Wind flow up a mountain tends to enhance precipitation – when the air moves higher into the atmosphere it is cooled, which drops the saturation dew point, and therefore tends to make more moisture available.  Wind blowing down the mountain does the opposite.  If the atmosphere is sufficiently unstable, the lift provokes deep convection.  But in less convective situations, the extra moisture is squeezed out in the lowest layers of the atmosphere, mostly as the entirely liquid “warm rain process”.  Either way, orographic (mountainous) effects create fine-scale variability that challenges all observations:

• More rain gauges are needed to pick up the variability, but there are almost always fewer gauges in the mountains.
• Surface radars can’t see all the way to the surface due to blockage of the radar beam, ground clutter, and the increasing height of the lowest beam with height.
• Satellite radars can’t see all the way to the surface due to ground clutter.
• Satellite passive microwave radiometers over land depend on channels that sense frozen hydrometeors (snow, graupel, hail), but in countries such as Puerto Rico, these frozen hydrometeors are only present at altitudes of 4 or more kilometers above the surface.  So, if the orographic enhancement is in deep convection, then the satellite signal is relatively well-correlated to surface rain.  But when the extra boost is due to warm rain enhancement, the satellite signal might only have a weak relation to the surface rainfall.
• Satellite infrared sensors only see the cloud tops, and as for the passive microwave sensors, warm rain processes near the surface can be more or less decoupled from what’s happening in the upper reaches of the cloud.
• Various numerical model-based estimates have issues, which include the time/space scale of the computation being too coarse to adequately resolve the processes, approximations in the microphysics (conversions among vapor, droplets, and ice particles) that are not sufficiently accurate, and shortcomings in the input observations needed to feed the computation the correct upstream conditions.

Because of these factors, satellite based estimates of precipitation in mountainous regions may be less accurate than ones in non-mountainous regions. For the IMERG product, you can get an idea of the accuracy of a given precipitation estimate by checking the quality index. Learn more.