Investigating the upscaling of in situ soil moisture measurements for satellite validation

Abstract

Remote sensing instruments can provide soil moisture estimations at spatial scales impossible with in situ measurements. Assessment of remotely sensed estimations requires development of calibration and validation data sets with the in situ measurements and through calculation of an areal mean soil moisture (AMSM) using upscaling approaches. However, justification as to the choice of upscaling method is rarely given. This research evaluates six upscaling methods and their resultant AMSM values using soil moisture measurements from an in situ network in Kenaston, Saskatchewan. Applying the upscaling methods to network data obtained over the 2015 growing season, this research compares differences between methods, and their different AMSM values against the Soil Moisture Active Passive (SMAP) remotely sensed retrievals. Disparities between methods appear to be the result of a number of factors including soil texture influencing heterogeneous drying patterns, heterogeneous precipitation across the network and the methods weighting scheme for individual stations. These results demonstrate that method agreement is highest during saturated conditions and that high variation between stations within the network could indicate when to exercise caution regarding validation activities.