Fast approximate viewshed analysis based on the regular-grid digital elevation model: X-type partition proximity-direction-elevation spatial reference line algorithm

Abstract

Viewshed analysis using regular-grid digital elevation models (DEM) is the basis of many analysis applications in geographic information systems. However, XDraw and reference plane, which have, until recently, acted a foundation of many viewshed analysis methods, have problems with accuracy and error-point aggregation. The proximity-direction-elevation spatial reference line (PDERL) algorithm, which is twice as slow as XDraw, has no accuracy problem, but not all applications can sacrifice speed for absolute accuracy. This study developed an “X-type partition PDERL” (XPDERL) algorithm based on PDERL by adjusting the partition mode of the PDERL and the combination mode of its partition results to maintain or even exceed the computational speed of traditional approximate fast algorithms while improving accuracy. The computational speed of XPDERL is stable at elevated heights from ground, slightly faster than XDraw and slightly slower than the reference plane algorithm; however, at lower elevations, it is significantly faster than both, especially in mountainous areas near the ground. In addition, the algorithm does not produce false-negative errors (identifying visible points as non-visible points) and can significantly reduce the error rate and degree of error-point aggregation. XPDERL can effectively alleviate the longstanding contradiction between speed and accuracy in viewshed analysis algorithms while providing a possible means for accurate and reliable large-scale viewshed analysis.