We consider a novel, accessible potential target for water harvesting on the lunar surface and use a toy model to assess basic robotic mission requirements. The lunar polar permanently shadowed regions (PSRs) are known to contain water ice, and are a high-priority target for near-term resource extraction. However, operating in the PSRs presents a significant operational challenge. Recent modelling in planetary geology has suggested the presence of a large number of smaller PSRs, or micro-cold traps, with a size ranging from a few meters to tens of centimeters. Although one of these traps would not contain sufficient water to constitute a useful reserve, by moving over the surface and harvesting water from multiple micro-cold traps, a rover could build up a significant quantity of water. We construct a simplified model where the micro-cold traps are assumed to be craters of varying size. The spatial distribution of cold traps is modelled as a hard-core Matern process, with a variable radius given by a power-law distribution. We select a subset of craters, and plot traverse trajectories to determine the distance needed to collect ten tons of water. The viability of these missions as an alternate SRU strategy is discussed.