Investigating the synergies between soil properties and topography from multi temporal radar imagery

Moisture is a major determinant of the dielectric properties of most non-metallic materials and soils and vegetation in particular. Natural surfaces that are subjected to intense wet dry cycles will exhibit a change in dielectric constant and hence reflective capability. An investigation of these changes with time indicates that understanding the topographic setting alone will yield significant data about soil moisture and drainage patterns. Vegetation also exhibits a greater moisture content when in close proximity to the water level. Vegetation laterally separated from the water hays off due to lack of available soil moisture and becomes essentially invisible to radar. Analysis of three multi temporal radar images obtained by Radarsat during the period of 1998-1999 indicates significant lateral drainage of floodplain surfaces in Northern Australia. A region of the South Alligator River floodplain from within Kakadu National Park was surveyed to ascertain soil properties and topography. The images were co-registered and segmented by means of a Gaussian Markov random field model in order to highlight areas of similar reflectance characteristics for each of the three images. Segmentation allowed the comparison of boundaries between flooded and dry areas of the floodplain over the course of the wet dry cycle. The drying effect exhibited in the segments is directly related to the topography of the floodplain surface. Results indicate a novel new method for vegetation, soil and topographic mapping in the wet/dry tropics of Northern Australia.