Surface soil moisture is essential to global water cycle monitoring, weather forecasting, prediction of drought and flood, and modelling of evaporation. The European Space Agency (ESA) launched the Soil Moisture and Ocean Salinity (SMOS) satellite in 2009, as the first-ever soil moisture dedicated satellite. It uses the passive microwave (radiometer) remote sensing technology due to the direct relationship with soil moisture, but due to technical limitations the spatial resolution is approximately 40 km. This places limitations on hydro-meteorological applications such as regional weather forecasting, flood prediction, and agricultural activities that have a resolution requirement of better than 10 km. Active microwave (radar) remote sensing provides a much higher spatial resolution capability (better than 3 km), but it is less sensitive to changes in soil moisture due to the confounding effects of vegetation and surface roughness. Consequently, NASA has developed the Soil Moisture Active Passive (SMAP) mission, scheduled for launch in January 2015, which will merge passive and active observations to overcome their individual limitations, thus providing a soil moisture product with resolution better than 10 km at a target accuracy of 0.04 cm3/cm3. The rationale behind this mission is to use fine resolution (3 km) radar observations to disaggregate the coarse resolution (36 km) radiometer observations into a medium-resolution (9 km) product. The downscaling algorithms for this purpose have so far undergone only limited testing with experimental data sets, and have therefore been tested mostly using synthetic data and a limited number of suitable experimental data sets mostly in the continental United States. Consequently, this thesis presents an extensive evaluation of soil moisture downscaling algorithms with an experimental data set collected from the Soil Moisture Active Passive Experiment (SMAPEx) field campaigns in south-eastern Australia. This research affords a unique opportunity to undertake a comprehensive assessment of the various downscaling approaches proposed, having applicability to the forthcoming SMAP mission. In particular, each approach is comprehensively assessed using a consistent data set collected over a diverse landscape exhibiting a range of conditions, and then inter-compared with the results from the others. A particular focus is placed on the SMAP baseline algorithm as this is currently the preferred algorithm and scheduled for implementation by NASA immediately upon launch. A preliminary study on the SMAP baseline algorithm was conducted by using existing satellite data; results from which suggested that a better representation of the SMAP data stream characteristics was required. Consequently, a study was undertaken on how to prepare the simulated SMAP data stream from the airborne data set collected from the SMAPEx field campaigns in Australia. These data were processed in terms of spatial aggregation, incidence angle normalization and azimuth effect analysis so as to be in line with the characteristics of the SMAP observations. Results indicated that data from SMAPEx could be reliably processed to represent the characteristics of the SMAP observations. The baseline algorithm was then tested using the simulated SMAP data set. Results showed that the baseline downscaling algorithm had the ability to fulfil the error requirement of medium resolution (9 km) brightness temperature product of SMAP over relatively homogenous area, but it had greater error than the requirement over the heterogeneous cropping area. Consequently, the baseline algorithm was assessed at higher resolutions in order to study the effect of land cover type and surface heterogeneity on the resulting downscaling accuracy. The medium resolution (9 km) brightness temperatures obtained from the baseline algorithm were then converted to a medium resolution soil moisture product, and results compared with other linear methods including the optional downscaling algorithm and a change detection method, and with a non-linear Bayesian merging method. The comparison of these different soil moisture downscaling algorithms suggested that the optional algorithm and the Bayesian merging method had a similar performance in retrieving medium resolution soil moisture products, with the lowest error and highest correlation between downscaled and reference soil moisture, amongst the downscaling algorithms tested. However, unless further improvements can be achieved with the Bayesian merging method the optional algorithm is recommended for application in SMAP due to its simplicity of approach and low computational requirement, thus making it simpler to apply in an operational context.