Existing radio-based satellite up-and downlinks are reaching their limits in terms of transmission rate. In order to enable higher data rates, optical links are being researched and developed. On ground, spatially displaced multi-aperture systems using different wavelengths can mitigate atmospheric scintillation and increase the uplink transmit power. A four-aperture free-space optical communication system was demonstrated at Fraunhofer HHI over a bidirectional link. For the representative uplink, each array aperture sent a beam carrying the same data but over a different wavelength to avoid inter-beam interferences. At the receiver, the combined signals are optically amplified, optically filtered, incoherently photodetected and electrically filtered. Signal path delays are pre-compensated on the ground-station site. This system shall be scaled up for a long-range demonstration within the EU-funded H2020-project VERTIGO. In this paper, we investigate the performance and scalability of this uplink configuration. When targeting a capacity of Terabits/s, spectral efficiency must be carefully addressed. Direct detection with optical pre-amplification of OOK or DPSK signals is a conventional optical fiber receiver technique, for which the bandwidth of the optical filter can be kept larger than the signal bandwidth only to a certain extent. Above a certain optical filter bandwidth relative to the signal bandwidth, degradation of the signal-to-noise ratio (SNR) becomes significant because of a stronger ASE×ASE noise term. We focus on optimizing power and spectral efficiencies as these are impacted by the introduction of diversity channels. Because these optimizations are essentially independent of the fading channel, we do not assess the reduction of scintillation through channel diversity. We formulate the SNR in terms of number of uplink beams, filter bandwidths (optical and electrical), wavelength channel spacing and received power (in photons per bit). Although the SNR is ideally proportional to the received power, we show how increasing the number of beams while keeping the total received power constant will result in a significant SNR degradation at some point. An optimum for the optical filter bandwidth and for the spacing between the wavelength-division diversity channels is investigated. Polarization-division spatial diversity is kept as an option.