Abstract

To adjust the geodetic networks effectively by computer, people often apply the algorithm of recursive/sequential adjustment. Of which the most commonly used is the Q algorithm. The Q algorithm maintains iterative computation on the symmetric variance-covariance matrix, which is considered to be sometimes difficult in numerical computations and reduces the accuracy of adjusted results. This paper has studied square root algorithms with better numerical stability and applied them to geodetic network adjustment. Based on the derived algorithms, we have designed MATLAB programs to use the least amount of computer memory, and use an illustrated example of a leveling network to demonstrate their correctness.