English {\rtf1 Measurement issues you may be overlooking. In \par particular I am going to talk about moving boat \par ADCP discharge measurements.}{\rtf1 Here we see multiple screenshots from WinRiver \par II with what hopefully for most people collecting \par ADCP moving-boat measurements, would }{\rtf1 provide them with immediate feedback that there \par are issues with the data being collected. I would \par like to think YOU are not missing these kinds of }{\rtf1 issues, as they should be hard to OVERLOOK. In \par this brief talk I’m going to focus on issues that you \par may not notice at first glance, but are issues I }{\rtf1 have noticed fairly often being overlooked even by \par those more experienced ADCP operators. I will \par highlight a few of these so you might notice them }{\rtf1 better, at least on second glance.}{\rtf1 When I started creating this presentation I was \par trying to think of wide ranging issues I could cover \par that would be applicable for all conditions. Then }{\rtf1 the reality of how short 5 minutes is really is sunk \par in. So what I am going to focus on applies more to \par the small streams than the large rivers. However, }{\rtf1 more than half of the USGS moving-boat ADCP \par discharge measurements are made in cross \par sections with mean depths less than 10 ft! So my }{\rtf1 focus today is on something that tends to be more \par significant in smaller streams, but does affect all \par ADCP moving boat discharge measurements.}{\rtf1 So… at first glance here… nothing really jumps \par out at you, especially compared to the \par screenshots from the measurement with the }{\rtf1 invalid data before. Most of the data collected \par here looks to be valid. So what might we be \par missing that could affect the quality of the final }{\rtf1 discharge?}{\rtf1 The answer… the potential for errors that could be \par in the unmeasured areas. In this screenshot I have \par changed the x axis to length and adjusted its }{\rtf1 scale so it includes the distances to the edges. \par We can see that the left edge is still relatively \par deep and relatively fast for this cross section. For }{\rtf1 the left edge this results in a Q being almost 14% \par of the total discharge. How accurate is this \par estimate for the edge? Well, it's really difficult to }{\rtf1 say without additional external observations made \par and noted about the edge conditions and why you \par couldn’t actually get the ADCP closer to that }{\rtf1 edge. }{\rtf1 In the USGS we have policies designed to get \par people to pay close attention to these edge \par discharges…. We ask people to always check }{\rtf1 that the edges seem reasonable and if any edge \par discharge is more than 5% of the total discharge \par that they verify the edge discharge using some }{\rtf1 secondary method.}{\rtf1 In addition to the edge discharges, the other \par unmeasured areas that can affect the final \par discharge are the top and bottom unmeasured }{\rtf1 areas. The potential errors here could be from \par either an incorrect ADCP draft or incorrect \par extrapolation methods. We are looking at the }{\rtf1 same measurement that had the 14% of the total \par discharge estimated in that left edge. With this \par measurement loaded in the USGS Extrap }{\rtf1 software we can see there may be some potential \par for bend-back of the velocity profile near the \par surface and perhaps a larger than typical power }{\rtf1 exponent near the bed. How do we know what is \par correct? Well, if the data is fairly noisy, or the \par profile may be changing across the channel, it }{\rtf1 can be difficult to determine. In this case the \par difference between a standard power profile \par profile velocity and one using constant-no-slip is }{\rtf1 almost 7%. If you are uncertain about the \par extrapolation method and it makes a significant \par difference we would rate the measurement of }{\rtf1 a lower quality. }{\rtf1 Back to edges briefly, as I mentioned we ask for \par our users to pay close attention to the edge \par discharges and if any edge discharge is more }{\rtf1 than 5% to get a secondary estimate. However, \par here is an example were the discharge is only \par about 3 % of the total flow and if you look at the }{\rtf1 measured data the estimated Q seems \par reasonable. When we take a look a the notes \par made during the measurement, however, we }{\rtf1 notice an issue… the notes say there is a large \par eddy at the left edge of the water… which should \par mean a flow reversal, but when we look the edge }{\rtf1 estimates they are positive for the transects… so \par this edge estimate might be adding in discharge \par and could result in error that is more than 3% of }{\rtf1 the final discharge reported. }{\rtf1 One final tip related to the unmeasured… even \par though the ADCP collects data at a high rate and \par we get lots of ensembles in our transects, those }{\rtf1 ensembles are still just snap shots and that data \par is used to compute the discharge from the time of \par any given ensemble back to the previous good }{\rtf1 ensemble. The best way to ensure this \par computation is correct is to maintain a consistent \par speed and direction as you move across the }{\rtf1 channel. }{\rtf1 The error I often see especially on the short cross \par sections is to take a long TIME to move across \par the channel. Taking a long TIME to move across }{\rtf1 the channel can often result in the boat moving \par erratically as you try to move slowly. In many \par cases it is often better to go faster so that you can }{\rtf1 maintain both speed and direction and have that \par smooth motion.}{\rtf1 Hopefully I have reminded you that careful \par attention needs to be paid in the field not only to \par what you ARE measuring, but ALSO to what you }{\rtf1 are NOT measuring! Thank you}