Several versions of convolutional neural network (CNN) were developed to classify hyperspectral images (HSIs) of agricultural lands, including 1D-CNN with pixelwise spectral data, 1D-CNN with selected bands, 1D-CNN with spectral-spatial features and 2D-CNN with principal components. The HSI data of a crop agriculture in Salinas Valley and a mixed vegetation agriculture in Indian Pines were used to compare the performance of these CNN algorithms. The highest overall accuracy on these two cases are 99.8% and 98.1%, respectively, achieved by applying 1D-CNN with augmented input vectors, which contain both spectral and spatial features embedded in the HSI data.

A domain decomposition-finite difference time domain (DD-FDTD) method is proposed to solve the problem of electromagnetic scattering from very large rough surfaces. The entire computational domain is decomposed into multiple subdomains. Each subdomain is simulated by using the conventional FDTD method to obtain the scattered fields on a Huygens' surface and the total fields on the rough surface. The latter is used to compute the scattered field on the Huygens' surface in adjacent subdomains, by invoking the reciprocity theorem. Simulation results are compared with those in the literatures to validate this method. The normalized radar cross sections (NRCSs) from very large surfaces of 66λ × 66λ are also presented to demonstrate the efficacy of this method.

A multistatic synthetic aperture sonar (SAS) configuration, composed of an active sonar, a towed receiver and a sonobuoy, is proposed to acquire the image of a moving target and estimate its velocity vector. The range model incorporates the time delay of an acoustic pulse between its emission and backscattered from the target to the receivers. An image is acquired from the received signals at one receiver after range cell migration correction (RCMC), range walk compensation (RWC) and compression, without prior knowledge on the motion parameters of the target. Range-frequency reversal transform (RFRT) and a modified second-order Wigner-Ville distribution (SoWVD) are used to estimate the chirp rate in the range model, and Radon transform is used to estimate the Doppler centroid. The velocity vector of the moving target is accurately estimated, within 3% of error and insensitive to noise, by using the Doppler centroids derived from the signals at the three receivers. Eight different scenarios are simulated to demonstrate the efficacy of the proposed method. The shape and size of the moving target can be clearly identified in all but one case. Three state-of-the-art methods are also used to verify the accuracy of the proposed method, the effects of noise are analyzed, and autofocus algorithms are applied to enhance the acquired images in some difficult cases.

The attenuation of millimeter-wave signals in a sand and dust storm (SDS) can be used to retrieve SDS parameters. A parabolic wave equation method is proposed to simulate millimeter-wave propagation in an SDS, including the effects of absorption and Rayleigh scattering by dust particles with given height profiles of particle-size distribution and total number density. Three different aperture fields of the transmitting antenna are also chosen to simulate wave attenuation under different sets of SDS parameters.

The split-step Fourier (SSF) algorithm is applied to simulate the propagation of radio waves in an atmospheric duct. The refractive-index fluctuation in the ducts is assumed to follow a twodimensional Kolmogorov power spectrum, which is derived from its three-dimensional counterpart via the Wiener-Khinchin theorem. The measured profiles of temperature, humidity and wind speed in the Gulf area on April 28, 1996, are used to derive the average refractive index and the scaling parameters in order to estimate the outer scale and the structure constant of turbulence in the atmospheric boundary layer (ABL). Simulation results show significant turbulence effects above sea in daytime, under stable conditions, which are attributed to the presence of atmospheric ducts. Weak turbulence effects are observed over lands in daytime, under unstable conditions, in which the high surface temperature prevents the formation of ducts.