198 
EXTENSIBLE LAND USE AND LAND COVER CLASSIFICATION FRAMEWORK 
DESIGN BASED ON REMOTELY SENSED DATA 
Wang Juanle 
Institute of Geographical Sciences and Natural Resources Research, Chinese Academy of Sciences, 100101, Beijing, 
China - wangjl@igsnrr.ac.cn 
KEY WORDS: Land use, Land cover, Classification system, Extensible model, ISO 19115, Remote sensing interpretation 
ABSTRACT: 
There are lots of land use/land cover classification systems in the world facing to different research targets and objectivities. And 
with the crossing of multi-disciplines and expanding of the research area, more and more land use/land cover classification systems 
are designed. This brings the difficulties for datasets’ exchange, integration and fusion which use different classification systems. 
The problem faced to land use/land cover classification systems are similar with the problem of metadata standards. ISO TC211 
designed an extension metadata standard framework for geographic data, and drew up the ISO Metadata Standard 19115. 
Enlightened by the ISO 19115 metadata extensible model, the paper brings forward the land use/land cover extensible classification 
framework. The framework is composed of 3 tiers, i.e., land use/land cover core classes, corpora classes and application profile 
classes. The core and corpora classes are derived from the main international land use/land cover classification systems in the world. 
Extension principle and methods are designed in this framework. This ensures many application classification profiles can be 
extended from the core classes and corpora classes easily. 
1. INTRODUCTION 
1.1 General Instructions 
The Land Use and Land Cover Change (LUCC) Project is a 
Programme Element of the International Geosphere-Biosphere 
Programme (IGBP) and the International Human Dimensions 
Programme on Global Environmental Change (IHDP). LUCC 
was launched in 1995 and now it has been one of the key 
research tasks in Global Environment Change Research area. 
Land use/land cover classification system is not only the basic 
of LUCC research, but also the interface for the datasets 
exchange, integration and sharing in different countries and 
regions. But, there are lots of land use/land cover classification 
systems in the world facing to different research targets and 
objectivities. These existed and being formed classifications 
have their own content, syntax and semantic structures. This 
brings the difficulties for datasets’ exchange, integration and 
fusion which use different classification systems. With the 
crossing of multi-disciplines and expanding of the research area, 
more and more land use/land cover classification systems are 
proposed. This adds the difficulties mentioned above. Face to 
the problem, this paper try to design an extensible land use/land 
cover classification framework, which will adapt to the change 
of various research subjects and research objectives. With the 
development of EOS, especially the development of GEOSS, 
remotely sensed data has been one of the most important data 
sources for LUCC research nowadays. The paper focuses on the 
extensible classification framework for the application which 
uses remotely sensed data. 
1.2 Land Use/Land Cover Classification System Progress 
There are many kinds of land use/land cover classification 
systems based on remotely sensed data sources in the world. 
Some popular ones used remotely sensed data sources are listed 
as follows. 
1. USGS land use/land cover classification system 
USGS land use/land cover classification system was designed 
in 1976. Because Anderson is its main designer, it is also called 
Anderson classification system. It is divided into 4 levels. Level 
I focus on the land use/land cover macro types and land 
resources natural & ecological background. This classification 
information can be acquired by remotely sensed data 
interpretation. Level II is the detail types of level I. This 
information can be acquired by the interpretation of Aerial 
photographs. Level III and IV is designed based on level II. 
Anderson classification system is used widely in America and it 
has important influence in the world. 
2. FAO/UNEP, IGBP and UMd classification system 
Land Cover Classification System (LCCS) was proposed in 
1996 by FAO/UNEP. LCCS actually is a decision tree which 
includes two phases, i.e., Dichotomous phase and Modular 
hierarchical phase. Although LCCS aims to promote its 
application widely, it is restricted because it pays more 
attention to the geographical special features and the 
classification types are defined differently in various extended 
systems. In 1990s, according to the global change research 
requirements, based on AVHRR remotely sensed data IGBP 
built the land cover classification system including 17 primary 
types, and University of Maryland built the UMd land cover 
classification system including 14 primary types. While, these 
two classifications are restricted in macro global land cover, 
they are not used in regional land use application with high 
resolution. 
3. Chinese Academy of Sciences land use classification 
system 
Chinese Academy of Sciences land use classification system 
was designed in 1991. This classification was used in “National 
Resources and Environment Macro Investigation and Dynamic 
Change Research based on Remote Sensing” which is one of 
the key application projects in eighth five year plan of China. It 
is divided into 2 levels. Level I includes 6 primary types and 
level II includes 25 types. It is suitable for regional land use