Pointing performance of a radio telescope antenna is important in radio astronomical observations to obtain accurate intensity of a target source. The pointing errors of the ALMA ACA antenna are required to be better than 0.6 arcsec rss, which corresponds to 1/10 and 1/20 of the field of view of the ALMA ACA 12-m and 7-m antenna at 950 GHz, respectively. The pointing verification measurements of the ACA antenna were performed using an Optical pointing telescope (OPT) mounted on the antenna backup structure at the ALMA Operations Site Facility at 2900m above the sea level. Pointing errors of these OPT measurements contain three different origins; originated from antenna, originated of atmosphere (optical seeing), and originated of OPT itself. In order to estimate pointing errors of the antenna origin, we need to subtract the components of optical seeing and OPT itself accurately, while we need to add components that cannot be measured in the OPT measurements. The ACA antenna verification test report demonstrated that all the ACA 7-m antenna meets pointing specification of ALMA. However, about one-third of datasets, values of estimated optical seeing is larger than measured pointing errors. We re-examined a procedure to estimate optical seeing, by investigating the property of optical seeing from the high-sampling OPT pointing measurements of long tracking a bright star for 15 minutes. Particularly, we examined the relation between optical seeing and sampling rate derived from Kolmogorov PSD. Our analysis indicated that the optical seeing at ALMA site may have been overestimated in the verification test. We present a new relation between optical seeing and sampling rate proportional to average wind velocity during measurement. We used this new relation to derive the optical seeing and as a result the number of datasets becomes half in which the optical seeing is larger than measured pointing errors. As a result, we successfully develop a new verification method of optical seeing that has high reliability.