Ground-based remote sensing is now essential for wind energy purposes. Currently available remote sensing instruments construct a wind vector from wind components measured at several spatially separated volumes, leading to errors on complex terrain where the flow is inhomogeneous. Wind estimation errors are found to be fully described by only two parameters: the flow curvature and flow inclination above the instrument. However, neither parameter is measured directly, nor are they simple products of flow models, so the challenge is to adequately estimate them. Linearized flow models are attractive in being fast and requiring few inputs, but make several limiting assumptions that can lead to their failure to adequately predict corrections for remote sensing. It is found that sophisticated CFD models can also over-correct. The status of such corrections is reviewed here, from a number of diverse measurement campaigns, and it is found that generally remote sensed winds can be corrected to within 1.5% of nearby mast winds. Alternative methods, using multiple receivers sensing several wind components within one volume, are also reviewed. Such systems show promise but are under development and further improvements are likely.