Since its first experimental demonstration in 1995, ghost imaging has attracted a great deal of attention due to its potential applications. In Yanhua Shih’s paper (Meyers R E, Deacon K S, Shih Y H 2011 Appl. Phys. Lett.98 111115; Meyers R E, Deacon K S, Shih Y H 2012 Appl. Phys. Lett.100 131114), he pointed out that " one of the useful features is the turbulence insensitivity of thermal light ghost imaging, i.e., atmospheric turbulence would not have any influence on the ghost images of sunlight”. However, in Jeffrey H. Shapiro’s view (Shapiro J H 2012 arXiv:1201.4513 [quant-ph]), lens-less pseudo-thermal ghost imaging is not immune to spatial resolution loss from the presence of atmospheric turbulence along the propagation paths, unless the source diameter is less than the source-plane turbulence coherence length. In the present paper, we find that the second order self-correlation of sunlight intensity is the case that can be satisfied with both theories of Shih and Shapiro. In this paper, the second order self-correlation of sunlight intensity (the intensity variance of the images acquired by the CCD camera), rather than the total intensity correlation between images and bucket detector signals (the traditional ghost imaging method), is calculated to recover the high-quality images in turbulent atmosphere under a few hundred measurements. Inspired by Boyd’s paper, titled " thermal ghost imaging with averaged speckle patterns” (Zerom P, Shi Z, O’Sullivan M N, Chan K W C, Krogstad M, Shapiro J H, Boyd R W 2012Phys. Rev. A86 063817), turbulence insensitivity of sunlight ghost imaging is demonstrated both theoretically and experimentally in this article. It is found that thermal ghost imaging system whose coherence time need not to be controlled to match the speed of the detectors, ghost imaging with sunlight, which has always been considered intriguing and highly desirable, now is realizable. We present theoretical and experimental results showing that a sunlight self-correlation ghost imaging system can produce high-quality images even when it uses an slow detector and passes through the turbulence near ground, as long as the signal variation is predominantly caused by the fluctuation of the sunlight intensity rather than other noise sources. Our scheme can also be used to improve the image quality in other wave bands such as infrared and ultraviolet, in the case where an poor image quality results from the turbulence or other random disturbances on the wavefront.