The quality of big (geo)data

Abstract

Big data is distinguished by volume, velocity, and variety. A large proportion of all big data is likely to be geographically referenced, and much may be real time. While examples can be found of high-quality big data, problems arise in meeting the normal scientific standards of replicability and rigorous sampling. These standards can be relaxed in certain stages of science, during hypothesis generation and exploration. Three methods of quality improvement and assurance are proposed. Only the third is sufficiently robust and rapid, especially in time-critical situations.