A channel-estimation (CE) scheme is proposed to estimate the complex amplitude, Doppler shift, angle-of-departure, and angle-of-arrival of the channel taps for sparse and doubly selective channels for hyperspectral image transmission from unmanned aircraft vehicles (UAVs) to ground stations. The proposed method is dubbed as compressed-sensing joint parameter estimation (CS-JPE) and finds the channel parameters matrix by employing a compressed-sensing (CS)-based method. Afterward, a modified version of the joint parameter estimation (JPE) is proposed as CS-JPE and is dubbed as M-CS-JPE, which employs the elevation-azimuth angles of the line-of-sight channel tap to estimate the channel parameters with higher accuracy and lower computational complexity compared to the CS-JPE scheme. For higher accuracy of the M-CS-JPE, an elevation-azimuth angle estimation is proposed and is dubbed as fractal-structurearray since it uses a fractal structure for the placement of the UAV antennas. The performance of the CE methods is appraised by simulating transmission of AVIRIS hyperspectral data via the communication channel and evaluating their accuracy for the classification after demodulation. Compared to the least-square method, the simulation results indicate up to 30-dB figure of merit in the bit-error-rate and 10 times improvement in the hyperspectral image classification fidelity.