geoshow versus pcolorm versus meshm (and versus contourfm)

I have a geophysical dataset on a [lat,lon] grid. I would like to plot a map (such as with pcolorm, meshm) that shows the correlation between a time series at each grid point and another time series. However, as I understand pcolorm and meshm use the [lat,lon] coordinates as cell vertices rather than centers. To avoid this, I tried geoshow instead, and compared the results to shifting [lat,lon] manually by half the grid cell width. The results are not the same (boxes are slightly shifted). Also, I cannot use contourfm, which should assume centered [lat,lon] points, because my data cannot be smoothly interpolated. Please advise me on which function is correct. Here is an example.

figure; worldmap([0 75],[-85,0]); colormap('jet');

[lonsmesh,latsmesh]=meshgrid(lons,lats);

%pcolorm(latsmesh-2.5,lonsmesh-2.5,coeffs); geoshow(latsmesh,lonsmesh,coeffs,'DisplayType','texturemap');

hg=geoshow('landareas.shp', 'FaceColor', [0.5 0.7 0.5]); load coast;

caxis([-1 1]);