It might be easiest to not deal with datetick for this particular issue. Instead, set tick values manually like this:
% one-second intervals from 10:53 to 11:05:
t = datenum(2014,3,15,10,53:1/60:65,0);
x = rand(size(t));
% Plot using datetick:
subplot(211)
plot(t,x)
datetick
% Enter times you want to label here:
labeltimes = {'10:57';'11:01';'11:03'};
% Get datenums of times you want to label:
labeldatenum = datenum(strcat('March 15, 2014, ',labeltimes))
% Plot with manual labels:
subplot(212)
plot(t,x)
axis tight
set(gca,'xtick',labeldatenum,'xticklabel',labeltimes)
Other tips for making your figure more clear:
1. Do not use colormap(jet). The jet colormap is not terribly intuitive for humans to interpret and it puts a weird bright spot arbitrarily about 60% of the way up the scale. The bright spot makes the yellow seem like the most important data. I recommend using the parula colormap if you have a new release of Matlab (or paruly for old releases). Or experiment with Cynthia Brewer's colormaps or brewermap. For data that goes in one direction use a sequential colormap to make it clear that this dataset goes from less to more. For data that diverges (if your velocity scaled from negative values to positive) use a diverging colormap.
2. Place the colorbar horizontally (e.g., with colorbar('location','south')), which will make your plot shorter in height and wider because the human eye likes slightly more landscaped images. This is especially important for time series which we want to see plotting along in time rather than plotted as a simple square.
3. There's an interesting peak in your data. Label that time explicitly.
