DBSCAN with huge number of points

We have a dataset of approximately 30 million new points per month (latitude, longitude, and a measurement associated with that point).

So far, we have applied DBSCAN to generate a monthly map such that the plotted points are:

• A "centroid" for each group generated by DBSCAN (calculated by us from the points of each group).
• A centroid color equal to the mean of the measurement associated with all the points used to calculate that centroid.

The problem is that from now instead of making a different map each month with the new data, we want to make a map that includes the new data.

Therefore, I have one question:

Would it make sense to apply DBSCAN using both the new data from one month plus the previously calculated centroids, as if these centroids were a raw dataset? (would be an incremental approximation)