There is an urgent need for improved soil moisture measurements and modelling due to the relevance of surface moisture to the local energy balance (and hence, for example, air temperature forecasting) to flood prediction and to a range of agricultural applications including plant stress and accessibility of heavy machinery onto the land. The UK Met Office Surface Exchanges Scheme (MOSES) represents land surface processes in the Met Office's Unified Model (MetUM). In this research, MOSES was used to simulate soil moisture over an agricultural site in Norfolk, United Kingdom and validate the model estimations against ground truth soil moisture measurements. Two versions of the MOSES model have been used: the Stand-Alone MOSES (for site-specific modelling) and the Nimrod nowcasting products (5 km grid data) based on MOSES Probability Distributed Moisture (MOSES-PDM). The validation results show that both versions of MOSES perform well in soil moisture estimations in general. However, the MOSES model produces a relatively large seasonal bias in the late summer and autumn seasons and the Nimrod-MOSES-PDM tends to underestimate soil moisture in general. A model sensitivity study shows that the soil parameters are most likely to contribute to the errors. In addition, the MOSES model treats the vegetation fraction as a constant throughout a year over agricultural fields, which was found partly to cause the seasonal bias. Recommendations on how to improve MOSES soil moisture estimation are discussed, as are the likely application benefits.