Address correspondence to:
Dr. Milton O. Smith
Dept. of Geological Sci.
AJ-20, Univ. of Washington
Seattle, WA 98195

A remote-sensing approach was used in conjunction with field measurements to examine local and regional-scale environmental processes that covary with the abundance and distribution of vegetation in a semiarid ecosystem. Images of the fractional abundances of vegetation and soils were constructed by spectral mixture analysis of Landsat Thematic Mapper (TM) satellite images, covering a 150-km segment of Owens Valley, California. These images, along with a TM image of the radiant temperature, a digital elevation image and ground-based measurements of precipitation and evapotranspiration, were examined to isolate the effects on vegetation of the covarying factors, net radiation, temperature, elevation, soil type, and precipitation. On a regional scale the abundance of desert scrub on the bajadas of Owens Valley appears to be influenced most by the mean annual precipitation. Also regionally, vegetation cover is sensitive to the differences between the gravelly fanglomerates of the bajadas and the alluvium of the valley floor. Other edaphic and ground-water effects are important but localized, and are most pronounced on the valley poor. They produce patterns in vegetation abundance that are independent of and superposed on the regional precipitation-controlled pattern. Temperature covaries with vegetation less well than precipitation, and appears not to be the major influence on either the amount of vegetation or the boundaries between major vegetation communities. The image-derived measure of vegetation cover correlates closely with ground-based measurements of evapotranspiration. The study demonstrates that local observations cannot be extrapolated safely to the regional scale, and that a combination of local field measurements and the regional measurements provided by remote sensing is required to determine the environmental factors that control vegetation distribution.

INTRODUCTION

As an example of the scaling problem, Jarvis and McNaughton (1986) pointed out that predicting regional evapotranspiration requires more than extending local models to landscapes. They showed that, at the scale of the leaf or the plant, evapotranspiration (ET) is effectively regulated by stomatal conductance, but at the regional scale net radiation and average temperature are the controlling factors. Stomatal regulation at the level of individual leaves leads to compensating microclimatic feedbacks at the community level. Measurements of stomatal conductance are not sufficient to estimate ET at the community level.

Satellite multispectral images are now beginning to be applied to ecological studies (e.g., Waring et al., 1986). Because digital images consist of massively repetitive discrete spatial measurements, they potentially provide one way to extend information from local to regional scales. Data from several Earth-observing satellites are presently available, and new systems are under development.

Satellite measurements have potential for depicting parameters such as R_n and T_s, as well as other parameters relating to the interaction of electromagnetic radiation and the Earth's surface. However, this potential can be realized only if the physical significance of the remote measurements is established ecologically. Thus, for example, it may be significant to determine the percent cover of green vegetation for an area on the ground, but it is not of direct value to the ecologist to know the radiance at various wavelengths for the same area. The statistical classifications and correlations that commonly are used on multispectral images are performed on the radiance or reflectance data, and, lacking a physical basis by themselves, are not sufficient to connect the satellite measurements to materials and conditions on the ground.

In the present study Landsat Thematic Mapper (TM) satellite data and field measurements from Owens Valley, a region encompassing the transition between the Mojave and Great Basin deserts [Fig. 1b)], are used to examine environmental patterns at several spatial scales. The satellite measures scene radiance in six different spectral bands in the visible and near-infrared wavelength regions, and one band in the thermal infrared region. Radiance values in the visible and near-infrared regions respond to variations in the chemistry and structure of materials at the Earth's surface, but they may not respond directly to the same parameters that an ecologist would measure in the field, such as vegetation cover, soil moisture, or community composition. In the field and laboratory most plant species and communities are identified by morphological rather than chemical criteria, and the resolution of a single (30 m x30 m) TM picture element (pixel) is far too coarse for direct morphological identification (Part I). Aggregates of species may be identified from satellite images if they possess unique reflectance spectra, but this is not often the case. More typically, species are inferred from spatial and/or temporal patterns and from context.

In Part I we described a method for analyzing Landsat images that isolated the spectral radiance contributions of vegetation, soil, and shade in the pixel data. Radiance measurements from the six visible and near-infrared image channels were mathematically transformed into the relative fractions of a few endmember spectra which, when mixed together, accounted for the observed spectral variation in the scene. Using this spectral-mixture framework, we then mapped the relative abundances of two vegetation and two soil types in Owens Valley. The goal of Part II, the present paper, is to examine the relationship of the scaled vegetation fraction (VF_s) measured by Smith et al. (1990) to five environmental factors: net radiation (R_n); radiant temperature (T_s); elevation; mean annual precipitation (MAP); and soil type. All of these environmental factors can be determined and expressed in image form along with the vegetation abundance, using a TM image, reference spectra, ground-based meteorological data, and digital elevation images. We examine and compare these spatial data sets, and test whether local correlations among environmental parameters that have been documented in the field extend to regional scales. We then use the spatial patterns of the observed regional gradients to assess to what extent each of the factors affects the distribution and abundance of vegetation.

A specific objective in Part II is to determine whether temperature, water (soil moisture), or soil composition has a dominant effect on the abundance and distribution of vegetation in Owens Valley. Soil nutrients, moisture, and temperature have all been implicated as limiting vegetation resources in semiarid lands (among others, see Merriam, 1898; Billings, 1949; Shelford, 1963; Beatley, 1974; 1975; Schreve, 1912; 1934; West, 1983a, b, c; MacMahon and Wiebolt, 1978; MacMahon and Wagner, 1985). Some authors stress the significance of minimum/maximum air temperatures in winter and summer in controlling the distribution of semiarid vegetation, or utilize day-degrees (integrating temperature over time) as factors to separate vegetation zones (Hastings and Turner, 1965). Scrub cover has been shown to be proportional to precipitation in some semiarid regions (Beatley, 1974; 1975; Goldberg and Turner, 1986), and a review by Shimeda (1985) concludes that Leaf-Area Index (LAI) and productivity are generally proportional to precipitation.

A high correlation among precipitation, vegetation cover, and elevation commonly has been observed in semiarid landscapes. Because precipitation varies with both air temperature and elevation, it has been difficult to isolate the effects of each on plant distribution. Previous studies have relied on extensive temporal sampling and have involved only limited spatial sampling. For example, Beatley (1974) concluded that minimum daily air temperatures control the transition between the vegetation communities of the Mojave and Great Basin deserts, based on relatively limited spatial/temporal data acquired at local scales. We argue that causal relationships among these interrelated parameters may best be revealed on a regional scale, by correlation of spatial patterns of variability in registered maps and images of VF_s and the candidate driving parameters. Using the TM images, we can study vegetation distribution and abundance using a 100%-sampled spatial data base.

Soil moisture cannot be measured directly by TM; however, we hypothesize that if the abundance of green vegetation (which we derived from the six-band radiance data) is an indication of availability of soil moisture, then the vegetation fraction (VF_s) and related spectral parameters are indirect measures of evapotranspiration (ET), integrated over a period of time characteristic of the vegetation community. These hypotheses are tested in terms of ground micrometeorological measurements and the patterns in the image data.