Monitoring soil moisture is important over agricultural regions for improving the skill of regional crop yield, flood and drought forecasting, and for assisting producers in making effective land-management decisions. Soil moisture data are available from Earth Observation missions, which use active (radar) and passive (radiometer) microwave remote sensing. To take full-advantage of these data it is necessary to obtain a realistic understanding of soil moisture errors (i.e. validation) and to develop methods to account for the inherent lack of sub-footprint variance associated with remote sensing data (i.e. scaling). The aim of this thesis was to address these issues through an analysis of in-situ and remote sensing datasets over an agricultural study region in Canada. The role of in-situ networks in satellite soil moisture validation was assessed through comparison to distributed field-sampling. A dense network of 31 stations and a network of 9 stations both represented up-scaled conditions over a coarse satellite footprint at 0.04 m3m-3 root mean square error (RMSE), but the majority of stations were unrepresentative of local soil conditions. As well, the dense network showed important differences in its characterization of soil moisture spatial variability relative to field-sampling with implications for inferring downscaling properties, especially in terms of vegetation. Measurements were compared from the network of 9 stations over two-growing seasons between 0-6 and 3.5-6.5 cm depths, and further assessed using SMOS L2 soil moisture estimates, showing that in-situ instrument configuration imparts a significant bias in satellite validation of 0.018 m3m-3 RMSE. Synergies between C-band synthetic aperture radar (SAR) and L-band passive microwave retrievals were then assessed. A unique method is proposed to downscale SMOS observations using field-scale SAR retrievals prior to vegetation emergence. This resulted in considerable errors, but results using a soil model suggest promise for the method with improved backscatter modelling. Multi-temporal similarities are also compared over changing vegetation conditions during the growing season using coincident aircraft and satellite data. For corn and canola fields, SAR linear backscatter was significantly correlated with changes in brightness temperature, with implications showing potential of C-band SAR to characterize vegetation contributions (e.g. vegetation water content) to L-band brightness temperature.