So, as per my first message, I really do recommend using a different tool. Firstly, you’re using an all-atom force field but gmx order doesn’t take the positions of the actual hydrogen atoms into account when doing the calculation (it assumes you’re using a united-atom force field). Secondly, it averages the two different hydrogen atom (pro-R and pro-S) order parameters. Finally, it won’t be able to calculate the terminal methyl group order parameters. Have a quick look at the paper I linked to, it mentions many different tools to do the all-atom analysis. In fact you can quite easily write something yourself to do the all-atom analysis (e.g. I’ve done it before within MDAnalysis).
As to your plot, the overall shape looks fine (bar the unsaturated bits not being right, as expected). If you could calculate the order parameters for the unsaturated carbons they would create a drop at atoms 7 and 8 in your plot. I can see from your naming that you’re simulating POPG, so you should expect the drop primarily at carbon atoms 9 and 10 with 16 points in the plot; there should be 16 data points as the oleoyl tail is 18 carbon atoms in length but carbon 1 is the carbonyl carbon and, as mentioned, gmx order cannot calculate the order parameters for carbon 18.
I think there are two issues are going on with your analysis. Firstly is that the labeling on the x-axis should start at 2 not 1. And secondly, I don’t think you are calculating the order parameters for carbon 2 in this tail (probably just an index group missing). In other words, I believe the data point at atom 1 is actually for carbon 3. But, as I said, you’re better off just using another tool as fixing these still won’t sort the unsaturated groups out.