Wind-driven desertification: process modeling, remote monitoring, and forecasting

<p>Arid and semiarid landscapes comprise nearly a third of the Earth's total land surface. These areas are coming under increasing land use pressures. Despite their low productivity these lands are not barren. Rather, they consist of fragile ecosystems vulnerable to anthropogenic disturbance.</p> <p>The purpose of this thesis is threefold: (I) to develop and test a process model of wind-driven desertification, (II) to evaluate next-generation process-relevant remote monitoring strategies for use in arid and semiarid regions, and (III) to identify elements for effective management of the world's drylands.</p> <p>In developing the process model of wind-driven desertification in arid and semiarid lands, field, remote sensing, and modeling observations from a degraded Mojave Desert shrubland are used. This model focuses on aeolian removal and transport of dust, sand, and litter as the primary mechanisms of degradation: killing plants by burial and abrasion, interrupting natural processes of nutrient accumulation, and allowing the loss of soil resources by abiotic transport. This model is tested in field sampling experiments at two sites and is extended by Fourier Transform and geostatistical analysis of high-resolution imagery from one site.</p> <p>Next, the use of hyperspectral remote sensing data is evaluated as a substantive input to dryland remote monitoring strategies. In particular, the efficacy of spectral mixture analysis (SMA) in discriminating vegetation and soil types and detennining vegetation cover is investigated. The results indicate that hyperspectral data may be less useful than often thought in determining vegetation parameters. Its usefulness in determining soil parameters, however, may be leveraged by developing simple multispectral classification tools that can be used to monitor desertification.</p> <p>Finally, the elements required for effective monitoring and management of arid and semiarid lands are discussed. Several large-scale multi-site field experiments are proposed to clarify the role of wind as a landscape and degradation process in dry lands. The role of remote sensing in monitoring the world's drylands is discussed in terms of optimal remote sensing platform characteristics and surface phenomena which may be monitored in order to identify areas at risk of desertification. A desertification indicator is proposed that unifies consideration of environmental and human variables.</p>