Principles of Modern Radar - Volume 2 1891121537

7.6 Operational Considerations 325cross-track (down-range) as well, if a deep swath is desired. A CRP is designated foreach subscene, and a subset of the measured data is sequestered for spotlight imaging ofeach subscene. The resulting set of spotlight images are stitched together to form the fullstripmap image.The spotlight images must be small enough to ensure discontinuities across imageboundaries are minimized. Also, as discussed already, ample along-track data must be processedto meet resolution-driven integration angle requirements over an entire subscene.This combination of small subscenes and sizable data subsets means there is considerableoverlap between the subsets applied to different subscenes. In other words, measurementson any given pulse are applied to the image formation of at least two subscenes. The applicationof each measurement to several spotlight images hints at some of the inefficienciesin this approach as compared to a dedicated stripmap algorithm.There is a rich history of partitioning a long stripmap scene and imaging each ofthe subscenes with overlapping blocks of data. For example, this approach was used toovercome some of the limitations to SPECAN (DBS with AD) discussed at the end ofSection 7.3.3. The step transform processing method [3] applies short, overlapped blocks ofdata to DBS-AD to generate overlapped, coarse-resolution subimages along-track. Thesesubimages are registered along-track with respect to one another, and a second Fouriertransform is applied through the subimages to realize the full, fine cross-range resolutionpotential of the collection.Finally, stripmap data acquisitions may be applied to the backprojection image former.As discussed in [1], backprojection is the brute-force, space-time domain realizationof the generalized matched filter in (7.50). Backprojection garners no computationallyefficiency through FFT processing, so image formation may take much more time thanRDA and RMA. On the other hand, backprojection accommodates nonlinear and nonuniformlyspaced acquisitions, while RSA, RDA, RMA, and all DBS methods require linear,uniformly spaced sampling along-track. Therefore, backprojection may be the only imageformation option given highly unstructured collection geometries. For example, datacollected by BoomSAR, a ground-based low-frequency SAR system, suffered from severemotion errors generated by the swaying of the extended boom upon which the radarantenna array was mounted [16]. While the nominal vehicle motion was linear, large alongtrackand cross-track antenna motion precluded Fourier-based image formation, leavingbackprojection as the only suitable imaging algorithm for this stripmap collection.7.6.6 Squinted OperationA stripmap collection was defined in the Introduction to this chapter by three attributes.Two of these properties, the platform moves in a straight line and the antenna beam patternsare not steered during the collection, will continue to hold. The third, radar antennas arepointed to broadside, will be relaxed in this section.In the context of stripmap SAR, broadside means normal to the platform velocityvector defined with respect to the imaged ground swath. Operation with transmit andreceive antenna beams pointed away from broadside is termed squinted. Squinted stripmapmay occur for a number of reasons.For instance transmit and receive beams may be mechanically or electronically steered(and fixed to an angle) either forward or aft of broadside. Small squint angles offer no significantadvantage or disadvantage with respect to cross-range resolution or area coveragerate but serve to complicate image formation. In this situation the need to squint is often