Information acquisition

Decisions concerning how the survey and other information gathering activities will be performed should be made at the outset of the resource assessment. Although this will not affect the type of information that is collected, it will determine, in part, how it is collected and organized.

Survey Approaches

Resource assessment might be approached in two ways: the component approach, in which each land characteristic is mapped individually, and the landscape or land systems approach, in which land is viewed as an integrated whole and the units that are mapped are more or less homogeneous. In both approaches, land characteristics are analyzed together to derive an estimate of land capability.

Component Approach

In much of the world, including the United States, agencies have been established to study individual resources (for example, climate and soils) within the country, or have been assigned to manage specific types of land use (for example, forest and rangeland). Both types of agencies conduct their own mapping activities. Thus, in a range improvement project, it is common to find that one or more land characteristics (for example, geology) have been mapped already for much of the project area. With this pattern established, a project will likely continue mapping land characteristics individually in the interests of consistency and economy. Moreover, project mapping may be carried out by several groups on a component basis because of the distribution or resource responsibilities among participating agencies.

Because it is difficult to map certain land characteristics individually, some characteristics often will be combined in one map (for example, climate and vegetation; vegetation and land use/land cover). Because of the constraints of scale, mapping units will tend to be somewhat heterogeneous, but will be designed around naturally linked groupings of land characteristics (for example, associations of soil or vegetation).

Landscape Approach
The need to adopt a systems approach during most phases of a range improvement project is emphasized throughout this volume. Two considerations argue for the systems approach at the resource assessment level. First, as noted above, the region should be understood in terms of its differences in land capability. This quality of the land is derived from physical and biological characteristics such as climate, soils, and vegetation. Although these characteristics may be surveyed individually, they must be considered together to determine land capability. Moreover, the units of land that are mapped function more or less like systems and not simply as a collection of independent components. Change in any one of the components will affect or be affected by other components in varying degrees. Thus, at the resource assessment level, it is desirable to consider the units of land to be managed as integrated "landscapes" with a distinct set of related characteristics.

Second, from a practical standpoint, significant economies of effort and improvements in product quality can be achieved by combining related or complementary aspects of the resource assessment. For instance, in a landscape approach, an interdisciplinary team (perhaps a soil scientist, geomorphologist, and plant ecologist) performs the mapping and analysis tasks as a group, rather than producing a set of individual maps and reports. Field expenses are reduced by combining activities, and mapping consistency and analysis quality are improved by complementary collaboration.

The landscape or land systems approach to resource assessment was developed and applied first in Australia after World War II. The problems faced there were not unusual: large areas of the country had to be surveyed quickly and accurately to determine their agricultural potential. This highly successful approach is still used in Australia. Comparable approaches have been developed by other countries in many parts of the world for agricultural, military, and engineering purposes. In the following discussions, the Australian terminology is employed.

The land systems approach has a hierarchical structure of units (figure 4-1). The smallest unit of land recognized is the land element. It is defined primarily by slope, and is essentially homogeneous in all properties, corresponding to the concept of "site." Because of its limited extent, it is never mapped at the resource assessment level, but rather is the primary focus of site evaluation, as described in the following chapter. The next largest unit of land is the land facet, which consists of a set of related land elements, commonly on the same landform. It is seldom mapped in a resource assessment. The land system is the largest unit, and consists of geomorphologically and geographically associated patterns of land facets. The land system is the most commonly used mapping unit in this approach and is well suited to the general purposes of the resource assessment.

Information Acquisition Methods

Archival Research

A search of archival sources of information is done at the beginning of the project to gather the work that has been done to map and describe land characteristics of the region. This avoids a duplication of previous work, and builds on the experience and insights of other workers in the region.

Published information on land characteristics exists for essentially all parts of the world. The types of information that can be found include maps and descriptions of climate, soil, and vegetation. Most of this information is extremely small in scale (for instance, world or regional maps produced by the United Nations Food and Agriculture Organization). Although continental-scale maps are not suitable for a final assessment, they are a useful starting point.

FIGURE 4-1 An example of land systems mapped in Jordan. (Mitchell and Howard, 1979)

Moreover, general reports usually contain extensive bibliographies that may lead to more specific studies for the project region.

Many papers are published in professional journals as a product of scientific studies conducted in a region. Commonly, these papers will describe one particular aspect of the region, such as overgrazing and erosion, that will not be critical to the total resource assessment, but may provide some insight into a particular problem in one part of the region. Descriptions of these papers can be found in science indexes and abstract journals, either by topic or by geographical location.

As noted above, maps and reports describing various land characteristics in a region are produced by regional, national, and international agencies. Because these materials may not be widely distributed, and their existence may not be generally known, inquiries must be made at all levels to find what work may have been done within any one region. Finally, some types of information, such as land ownership or census data, may be available only from local or regional archives.

All of the information needed for a resource assessment probably will not exist in a usable form at the start of a project. For example, vegetation maps are relatively uncommon, and any existing maps for a region may have been done for a purpose that is not compatible with the objectives of the project (for example, a map of forest resources will be quite detailed for forested areas but may describe non forested lands only as "rangeland" ). In other instances, existing maps or data may be out of date. Thus, in most resource assessments a good deal of map and supporting information must be gathered and compiled independently during the project. Some of the more commonly used techniques for generating this information are discussed below.

Interviews

Discussions with local administrators, researchers, and especially land managers can be conducted to gather information on those characteristics that cannot be directly observed and that are probably not recorded, such as land use, land management practices, animal management practices, general management issues, economic conditions, land capability, local perceptions of resources, and any other information that may support specific project objectives.

As suggested above, background information can be extremely

important in estimating land capability in terms of indigenous practices. It also should provide some clues about the acceptability of proposed changes in management practices.

Interviews may be conducted formally, and may rely on the use of questionnaires if the objectives of the project call for a quantitative description of some features of local culture. However, informal interviews are done more easily and may serve equally well.

Ground Sampling

Much of the information required for the resource assessment can be gathered only through direct observation. This information may be used to develop maps, to develop estimates of the magnitude of other characteristics that are not ordinarily mapped (for example, population), or to describe the composition of mapping units (for example, vegetation species and cover, and soil type and depth) that have been recognized by other means (see the following section on remote sensing).

Maps and estimates of land characteristics may be developed in two ways. First, ground samples may be gathered in a sampling pattern such as a grid. Reasonable maps or estimates may be developed from such data. However, the accuracy of this approach is dependent upon the complexity of the region and the density of sample points. At the resource assessment level, it is unlikely that a project could afford the expense of allocating enough samples to characterize a large region. Thus, systematic ground sampling is used only for very intensive studies, such as irrigation soil surveys, or where the features of interest are assumed to be poorly correlated with other observable features, such as archaeological sites.

A second approach employs stratified sampling and is used where it is possible to assume a reasonable correlation between two or more characteristics. For example, in a landscape approach to survey, it is possible to stratify an area according to landform and elevation if good topographic maps of a region exist. Ground samples are allocated to each stratum, according to its importance or complexity. Maps and estimates developed in this way are reasonably accurate and more efficient than a systematic approach. However, dynamic land characteristics, such as land use and vegetation, may be inadequately sampled because of their high variability. For example, major changes in vegetation resulting from clearing or fire may be missed because they are not necessarily correlated with landform.

Remote Sensing

Although maps and estimates of land characteristics can be produced by ground sampling alone, it is seldom done for a resource assessment because of the expense and the likelihood of missing important features.

Remote sensing is the most commonly used tool for gathering information for large areas As defined here, remote sensing includes the uses of aerial photography and satellite imagery to study the earth's surface. Remote sensing data are unique because they (1) provide a comprehensive picture and permanent record of surface conditions at one point in time and (2) present a vertical perspective in which all features are represented, essentially in their true geometric relationship with all other features. There is no ideal remote sensing system. Thus, a primary task in remote sensing is to select a system that best meets the needs of the project.

Principles of Remote Sensing Remote sensing exploits the differences that can be detected among surface features on an image of the earth. The ability to distinguish among features is conditioned by several factors. Foremost in many applications is the feature's tone or color. Earth materials reflect or emit electromagnetic radiation, including light, in different ways (figure 4-2). For example, vegetation has a unique pattern of reflectance, with moderate reflectance in the green part of the spectrum, low reflectance in the red part of the spectrum, and very high reflectance in the infrared part of the spectrum that is just beyond what is visible to the human eye. Second are those inherent properties of a feature that determine how it appears, or what is sometimes called a feature's "signature" or "response." The characteristic shape of a surface feature when viewed from above (for example, a folded geologic structure) or its relative size (such as tree versus shrub) are two such properties. Other important characteristics are less obvious, such as a feature's “texture" on an image (say the difference between the smooth texture of a meadow and the rough texture of a forest canopy), or the association of one feature with others (such as pine forests on steep north-facing mountain slopes).

FIGURE 4-2 Spectral characteristics of different earth materials.

The type of remote sensing system used also affects the ability to distinguish among earth features Scale - the relationship between the size of the image and the area on the ground it portrays - largely determines what can be seen, especially if geometric properties such as size, shape, and even texture are noted. Other important characteristics of remote sensing systems have to do with system resolution, or its "sharpness" in several dimensions - spatial, temporal, and spectral.

Spatial resolution broadly describes the quality of the system that determines the smallest feature that might be detected. Thus, spatial resolution influences the ability to detect features based upon geometric characteristics. For example, to determine tree densities in savannah vegetation, relatively high-resolution images would be required to see individual trees, while simply to map the boundary between forest and savanna, low-resolution images might be preferred to enhance differences in total tree cover rather than the location of individual trees.

Temporal resolution describes how often a system acquires images for a single point. Although not usually a consideration for aerial photography, temporal resolution is an important characteristic of satellite systems because they operate continuously. For example, mapping forests in a large area and also distinguishing between evergreen and deciduous types would require a system that acquires images frequently enough to assure at least one cloud-free image from both summer and winter.

Spectral resolution describes the location and width of the parts (bands) of the spectrum in which the system records. Earth materials reflect and emit electromagnetic radiation in different ways. To improve the ability to discriminate features it can be useful to employ a "multispectral" approach by examining several different parts of the spectrum. The value of this approach is easily appreciated when comparing a black-and-white panchromatic photograph with a color photograph. Color photography provides much more information than black-and-white, but it tends to have poorer spatial resolution, is more expensive, and is difficult to process in some parts of the world because of the lack of proper equipment.

There is much information outside the visible spectrum that would be useful for studying vegetation (table 4-1). Photographic films have been developed that are sensitive to infrared radiation that is just beyond the visible part of the spectrum. Color infrared (CIR) film is the most common type. To record infrared energy using conventional photographic technology, colors from the natural environment are assigned to other colors on the CIR film. Thus, the final product is sometimes called a false color image: blue is filtered out, green is recorded as blue, red is recorded as green, and infrared is recorded as red. Because plants reflect more infrared light than green light, green vegetation appears as various shades of red or pink. Red soils are yellowish-green, and urban areas are bluish-gray. CIR photography is especially effective for mapping vegetation, but is expensive and sometimes difficult to expose and process.

Nonphotographic sensing systems can be carried by aircraft and spacecraft and provide similar kinds of spectral information. Nonphotographic systems have a number of advantages. For example, parts of the spectrum that are critical in some applications and that are beyond the capability of photographic systems, such as the thermal and microwave (radar), can be sampled. Also, many nonphotographic systems record images digitally, which allows several processing options (see below). With the exception of Landsat satellite data, however, nonphotographic imaging systems will be used in few resource assessments because of the expense of processing and the need for special computer facilities.

Satellite Systems Since 1972, satellite imagery suitable for land resource assessments has been produced continuously for most parts of the world. Landsat was the first satellite to provide regular and universal image data and continues to be the most widely used system.

Table 4-1 Comparison of Landsat MSS, TM, and NOAA AVHRR System Characteristics

Table 4-1 Comparison of Landsat MSS, TM, and NOAA AVHRR System Characteristics

* Instantaneous Field-of-View Resolution

SOURCE: Adapted from the Final Report of the Panel on the National Oceanic and Atmospheric Administration Climate Impact Assignment Program for Africa, BOSTID, National Research Council, Washington, D.C., January 1987.

The primary instrument on Landsat is the Multispectral Scanner (MSS), which records images of the earth in four spectral bands (see table 41). Images are recorded digitally but are produced in both digital and photographic formats. The MSS creates an image by recording the relative brightness of each element or cell of a large array. Each picture element (pixel) equals an area on the ground of approximately 0.5 hectares (60 m x 80 m). An image is created for each band in the green, red, and two infrared parts of the spectrum. Images from each band may be combined to create an image that is similar in color renditions to a conventional CIR photograph, and may be interpreted manually. Because they exist in digital form as well, images may be processed statistically using a computer. Although spatial resolution is relatively low, the MSS is well suited to resource assessment because a single image covers a large area and, as suggested above, it is desirable sometimes to avoid the confusion introduced by detailed data.

The most recent series of the National Oceanic and Atmospheric Administration (NOAA) weather satellites has carried the Advanced Very High Resolution Radiometer (AVHRR) instrument. The AVHRR has low resolution (see table 4-1) because it was intended to complement conventional very-low-resolution weather satellite systems by acquiring data that could be used to describe general land surface conditions. Although AVHRR has been used mainly for monitoring studies, it might provide useful information for exceptionally large regions.

Aerial Photography Aerial photographs are the most widely used form of remote sensing data. They are routinely acquired in most parts of the world for a variety of purposes, including geophysical surveys and the production of topographic maps.

The principal advantages of aerial photography are its high quality (conventional image format is about 23 cm x 23 cm) and the ability to schedule photographic missions at the proper time and appropriate scale using the desired films and filters. Moreover, aerial photography firms can be contracted to fly missions over almost any area in the world. The primary disadvantage of conventional aerial photography is its relatively high cost.

To counter the high costs of conventional aerial photography, increased use has been made of 35 mm cameras for aerial photography (figure 4-3a and 4-3b). Because the film format is so small, most 35 mm photography has been acquired at very large scales from low flying light aircraft. Images from these systems are of somewhat lower quality than conventional aerial photography and they must be used with some care. The advantages in cost and flexibility, however, seem to offset most other considerations when the 35 mm system is used for resource assessment.

Use of Remote Sensing Data
Aerial Photograph Interpretation Satellite images or aerial photographs can be interpreted manually by a trained analyst who is familiar with the study region (figure 4-4). As in any mapping exercise, the objective is to recognize areas that are more or less homogeneous in one or several properties. Using knowledge of the region and the image characteristics, the analyst examines the image and manually delineates areas judged to be homogeneous. areas or mapping units that appear to be related (for example, sandy alluvial fans) are labeled accordingly. These mapping units can serve as strata for subsequent sampling. Boundaries are determined, and mapping unit descriptions are generated by examining large-scale aerial photographs or by analyzing ground samples.

Single images are rarely used in remote sensing. The preferred approach is to examine a number of images at scales larger than that of the base map; larger-scale images are used to refine or label mapping units defined at higher levels Ultimate verification is provided by ground sampling This multiple strata sampling scheme is called the multistage approach to remote sensing (figure 4-5). The multistage approach is used with all types of remote sensing data but is especially effective when satellite images are being used.

Photo Sampling Several other types of information must be gathered during the resource assessment that are not normally mapped by remote sensing (including human population, herd sizes and composition, and detailed land use). Although much of this information might be gathered through ground sampling or interviews, other techniques have been developed for using large-scale (larger than 1:1,000) 35 mm photography as a supplement for ground sampling. For example, a rich methodology has evolved for the use of low-level photography for studying rangeland and large animals in East Africa (International Livestock Centre for Africa, 1981). These systems use the aerial photograph as a sample point in a systematic sampling scheme. Detailed interpretations of the photographs are used to develop accurate estimates of the sizes of domestic and wild animal herds, human populations, crops, and land use at the ranch level for very large regions at comparatively low costs.

Digital Processing Satellite images are available in digital format. These data may be selectively enhanced to produce images that are more easily interpreted by the analyst. For example, if the project is located in a region dominated by bright sandy soils, the contrast at the brighter end of the tonal range can be increased at the expense of the darker. This type of enhancement would bring out subtle differences in brightness that otherwise might be overlooked (figure 46).

Applicability and Availability of Remote Sensing Data Remote sensing may be the only way to acquire information about basic land resources in many parts of the world. Maps of natural resources and land use are relatively rare in many nonindustrialized countries. In fact, adequate base maps that describe topography, roads, and cultural features may not exist for many areas.

FIGURE 4-5 The multistage approach to sampling as applied in remote sensing (after Townshend, 1981).

Conventional aerial photography is usually the preferred source of information for mapping land resources. As previously noted, however, existing aerial photography may be quite old and thus of relatively little value for mapping current conditions. Even current aerial photography, though, may present difficulties. Because of the great amount of detailed information they contain, conventional aerial photographs are commonly perceived to have considerable intelligence value. As a result, their distribution is sometimes controlled by military authorities and thus may not be available.

If the project is large enough, acquisition of aerial photography may be a major activity and warrant special attention. A large number of private firms provide aerial photography services. Their names and addressses can be found in the telephone books of major cities or in directories of the journals of major professional societies, such as Photogrammetric Engineering and Remote Sensing, the journal of the American Society of Photogrammetry and Remote Sensing.

FIGURE 4-6 A method for using digital classification of Landsat imagery for vegetation inventory. Low-level, large-scale aerial photographs were acquired along randomly selected transects. The photographs were used to estimate biomass and soil condition along the sample transects. These values were correlated with the various Landsat spectral classes along each transect. Summary values then were calculated by multiplying the area of each spectral class by the vegetation and soil values derived from the sample transect data for the entire study area. (Courtesy of Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, McLeod and Johnson, 1981)

The United States has maintained an "open skies" policy in the acquisition and distribution of data from the Landsat satellites. Since the initiation of the Landsat program, ground-receiving stations have been established in a number of countries around the world (see list below). Images from these stations are available at a modest cost and cover most parts of the earth (see table 4-1).

Where large areas are studied and extensive ground sampling poses a problem, large-scale aerial photography may be required to supplement satellite data and conventional aerial photography. This is particularly true when the efficient multistage sampling approach is employed. Nonconventional systems have been developed (see figures 4 3a and 4-3b) that provide excellent data at low cost for large areas (International Livestock Centre for Africa, 1981). Where budget presents no problem, conventional aerial photography can be purchased from a service.

The variety of remote sensing systems currently available ensures that basic information on land resources can be produced quickly at reasonable cost for almost any country. Should training or re-training be necessary to benefit from these systems, practical training courses are available at most major agricultural universities, and at many remote-sensing centers. Aside from short-term courses (less than one month duration), university training is generally part of a longerterm advanced-degree program. However, a variety of variable-duration, comprehensive training programs in all aspects of remote sensing are available under Unesco auspices through the International Institute for Aerial Survey and Earth Sciences (ITC), 350 Boulevard 1945, P.O. Box 6, AA Enschede, The Netherlands.