An average person who walks down the street would not notice these small tremors, which vary in magnitude between -2.0 and 1.7.
Even researchers who study earthquakes had difficulty detecting these "hidden" earthquakes.
Background noise, such as shaking car traffic or building construction, appears in seismic data and can mask the signals of smaller earthquakes.
To detect these evasive tremors, the scientists used a technique called template matching. From seismic data of known earthquakes, they identified patterns as to how the signal of an earthquake should be. Using this information, the researchers then scanned seismometer records to find small earthquakes.
The template comparison method has been used in seismology since around 2006, but was limited mainly to the analysis of small data sets over a couple of weeks.
"The computational load of using this method is great," said Zach Ross, the study's principal investigator. "It requires large computers, which limit their use to small pieces of data.When we started with this project, we wanted to apply this on a scale … significantly larger than anything that has been done before."
Scientists now have a lot of data to learn
The expanded data on earthquakes will help scientists understand how the swarms of earthquakes and previously unknown earthquakes that precede large earthquakes evolve.
"The reason we are interested in smaller earthquakes is because we do not have enough earthquakes in the file to observe the long-term evolution of earthquakes and faults," Ross said. "The (smaller earthquakes) begin to fill all the gaps between the largest."
Marine Denolle, a seismologist at Harvard University who was not involved in the study, said she is excited about future studies that will generate this new data.
"This is the largest earthquake catalog of all time," Denolle told CNN. "This will help us find where the earthquakes come from, and they can highlight new fault systems that we could not see before and reveal new tectonics that we did not know about before."