission VII, WG VII/2 
KEY WORDS: Forestry. Management. Ecosystem. Landscape. DIM, GIS 
ABSTRACT: 
This studv. focuses on using an ecosystem approach for forestland classification of the Liukuei Experimental Forest of Taiwan 
Forestry Rescarch Institute. The content includes the delincation of ecosystem units using DTM, the development of’ a hierarchical 
ecosystem classification system using GIS and multivariate statistical analysis. the establishment of a forestland classification 
decision support system (DSS) and its application on site selection of a Taiwanese native. species — 
Taiwania (7aiwania 
cryptomerioides). Vhe results indicate that D'TM is a fast. easy. feasible. and automatic approach for delineating ecosystem units of 
different spatial scales. 
The developed hierarchical ecosystem classification is a satisfactory scheme for Liukuei's forestland 
classification because the developed scheme coincides with the terrain characteristics along a continuum. The established DSS can 
effectively and feasibly analyze forestland classification under different spatial scales. Meanwhile. the system can easily perform 
site selection for Taiwania. From the results. it is concluded that techniques such as DTM. GIS. and DSS are useful for forest 
managers in the reasonable planning of forestland classification and management practice. In addition. the ecosystem approaches 
obtained from the Experimental Forest will be extended to island-wide forestland classification in Taiwan. 
I. INTRODUCTION 
There is a growing consensus that ecosystem management is 
essential to achieve desired future conditions of sustainable 
forests (Salwasser ct al. 1992: Gregg 1994). The prerequisite 
process for ecosystem management is forestland classification 
using an ecosystem approach. To achieve forestland ecosystem 
classification. the determination of ecosystem. units and the 
development of a hierarchical ecosystem classification scheme 
become important and necessary tasks. — Several pieces of 
literature point out that watersheds can be treated as the basic 
ecosystem (Odum, 1969: Mather and Doornkanp. 1970: Omi et 
al. 1979: Lotspeich 1980). and there are two approaches for 
delincating watersheds: the manual and automatic approaches 
(O'Callaghan and Mark. 1984: Jenson and Domingue. 1988: 
Morris and Heerdegen. 1988: Cheng 1995). As for the 
hierarchical ecosystem classification scheme. several countries 
have proposed and implemented schemes for recognizing such 
scale levels (Salwasser. 1990). Among them. Miller (1978) 
proposed 1 scheme at 3 scales of perception (i.e. site. landscape. 
and ecoregion). Rowe and Sheard (1981) advanced a similar 
scheme. Bailey (1987. 1996) proposed a hierarchy of 
ecosystem. units and suggested that there are 5 methods for 
identifying ecosystems: gestalt. map-overlay. multivariate 
clustering. digital-image processing. and control factors. For the 
controlling factor method. many possible primary factors are 
apparent. such as vegetation. soils. physiography. and 
watersheds. 
In addition. the decision support system (DSS) that helps forest 
managers manage and assess forests has grown tremendously 
and is commonly used for many aspects of forest management. 
for example, to provide support in the complex process of 
problem formulation and task analysis: to make effective use of 
available data and knowledge bases: and to support rational use 
of the results (Bulger and Hunt. 1991: Jankowski. 1995: Mulder 
and Corns. 1995: Walker and Lowers. 1997: Reynolds. 1998: 
Rauscher. 1999; Varma et al.. 1999). As for the DSS. Reynolds 
(1998) described how the USDA Forest Service Pacific 
Northwest Research Station in Corvallis. OR had developed an 
ecosystem management decision support (EMDS) system for 
ecological assessment. Varma et al. (1999) proposed a DSS 
using a combination of lincar programming and GIS for 
formulating forestland use strategies to improve sustainability. 
Rauscher (1999) reviewed ecosystem management decision 
processes and the decision support systems available to 
implement them for federal forests in the United States. 
In Taiwan, ecosystem management is greatly being emphasized 
currently. and. the. use. of. ecosystem classification to assist 
ecosystem management is underway. lo establish this scheme 
as soon as possible. the Liukuei Experimental Forest of Taiwan 
Forestry Research Institute (TERI) was chosen as a study site. 
The forestland classification of the Experimental Forest was 
originally finished in 1995 (Cheng. 1993). The method treated 
watersheds as. an ecological unit and used multivariate 
statistical analvsis for grouping watersheds. — Although the 
method is better for understanding similarities and relationships 
among ecosystem. it is still limited because of manual 
delineation of watersheds and lack of a hierarchical 
classification scheme. For this reason. à further modification on 
the delineation of ecosystem units and the development of an 
ecosystem classification scheme are certainly needed. Therefore. 
this study focuses on using an ecosystem approach and 
techniques such. as DTM. GIS and multivariate statistical 
analysis for forestland classification. Furthermore. the EMDS 
was applied to establish a forestland classification DSS for a 
case study of Taiwania site selection. The objective was to 
extend the forestland ecosystem approaches obtained from the 
181