# How can I find a representative point in 3D matrix (time,lon,lat)

I have temperature in a matrix of time, longitude and latitude. I need to find a way or criterion to find a point or location (lon *, lat *) that is representative of my entire area of interest (time, lon, lat). The idea is that by extracting the time serie from that locality (lon *, lat *), it is a time serie that is most similar to the time series of the other locations.

I was thinking of using correlation, that is, finding the point that is mostly correlated with all the other points (spatial autocorrelation is the name?), And thus get a single 2D map where I can do a contourf and see the point (or area) that has the highest correlation. The problem is that I do not know how to get a single 2D map !.

¿it is there a function in any language (hopefully R, Matlab or Python) or way to do that?

• How does a point have a correlation with other points? You may want to clarify what the data looks like and what the expected output is – Juho Kokkala Jan 26 at 7:23
• On a closer reading, it seems that you refer to the correlation between the temperature time series of different locations. However, it's unclear what this 2D map would be: there is no single 'correlation at a location', but rather a correlation with every other location – Juho Kokkala Jan 26 at 17:00
• Ok, sorry. It is annual satellite sea surface temperature, 25 km resolution from 1997 to 2017 and from 30 to 37°S and 70 to 79°W. The only NaNs are for the coast. I just want to get a point where I can extract the time series and be sure that that series is representative of the series of the other points. That is why I think that having the spatial correlation could be good idea, but I don't know how to compute for 3D matrix (each point correlated with the other points in the space create a lot of values of correlation only for that point) – Marco Sandoval Belmar Jan 27 at 16:13