Our knowledge of error in remotely-sensed surface soil moisture retrievals is currently hampered by a lack of soil moisture ground networks with sufficient spatial sampling density to provide direct comparisons with footprint-scale (10 to 40 km) satellite retrievals. This lack of information hampers both retrieval validation activities required to ensure products meet predetermined accuracy goals and the development of data assimilation systems to optimally merge surface soil moisture retrievals into land surface models. Recently applied to soil moisture for the first time, so-called “triple collocation” techniques offer a partial solution to this problem. These techniques are based on the acquisition of three different time series estimates of a given geophysical variable – each with mutually independent underlying errors. Given such independence, simple differencing and averaging can produce estimates of root-mean-square-error magnitudes in each individual product. Using a range of existing surface soil moisture products derived from modeling, ground-based observations and remote sensing data products this presentation will discuss two potential applications of triple collocation to SMAP preparatory activities. First, it will be applied to the point-to-footprint up-scaling problem in soil moisture validation to estimate the fraction of error inferred by comparisons between footprint-scale surface soil moisture retrievals and ground-based point measurements that can be attributed to spatial sampling uncertainty in the ground-based observations (as opposed to error in the satellite retrievals themselves). Results demonstrate that application of triple co-collocation techniques can improve our ability to validate SMAP data products using existing ground-based networks of sparse surface soil moisture observations. Second, triple-collocation estimates of error in passive microwave soil moisture retrievals will be used to improve the optimization of error parameters required as input by existing land data assimilation systems. Such improved optimization will be shown to enhance predictions from a data assimilation system designed to integrate surface soil moisture information into a land surface model.