The triple collocation method, introduced by Stoffelen (1998), is a well-established method for calculating the relative linear calibration coefficients and absolute error variances of a data set consisting of triplets of in space and time collocated measurements. The method has, e.g., been applied to ocean vector winds measured by scatterometers (McColl et al., 2014;Stoffelen, 1998;Vogelzang et al., 2011), to ocean surface wind speed from scatterometer and altimeter (Abdalla & De Chiara, 2017), and soil moisture from scatterometers and radiometers (e.g., Gruber et al., 2016b, and references therein).The triple collocation equations are solved assuming that the error covariances are zero. In most cases, this is not the case, because error covariances originate not only from correlated measurement errors, but also from differences in resolution between the various observing systems involved. The latter error covariances are known as representativeness errors, and they can easily be included in the multicollocation formalism by correcting the observed covariances for them. The first estimate of representativeness errors was given by Stoffelen (1998) in the spectral domain assuming a k −5/3 spectrum. This was refined by Vogelzang