appear in a continuous gradation from 0 (no use) to 100 (full use). The problem is where to place boundaries. How large an area should have a particular use before that use is acknowledged? How are the problems resolved in areas having heterogeneous mixtures of equally significant land use? Such questions are not easily answered.

Documentation and description of classification criteria help identify where on the gradation scale a land use change occurs. If the size of an area is a deciding factor, the size should be stated on the basis of the information source and the decision criteria.

For example, on land where agriculture has ceased there are various stages of reforestation. Much of the land cannot be properly classified as forestland because there is gradation of cover beginning with woody plant materials that start to grow when the land is dropped from crop production. There are areas with no discernible line where low sparse bush growth changes to stands of marketable timber. In the past, the solution has hinged on well-documented decisions about where the line should be drawn. Remote-sensing platforms can help by providing a vantage point that is not available on the ground.

The above example reemphasizes the significance of user needs in developing a classification system. The interest of commercial foresters differs from that of planners or transportation engineers. Expected use usually determines where boundaries are drawn.

Another problem is the interface of land uses. An example is the large area of suburban-influenced land around urban areas. Factors, including information sources, available funds, significance of land use area, and minimum map units, are used to decide how the land use should be classified and mapped. Similarly, in areas of multiple uses (such as grazed woodlands), several factors are considered. Sometimes a separate class is used to avoid making a decision favoring one class over the other.

Problems also arise in satisfying inventory users when there is a conflict between cover and activity. Many recreational activities are in this category. For example, hunting is a common recreational land use, but most hunting is done on land that is better classified as forestland, rangeland, or agricultural land. Consequently, hunting as a land use may require a different classification. Hunting does not easily fit into an inventory based on remote-sensing

sources.

efficiently by putting each use on a separate overlay, thus retaining the basic data for map users who are not interested in the combination of uses.

In other cases, confusion over large areas of multiple land uses can be clarified by including general statements or guidelines as marginal information. Technically, grazing is rangeland use whether it is in the East or the West. However, management practices in the East are substantially different from those in the West, and it is hard to resolve the difference in classification description. Deciding where the term "range" is to be used instead of the term "pasture" will resolve most of the conflict.

CLASSIFICATION SYSTEMS

Land use classification schemes are abundant. Some are very general; others are highly detailed, having as many as four levels. It is not practical to discuss all classification systems here. The examples included in the handbook were chosen because of their wide acceptance or appropriateness.

National in scope, the USGS land use classification system (Anderson and others, 1976) relies mostly on remote-sensor data. The system was developed for Federal and State agencies that needed. a current overview of land use and land cover with uniform categories at the more generalized levels I and II, and that is receptive to data from satellite and aircraft remote sensors. The system incorporates features from widely used classification systems that are compatible with data derived from remote sensing. The classification is intentionally left open-ended so that Federal, regional, State, and local agencies can be flexible in developing more detailed classifications at the third and fourth levels to meet their needs, and so that they can remain compatible with each other and the national system. The revised system incorporates the results of extensive testing and reviewing of categorization and definitions. The classification is shown in table 3.

An example of how the USGS classification system can be used as a framework for more detailed categorization is the system that was developed in the Land Inventory and Monitoring Division of the Soil Conservation Service (table 4). Categories such as cropland in conservation use, or temporarily idle and grazed and nongrazed commercial and noncommercial forestland cannot be identified solely from remote-sensor data. Therefore, other sources of information are needed.

Many States have developed and used their own classification system or have modified existing sys