With the huge dependence of soil moisture on topography, land cover, soil properties and atmospheric forcing - some or all of which show both spatial and/or temporal variability at different scales - it becomes imperative to understand the propagation and evolution of the prediction uncertainties of soil moisture and revelent heat fluxes across different scales. Our attempt is to use the remotely sensed data and improve the existing model predictions using efficient, flexible and robust assimilation algorithms, which will give us a better understanding of the underlying nonlinear dynamical system. The temporal assimilation involves carrying out point assimilation of near surface soil moisture into a 6-layer stand alone soil column to obtain the posterior estimates of the whole soil moisture profile and relevent energy fluxes. We use the extended Kalman Filter (EKF) for this purpose. The spatial assimilation involves generating such assimilated time series at different spatial scales for the same domain. This provides us with an undertanding of how the soil moisture dynamics evolve at different spatial scales and also provides an insight into the dominant variables and their relevence at different scales. We use a multiscale kalman filtering algorithm for this purpose and study the dynamics at scales ranging from 1 km to 32 km.