Developing a reversible rapid coordinate transformation model for the cylindrical projection
Si-jing Ye,Tai-lai Yan,Yan-li Yue,Wei-yan Lin,Lin Li,Xiao-chuang Yao,Qin-yun Mu,Yong-qin Li,De-hai Zhu
DOI: https://doi.org/10.1016/j.cageo.2016.01.007
IF: 5.168
2016-01-01
Computers & Geosciences
Abstract:Numerical models are widely used for coordinate transformations. However, in most numerical models, polynomials are generated to approximate “true” geographic coordinates or plane coordinates, and one polynomial is hard to make simultaneously appropriate for both forward and inverse transformations. As there is a transformation rule between geographic coordinates and plane coordinates, how accurate and efficient is the calculation of the coordinate transformation if we construct polynomials to approximate the transformation rule instead of “true” coordinates? In addition, is it preferable to compare models using such polynomials with traditional numerical models with even higher exponents? Focusing on cylindrical projection, this paper reports on a grid-based rapid numerical transformation model – a linear rule approximation model (LRA-model) that constructs linear polynomials to approximate the transformation rule and uses a graticule to alleviate error propagation. Our experiments on cylindrical projection transformation between the WGS 84 Geographic Coordinate System (EPSG 4326) and the WGS 84 UTM ZONE 50N Plane Coordinate System (EPSG 32650) with simulated data demonstrate that the LRA-model exhibits high efficiency, high accuracy, and high stability; is simple and easy to use for both forward and inverse transformations; and can be applied to the transformation of a large amount of data with a requirement of high calculation efficiency. Furthermore, the LRA-model exhibits advantages in terms of calculation efficiency, accuracy and stability for coordinate transformations, compared to the widely used hyperbolic transformation model.