Mineral deposits commonly occur within special geologic units or structures, such as fault zones, which can be detected and mapped from aircraft and satellite images. Modern techniques analyze multispectral images that record the way solar energy is reflected or emitted by the materials exposed at the Earth's surface. In sparsely vegetated regions, including most of the Western United States, mineral composition is determined directly by analyzing the spectral properties of rock outcrops. In more densely vegetated terrain, such as the Eastern United States, rock and soil composition can be determined directly in manmade exposures, such as plowed fields and construction sites, or much more general determinations can be made indirectly by analyzing the distribution and apparent health of naturally occurring plants. The association of certain plants with particular rock or soil types has been known for decades. For example, coniferous trees grow preferentially on well-drained sandy soil, whereas deciduous trees dominate on shaly bedrock. These two forest types reflect solar radiation quite differently and, therefore, are distinguished readily in conventional aerial photographs. More subtle plant-bedrock associations require digital multispectral image analysis to infer compositional information from the spectral characteristics of the forest canopy.