The secular variation in the global geomagnetic field was analyzed in terms of the annual differences in monthly means by using the hourly mean data from 18 foreign (outside China) observatories of the World Data Center (WDC) for Geomagnetism from January 2010 to January 2020 as well as 9 observatories in the Geomagnetic Network of China from January 2015 to April 2021. In addition, according to the correlation of noisy components from the observatories, a covariance matrix was constructed based on residuals between observations and the CHAOS-7.4 model to remove external contamination. Through a comparison before and after denoising, we found that the overall average standard deviations were reduced by 29.97% in China and by 41.4% outside China. Results showed the correlation coefficient between external noise (mainly the magnetosphere ring current) and the <italic>Dst</italic> index was 0.82, and the correlation coefficient between external noise and the Ring Current (RC) index reached 0.94. A geomagnetic jerk was globally discovered around 2018.0 on the geomagnetic eastward component <italic>Y</italic>. The jerk timing in China was around 2020.0, and the earliest one was in 2018.75, whereas the timing outside China was around 2018.0, and the earliest one was in 2017.67. This 2-year lag may have been caused by the higher electrical conductivity of the deep mantle. After more data were added, this jerk event was found to occur in an orderly manner in the northern hemisphere as the longitude increased and the intensity gradually increased as well. The variations in location of the jerk center were analyzed according to the CHAOS-7.4 model. Results revealed six extreme points distributed nearby the equator. The strongest was near the equator, at 170°E, and the strength gradually decreased as it extended to the northern and southern hemispheres. Another extreme point with the opposite sign was located at the equator, at 20°W, in the south-central part of the Atlantic, and the strength gradually decreased as it extended into Europe. The covariance matrix method can be used to analyze data from the Macau Science Satellite-1 mission in the future, and this method is expected to play a positive role in modeling and separating the large-scale external field.