LIDAR and IFSAR, which stands for Interferometric Synthetic Aperture Radar, are two remote sensing technologies that have been proven to be very effective in generating elevation data. All too often they are evaluated as one versus the other. It is now possible to integrate the two types of data to leverage the positive characteristics of both sensors. This concept was successfully demonstrated recently in a project for the Murray-Darling Basin Authority in Australia. This basin in Southeast Australia drains one-seventh of the land mass in the country and covers the main agricultural area in Australia.
The authority acquired IFSAR data of a major portion of the basin to support a number of hydrologic modeling applications. They also wanted to integrate several existing LIDAR datasets into the overall elevation model. The existing LIDAR data sets in the project area had been collected over a number of years to different specifications by different contractors. As you might expect, these datasets had a number of inconsistencies. One of the goals of the project was to evaluate the IFSAR/LIDAR fusion process as a means of adjusting the disparate datasets into a consistent vertical framework.
The demonstration area was approximately 1900 sq. kilometers and contained six different LIDAR coverages collected from 2003 to 2007. As the first step, these six LIDAR coverages were evaluated. They were found to be very consistent horizontally. The vertical consistency was tested and found to be lacking. They were within 50 cm of each other.
A two-step process was used to adjust the LIDAR to the IFSAR. Intermap used tools it developed to allow the adjustment of ancillary elevation models to the IFSAR elevation models. First the lightly vegetated or bald areas in the IFSAR data were identified. These are areas where the vertical accuracy of the IFSAR data was known to have the highest vertical accuracy. These bald areas provided control against which the LIDAR data in the obscured areas was adjusted and then substituted for the IFSAR elevations. In a second stage, the adjusted LIDAR became the control and the remaining IFSAR elevations were adjusted and substituted. All the processing was done at 5 meter postings and the finished product was posted to 5 meters.
A prescriptive surface was developed for use in adjusting the full resolution LIDAR datasets. This surface was the difference between the original LIDAR surface and the fused LIDAR. The prescriptive surface was used to adjust the full resolution LIDAR surfaces.
The resulting surfaces were then tested against 180 vertical control points. These check points were established in accordance with ICSM guidelines. The results demonstrated that the error in the LIDAR was dominated by mean error bias. The IFSAR data error is random and not dominated by bias. The results also demonstrated that the DTM is improved in the fusion process. The fused DTM is superior in accuracy to both the original LIDAR DTM and to the original IFSAR DTM. The following chart shows the results of the error analysis.
One of the assumptions important to the viability of IFSAR/LIDAR fusion is the high level of internal consistency within the IFSAR data. This has been proven with various data validation studies.
The conclusion from this demonstration project is that the fusion of IFSAR and LIDAR is a very viable technique when properly employed. It also provides a very cost effective method for generating a wide area elevation model particularly where there are areas requiring a high level of detail and other areas that do not require this detail.