When dealing with gridded map coordinates, the grid will either be rectangular in degrees or else rectangular in some physical distances. If it is rectangular in degree, then the angular distance covered by each cell is the same, but the physical distance might not be the same. For example, the cells might be consistently 1/10th degree longitude apart everywhere, but if the top is at (say) 60 degrees north and the bottom is at the equator then the physical distances between the east and west edges might be fairly different between the top and bottom.
If you have consistent angles, then typically it becomes easy to figure out which cell a particular sensor reading falls in --
long_bin_number = floor((sensor_long - west_edge_long) / long_width_of_bins) + 1;
and likewise for latitude. Or use discretize . This technique works as long as the bin calculation for one direction never depends upon the current coordinate in the other direction. If, though, what you have is a grid over physical distances (for example, the bins are 750 meters tall by 500 meters wide) but the sensors locations are in terms of lat and long, then often the easiest way to move between lag and long and bin numbers is to use histcounts2 and use the output bin numbers. Once you have latitude bin numbers and longitude bin numbers for each sensor reading, you can proceed in a variety of ways. The older (and still quite efficient) way is to use accumarray which works very well but can be difficult to understand the first several times you use it. These days functions such as groupsummary might be considered as an alternative. 
