Increasing interest is devoted to the development of solar powered plants in the countries of the Arabian Peninsula. However, the unique climatological conditions of the region, characterized by high dust concentrations, high humidity and modest cloud coverage, cause a challenge in solar radiation estimations. This study presents and compares yearly direct normal, diffuse horizontal and global horizontal irradiation maps for the years 2008 to 2010 over the United Arab Emirates using an ensemble artificial neural network model and a cloudiness index corrected semi-empirical model, both of which were derived from satellite images. A validation against ground measurements from one station is also performed.

The vast alluvial fans of Death Valley, California, provide an ideal environment to examine the remote sensing measurement of geologic surfaces. One of the objectives of the shuttle imaging radar C (SIR‐C) program in Death Valley is detection of the variation in surface microtopography with age of the surface. We present results of extensive field measurements of surface roughness together with an analysis of the effects of the surface microtopography on radar backscatter and visible and near‐infrared (VNIR) reflectance as measured by aircraft and satellite sensors. This subject is addressed in both the forward and inverse sense: surface simulation and forward modeling are used to determine expected roughness effects, while a method of inverse analysis that uses finite impulse response (FIR) filters is used to assess the potential for inversion of multifrequency, polarimetric, synthetic aperture radar (SAR), and multispectral VNIR imagery for surface roughness. The interaction of radar and VNIR radiation with the Death Valley surfaces is complicated. Simple roughness parameters such as rms height, and slope and offset of surface power spectra, do not represent a sufficiently complete description of surface roughness to predict the radar or VNIR signature uniquely. Multiple scattering, which is controlled to a large extent by aspects of the phase of the surface Fourier transform, also exerts a controlling influence on the observed signal. The phase aspect of surface roughness has not been considered in existing roughness characterization. Our inversions demonstrate retrieval of roughness parameters with almost equal success from both SAR data and Landsat thematic mapper (TM) data and indicate much potential for joint SAR/VNIR data analysis. The solutions are not, however, very stable and include effects of additional parameters such as intermediate‐scale topography and vegetation cover which masquerade as roughness variation. In designing a stable inversion of more general applicability, the multifrequency and polarimetric aspect of SIR C data is important. Nevertheless, high‐resolution roughness recovery will probably require hierarchical analysis of radar and optical images, and also SAR acquisition at multiple look angles and directions.