Vegetation height is an important parameter for biomass assessment and ecosystem structure monitoring. However, vegetation height data over large areas are difficult to obtain. Vegetation heights can be retrieved by using large-footprint satellite altimeter system GLAS waveforms, but the current method can only be used in relatively flat forest regions because over sloping mountainous areas, the effects of ground and canopy are mixing on the waveforms and cannot be distinguished. By establishing analytical model of forest target echo, the separation requirement of vegetation echo and ground echo from aliasing waveforms is derived. It is pointed out that the factors of leading to echo aliasing in addition to target roughness and slope, and beam divergence angle is also a factor. The broadening effect and waveform decomposition related to the beam divergence angle, surface roughness and topography slope is analyzed. By using the waveform simulation and GLAS actual waveforms, it is verified to decrease beam divergence angles can reduce the sensitivity of forest echo to sloping mountainous terrains and be more conducive to the retrieval of canopy heights. For laser altimeter to carry on satellite in the orbit of 500~600 km, the beam divergence angles is 40~60 μrad better. The result provides a meaningful reference for designing of satellite laser altimeter system for the retrieval of canopy heights over high sloping terrains.