Developing and Applying Shift Curves Using GRSAT

Video Transcript
Download Video
Right-click and save to download

Detailed Description

This video is a demonstration of how to develope hydraulically sound shift curves and apply them hydrologically in time using GRSAT. Shifts are developed and applied for an entire water year.


Date Taken:

Length: 01:10:59

Location Taken: Salt Lake City, UT, US


Hello, my name is Terry Kenney, and I'm a
surface-water specialist with the Water Science

Field Team.

I'm going to demonstrate how to develop hydraulically
sound shift curves for a base rating.

And how to apply those shift curves in time

This demonstration is going to be slightly
different than some of the previous ones you've

seen in this series.

We're going to be using GRSAT.

And we're going to be acting as though we
are working record up operationally in a continuous


I have already opened up GRSAT and loaded
the shift demo XML file.

And you can see we have a multiple offset
rating comprised of 3 straight line segments

connected by smooth transitions at the breakpoints.

We have a breakpoint at 11.4 feet of stage.

And a second breakpoint at 36.1 feet of stage.

As I indicated just a minute ago, we are going
to be working on shift curves in an operational

sense very similar to continuous record processing.

And so with that in mind, we are going to
turn off all measurements except for the very

first one, measurement 665.

So you see they're grayed out.

And I'm going to toggle off the Show-Hide
unselected measurements.

And they're gone now.

So we're going to pretend that we are getting
these measurements sequentially post field


And developing hydraulically sound shift curves.

And then applying them hydrologically in time.

And then we'll move onto the next measurement
as if we just collected some new data.

And we have hydrographs to look at and that
sort of thing.

We have measurement 665 made on October 16th
that we need to see whether we need to apply

or develop a shift curve for.

So let's go into the variable shift diagram
and zoom into that measurement.

And as we can see, the error bars of that
measurement do not overlap with the rating

curve as represented by the 0 line.

Now, we have a previous shift curve brought
into this record working period that starts

on October 9th.

And it has already defined some sort of a
shift curve strategy with a hinge point.

Shown here at 10.7 feet of stage.

And a merge point at 11.4 feet of stage.

Which you can see conveniently corresponds
to the breakpoint.

That location in which we are identifying
a change in offset.

And therefore a change in control segments.

So what's obviously happened is someone's
developed a strategy of having some sort of

a hinge point at this gauge height.

And obviously the merge point at this gauge

What we'd recommend folks to do is select
a hinge point -- that gauge height in which

the shift curve begins to work its way towards
the rating curve -- to be somewhere high up

in the area of the rating curve segment.

Somewhere up in here.

To which for most shifts that you would expect
would be a smooth transition into the merge

point at 11.4 feet for this example.

So what we need to do is define a hydraulically
sound shift curve.

Because someone's developed the strategy,
brought it into this record working period.

We are just going to clone this existing shift

And of course I want to do that.

I say yes.

My default GRSAT will assign a start date
of one day past the shift curve that you are


While I don't know exactly when I want to
apply this shift curve that I'm developing,

for the sake of starting out and defining
the shape of the shift curve, I typically

start it with a -- I begin it with a start
date of the measurement time.

So measurement 665 was made on October 16th.

So I'm going to adjust this to October 16th
at 12:45.

Type that all in.

And then I can add a remark.

And I'm going to put something very simple,
not too descriptive.

But it's for measurement 665.

We would like to see a more descriptive remark.

Something more specific to the control and
what's going on.

If it was something related to scour or to
fill, it'd be nice to put that in the remarks


I'm also going to change this color because
the background of my graph here is green.

So I'm going to make this one blue just for
the sake of simplicity and for viewing it.

So I'm going to hit apply.

And now I have a new shift curve, which is
exactly the same as the previous one.

So I have these inputted gauge heights already
set up for me.

I'd probably like to adjust them to get the
shift curve to go through this measurement

as best as I can.

So I'm going to zoom in here just to make
things easier to see.

And you can see that a shift of about minus
0.05 is probably going to be about what I

want to apply for this measurement.

So I have these input points already set for

I'm happy with them I think.

And I can just toggle these radio buttons
to go up by whatever the step in increment

is here.

I typically want to set the step here to be
100th of a foot.

So I'm going to change that.

And then I'm going to move these over -- at
least the first one -- until it gets to about

where I want it to be.

And I said .05 minus .05 is probably where
I want to be.

And then I'm going to do the same thing for
the second input point.

Which represents the hinge point.

That gauge height where the shift curve begins
to transition back to the rating at the merge

point right up here.

And then obviously my merge point, 11.4, is
going to be 0.

That's where the shift becomes the rating.

And then for all stages above that we are
on the rating as well.

So the shift curve looks pretty good.

I mean I think it looks good on the variable
shift diagram.

We've shared the same input points.

But what's really important is not that it
looks good on the variable shift diagram.

It's that it looks good on the rating curve

Particularly that the slope of the shift curve,
which becomes the shifted rating at the time

of application, mimics the base rating form.

Has a very similar slope.

It's a very smooth transition back into the
rating and that sort of thing.

So I'm going to zoom into the segment here
that I'm interested in.

The section control segment.

And I can see my shift curve is this blue

And it's smoothly going to zero at this breakpoint
of 11.4 feet.

Again, I could zoom in a little bit closer
just to make sure nothing is drastically changing.

And from what I can tell here, it's not.

And I would expect that after looking at what
my selected input points were in the variable

shift diagram and that sort of thing.

Our shift curve is hydraulically sound.

That is, it mimics the base rating, stays
true to the control that is being affected

by whatever is causing the shift.

And now it's time to define how we want to
apply this shift curve in time.

And that would be how it would be done by
looking at the hydrograph and thinking about

the hydrologic application of the shift curve.

So GRSAT has the time series view down here
where the last process to computed unit value

discharges are made available when you pull
them out of the database.

And you can see the hydrograph here.

Now, remember, we are working on this operationally.

Let's pretend we don't see any of the hydrograph
past this measurement because we wouldn't

have that kind of information available to

So we'll zoom into this time period just slightly
past the measurement.

And so this is typically what you would be
dealing with in an operational sense.

So what we can see here is we have a small
little event that happens and then we have

a measurement that's made at the recession
or very near the bottom of the recession itself.

And coming into this record working period
someone has decided to input a 0 ship at the

peak of this event.

So we can go with that information.

And then seeing that our shift associated
with measurement 665 is some sort of a fill

shift, right, it's a negative shift, then
we probably have some filling that happened

on the control during the recession of the

And so applying the shift associated with
measurement 665, we'd want to apply it at

the end of the recession essentially, right.

We have an amount of material that has filled
in on the control as energy dropped during

the hydrograph recession.

And we go from there.

And if you look at this measurement itself,
it's made pretty much at the bottom of the


So for this instance, it makes sense to probably
start or input this shift on the start time

or end time of the measurement, measurement

And we've already done that, right.

We cloned the previous shift.

And then we put a start date of the measurement

Now, I want to show you some functionality
associated with GRSAT in the time series window

and dealing with shifts and the application
of shifts in time.

Looking at the shift bar down here.

Right at this middle section with these orange
boxes essentially.

We have an orange colored box here with a
line beginning at the start time of that previous


The shift applied on October 9th corresponds
to this line right here.

This is the start time of that shift.

And similarly the line shown right below measurement
665 is the start time we assigned to that


Now, it's important to understand there's
two different colors of boxes that correspond

to how shifts are behaving between one another.

And these two obviously are the same color.

The same color orange indicate that the two
shifts are prorating into each other.

And so this start date -- this shift that's
inputted here -- is input at this point in


And then it prorates across till it gets to
this other line.

And then similarly this inputted value prorates
to whatever gets input later on.

Now, I said there's two different colors.

When you put an end date on a shift, you get
a different color orange down here.

And let me show you by putting an end date
on this first shift.

So I'm going to highlight this shift.

I'm going to double-click it.

And I am going to just arbitrarily put an
end date of -- let's go 10-15-2008; hit apply.

And you'll notice that the rectangle here
became a different color orange.

And you will also notice that there's a gap
or a white space between this end date and

this start date.

And that is to illustrate the fact that when
you put an end date on a shift, what you're

doing is whatever the value you inputted for
the shift on this date is the same all the

way across this box and goes to this end date.

And then after that end date, no prorations
happening till the next start date.

Essentially it's a null shift.

It's a 0 shift value going on here.

And so you have to keep that in mind if you're
going to be ending a shift and then starting

one right after that end date -- end date
and time.

You could potentially have a jump in unit
values because there's no proration happening.

So keep in mind what end dates do and what
they mean.

And what's nice about GRSAT is you have a
different color box down here in the shift

time series to see what's going on.

Something else that's nice to know or nice
to be able to do inside GRSAT is the ability

to adjust these dates graphically.

So I could click on this adjust date button

And I could click on this shift curve associated
with the start date of 10-9-2008.

And I can grab it and move the start date
anywhere I want.

See that?

And it gives you the times.

Pretty nice.

And I think we had it starting on 10-9 at

I can adjust it here if I didn't like the
way it worked.

Hit apply.

And then I can grab this end date.

So I can adjust the end date.

So if you look here, the end date pops up
up here, right.

And I can move it back and forth to wherever
I want to go.

I can slide it right up against the start
date of the next one.

But keep in mind, because there's a different
color orange here, there's no proration happening.

So let's go back and let's remove this end

Because I don't want to put an end date on

I have something happening hydrologically
that I'm going to blame this shift on.

Some sort of a fill during the recession.

So I'm going to turn off the end date; it's

Hit apply.

And the color goes back to the way it was.

And so when you see this color, remember,
whatever the inputted shift value is, the

front end of this box prorates to the shift
value that's input on this next box over here.

So getting back to our example, because this
measure was made essentially at the bottom

of the recession, I think I'm happy with starting
the new shift curve on that measurement.

Some other neat little tool the GRSAT has.

You can see how things are prorating from
one shift curve to the next shift curve.

If you click on this little time slider clock
right here, you have the ability to move this

line on the time series.

Which corresponds to being able to view what
the shift curve is -- or the shifted rating

is -- at that point in time.

So as you can see, I have a 0 shift that I've
input right here on October 9th.

The purple line is on the 0, just like what
was input.

And I can see over time how that 0 shift curve
is prorating out towards this minus 0.06 shift

curve that I've defined.

So I can see how the application of my shifts
and how things are prorating, if it's agreeing

with how I'm thinking the event or the cause
for the different shifts is happening.

So I can always use this to make sure that
my shift curves look good on the variable

shift diagram.

I can look and see how they are plotting on
the rating as well.

So I can zoom into the rating.

And I should be able to see that same purple
line in here.

Although everything's really close.

You can see that there's the blue line.

And then there's a purple line in here that
does change as I adjust that time slider.

You can see it's closer to the rating and
the red line there.

So I'm going to turn off that time slider
by clicking on it.

And we are ready to move onto our next measurement.

So let's zoom everything out, get rid of our

Autoscale everything.

And add our next measurement that we collected.

So now we have measurement 666 made on November

And what we can see is it was made at a gauge
height slightly lower than our last measurement.

And it plots in a fairly similar fashion,
at a minus 0.04 shift.

Now, what we can see is the error bars from
measurement 666 do cross over the shift curve

that was defined by measurement 665 and applied
at the time of measurement 665.

So we could just leave that shift curve active.

Because remember, when we have error bars,
we are assuming that the measurement is anywhere

in between those error bars.

But for the sake of illustration, I'm going
to develop another shift curve just to show

you how it's done and talk about how things
are applied in time.

The thing about working up records continuously
is you don't have hindsight.

You don't have the ability to draw a single
shift curve that splits multiple measurements.

Because, you know, they're all plotting kind
of in a similar fashion.

And typically you're going to have to develop
shift curves from one measurement to another


And then look at the hydrograph and apply
them in whatever fashion you need to.

So, again, I'm going to take advantage of
the cloning.

Remember, we have a nice strategy set up here.

And when we have multiple shift curves for
the same control, we want to share those hinge

points and merge points so that we have smooth

And we don't have shift curves crossing one

That'd be like having radiants crossing one

So it makes for a nice smooth way of prorating
from one shift curve to the next shift curve.

I'm going to highlight that measurement made
on October 16th.

I'm going to hit the clone tool.

And of course I want to clone it.

And remember, it's going to just by default
assign a date of one day later than the shift

that was cloned.

Again, we're not looking at the hydrograph

We're just getting started on developing the
shape of the curve.

And for that matter, I'm just going to assign
the same date as the measurement when it was


So 11-20.

I can use these radio buttons, bump it up
at 12:35.

And then of course when you clone a shift,
you're going to get the exact same remarks

from the previous one.

So keep that in mind.

I'm going to change this to my real simplistic

Again, I would like to see more.

When you are working stuff operationally,
give some cause and effect in the remarks.

Color is orange, which is going to work out
just fine.

And I'm going to hit apply.

And you can see looking at the variable shift
diagram it's the exact same shift as the previous


It's highlighted so now it's orange.

But if I clicked on the other one it would
be blue.

But it's the exact same shift curve.

And once again, we have already defined hinge
and merge points.

And it's just a matter of adjusting the amount
of shift value for the next measurement.

Now, I can use these radio buttons that I
used before.

But I also have some graphical tools down
in the variable shift diagram that are really

useful as well.

I have the ability to click on this button
that moves the shift point.

Now, that means I can grab the shift point
and move it anywhere: up, down, side-to-side.

Now, because I already have some established
inputted gauge heights, I don't want to do


'Cause I would have to, you know, fine-tune
the gauge height value as well as the inputted

shift value.

Now, I have these other two tools to the right
of the move shift point.

One that moves the shift curve only horizontally.

You can't move it vertically, which is what
I probably want to do here because I don't

want to move the shift point beyond or different
from the inputted gauge heights.

And then likewise, I have one that I can just
move it vertically.

And that's not what I want to use here because
I want to adjust the shift amount.

That would be the way to adjust the inputted
gauge height.

So I'm going to click on the move horizontally

And I can grab any one of these shift points.

I'm going to grab the bottom one first.

I'm going to slide it over to this minus .04

I'm going to grab the other one and move it

And you'll notice when I go up and down it
doesn't move at all.

And so what I've done is I've maintained the
same inputted gauge heights and just adjusted

the shift value.

And once again, I want to have the same merge
right here.

Now, I can show the previous shift curve or
any other shift curve on the variable shift

diagram just by clicking on the box.

And it will show up as a dashed line corresponding
to the same color that's shown on the number

right there.

And so I could highlight this one.

It'll be solid.

I could highlight the lower one.

It'll be a dashed orange line.

Pretty handy stuff.

Once again, I believe I've developed a hydraulically
sound shift curve by sharing the same input


And it looks good on the variable shift diagram.

But it's always important to look and see
how that shift curve looks on the rating curve


And same as I saw before, the lines look really

The slope is very similar to the base rating
and the previous shift curve.

And they're really close to each other.

Hard to kind of zoom in and see major differences.

But I can tell that it's merging nicely into
the breakpoint at 11.4 feet.

So I am happy again with the shape of the
curve, the hydraulic shape of this shift curve

that I've put in here.

And now, once again, it's time to look into
the time series view and decide how I want

to apply that in time.

Let's zoom in again to the area we're interested

We only get to see so far ahead of us because
we're working operationally here.

And what I can see is nothing hydrologically
has happened between measurement 665 and measurement


We have a very flat hydrograph going on here.

There's really no event -- there are no events
to deal with or to assign any responsibility

for changing the hydraulic control.

And so at this point, because it's flat, I'm
just going to allow that previous shift of

minus .05 feet to prorate smoothly into a
shift of minus 0.04.

And I am going to start this new shift on
measurement 666 because there's nothing happening

in between these measurements.

And so you'll see I have a start date, once
again, October 9th, October 16th, and then

November 20th right here.

Now, if I had wanted to, I can adjust that
start date for this last shift.

Or any of these shifts for that matter.

And I didn't really show you how to do that

So I can click on this adjust date button
and I can highlight a box.

And I can move it in time.

And because these are prorating into each
other, right -- there's no end dates -- there's

no gaps or anything.

It's just defining where the proration goes
from one to the other.

And so I'm going to pull this back to that

And I am happy with that.

Let's move along and do another quick field
run and get measurement number 667.

So I'm going to autoscale everything.

Go back to where I was.

Unzoom stuff.

And look at that.

We just got back from our field run and we
obtained measurement number 667.

Made it a much higher stage and discharge
than our previous two measurements here.

So let's look at the variable shift diagram
and see whether or not we need to define a

shift or what kind of shift we're going to
need to define for measurement 667.

So I'm going to unzoom.


And I can see that measurement 667, with its
error bars, overlaps with the rating at this

0 line.

So I don't need a shift to be defined for
measurement 667 other than something that's

a 0.

Now, when you start thinking about how these
shift curves looked for the lower section

control, the shift curve merged with the rating
at the breakpoint of 11.4 feet.

And therefore the shift above 11.4 feet was

And you can see that orange line right here,
right on the 0 line.

So I could just let the shift that was applied
for measurement 666 just carry forward in


And where I'm at right now with this higher
flow is right on.

And the stage discharge relation is being
reported properly at a 0 shift.

My preference at this point would be to at
least define a 0 shift shift curve for measurement


And then look at the hydrograph and decide
whether or not I want to keep it in or figure

out a way to prorate it.

Or remove it and allow for the previous shift
just to carry forward.

And so I'm going to just insert a new shift

Keep in mind, this is at a different control
segment of the rating curve.

And so these inputted points -- this hinge
and merge point -- aren't related at all to

measurements made in the channel.

Or even the overbank portions of the rating

So cloning this would not really save me any

So I'm just going to insert a new shift.

And it's going to ask me and of course I do.

By default it's going to assign another date
from the last one, just within one day.

I'm going to change this to January 1st -- or
January 8th, my mistake, 2009.

Bump that up.

I'm going to make this a different color.

I'm going to make it, oh, I don't know, let's
go with this darker green.

And, again, this is measurement 667.

And I'm going to hit apply.

And because I just added a new shift curve,
the inputted value defaults to a .1 foot of

stage and a 0 shift.

And the other input values aren't even highlighted
because we are on the rating curve itself.

Now, at this point you could just leave that
because it's a 0 shift.

Or you could make an attempt to establish
some sort of a shifting strategy for the channel

control section of the rating curve.

For example, if this was a site that typically
has truss-shaped shifts.

Those kinds of shifts in which it has two
merge points.

One at the lower end of 0 near the bottom

And then out to a certain hinge point in the
middle of the control section.

And then merges back to 0 at the upper breakpoint.

You could kind of define that or carry that
forward from previous times.

For our example here, I am not going to change

I'm just going to keep it at 0.

And then, you know, as I look at the hydrograph
and decide whether I'm going to keep this

shift in or not, I'm going to possibly come
back and refine my input values and that sort

of thing.

So it's 0 right now.

And let's look at the time series.

We want to zoom out a bit.

And then let's autoscale it so we see everything.

And then zoom back in to our area of our time
period of interest.

Again, operationally we kind of don't get
to see forward too much.

And what we're seeing here is that we had
-- since the last measurement, measurement

666 -- the hydrograph remained fairly flat
for a little while.

And then we had an event that came and raised

And we got a measurement near the -- I guess
it would be like the second peak of this event

right here.

And so we need to figure out how we're going
to apply the shift in time.

But looking back at measurement 666, we likely
want to hold that shift curve steady during

the flat period before the event.

And so I can figure out a way to put in that
minus 0.04 shift somewhere right in here.

My best approach to doing that would be to
clone the shift that was applied on November


So I'm going to click on it and I'm going
to clone it.

I want to clone it, yes.

And for now let's just arbitrarily call this
December 21st.

And then we can say carried shift 
from measurement 666 to beginning of rise

in hydrograph.

Maybe held's the right way.

Held shift from measurement 666 to the beginning
of the rise in the hydrograph.

Oh, we'll make this one -- we'll make it orange
as well.

Hit apply.

And I just arbitrarily put a date in there
because I want to adjust this graphically

right now.

So I'm going to adjust the date of this one.

And I'm going to move it to, I don't know,
right about there.

That looks about where the rise started.

So if this was operational, if you were really
working this up, you should look in at the

actual computed time series of stage values
and see where that exactly happened.

And that's where you're going to want to hold
that shift steady.

So what's going on here is we had input a
shift associated with measurement 666 of minus


And we cloned it and input it here as well.

And so what's going on is that 0.04 shift
is being held steady.

It's not changing.

It's prorating into itself.

So it's minus 04 here and minus 04 here.

And now we have a shift that was defined as
a 0 shift associated with measurement 667

that we need to figure out how we're going
to apply it.

Shifts vary with time and with stage.

We know that above a gauge height of 11.4
-- any time after this measurement -- we're

at a 0 shift.

And so everything seems to be working out
just fine.

So the hydrograph is above 11.4 most cases

And it will be rating direct, or on the base

Just as our measurement 667 had defined.

So we have some choices.

We can leave it as it is now, where these
prorate into each other.

A 0 all the way up and down the entire range
of the rating curve prorates to a minus 04

that starts here at the lower portion of the
rating curve.

But our flow is not affected when gauge height
gets above 11.4 feet.

So we can leave it as is.

It seems like it makes sense.

Or we could also take this shift value -- this
0 shift value -- and stick it somewhere near

the top here.

And just allow that lower minus 0.04 to kind
of work its way out and become more of a 0

shift as the flow came up.

So we could do this a number of different

For now, I'm going to leave it as is 'cause
it works out just fine.

'Cause I know what I have behind me in the
hydrograph and how the shift curve is going.

Whatever choice you make as far as how you're
going to prorate one shift into the next shift,

it needs to be documented and explained in
the station analysis.

So a big portion of applying shifts in time
is describing the decisions that you make.

Why you started a shift at a certain time.

Why you prorated into a different shift and
that sort of thing.

You need to provide that cause and effect
style of narration that describes what is

going on.

Okay, now, it's time to move onto our next

Measurement 668.

Just got back from the field.

Time to develop a shift curve and apply it
in time.

Looking at the variable shift diagram, we
can see that measurement 668 looks like it

plots off to the left and is not hitting the
rating curve.

So we need to develop a shift curve again
for measurement number 668.

Now, if you look on the overview rating plot,
we can see that measurement 668 was made under

section control.

Of course, our note sheet -- our field notes
-- would indicate that as well.

And so it's probably a good idea to develop
a shift curve that has a similar form as the

previous two that we made for measurements
665 and 666.

So why don't we clone one of those shift curves.

The one that was applied on November 20th
seems like a good candidate, right.

If we zoom down we can see that it is for
measurement 666.

So let's clone it.

And of course we want to clone it.

And let's give it a start date associated
with measurement 668, right.

So February 28th, 2009, at, whoa, 4:26 in
the morning.

Yowza, that's early.

And we'll just turn this to measurement 668.

It's going to be blue.

Let's hit apply.

And it shows up down here.

So let's work with it.

We want to zoom in down here at the bottom.

Let's get those shift amounts out towards
that measurement.

We're going to keep those same inputted points:
the hinge point and the merge point.

And we're going to hopefully see that measurement
668 fits nicely on that same sort of a curve,

just with a larger amount of shift.

So let's bump this up to where we think we're

Oh, minus 12 looks good there.

Stick the hinge point out there.

Take it up one.

Pretty close.

It's nice that the gauge height of measurement
668 is below that hinge point.

And it looks okay.

The thing to be concerned about potentially
is the fact that now this shift curve form,

with the common hinge and merge that we already
selected, the amount of shift is a lot more

than we've seen from the previous shift curves
we defined on this type of hinge and merge


So we would want to make sure, just like the
case with the other ones, that everything

is smooth on the rating curve itself and how
that shift curve is coming into the merge


So let's zoom into the rating curve.

And look at our rating zoom 1.

And it does look okay.

I mean it's certainly a larger shift down
here at the bottom.

But that's because measurement 668's plotting
that much further off of the rating curve.

And the place where we'd be most concerned,
again, would be right here where the shift

curve is starting to merge back into the rating
curve at the merge point of 11.4 feet.

And it looks good.

I mean it is out here fairly far off.

But it looks like it's a smooth transition
into the rating curve itself.

So that looks really good.

Now, we need to decide how we're going to
be dealing with applying this developed shift

curve hydrologically.

Let's look at the time series and decide how
we're going to apply our newly developed shift


So let's grab the pan button and give ourselves
a little bit more time series to view.

There's measurement 668.

And so what we have going on here is, remember,
for measurement 667 defined a 0 shift.

We input a 0 shift through the entire range
of the rating curve at the time of measurement


Shortly after measurement 667, it looks like
we had a fairly rapid recession.

A quick drop in stream flow.

And then it remained fairly steady up to measurement

Which defined that fairly large minus 0.12
foot shift to the section control.

So now we have to decide how we're going to
get from the 0 shift to this minus 0.12 feet

of shift.

And so we can look at this rapid drop in flow,

We can make some assumptions here.

Possibly blame a large amount of fill on the
section control to a sudden drop and loss

of energy associated with this rapid recession.

And then if we were to take that a little
bit further, we'd want to apply -- or at least

hold -- this 0 shift to about the beginning
of that recession.

And then we would want to hold the minus .12
foot shift back from measurement 668 to about

the point in time where the recession ended,

And so we would blame this quick drop to be
the cause of dropping material onto the section


With that, we'd probably want to take the
shift associated with measurement 667 -- that'd

be the shift on January 8th -- and clone it.

So we'll hit the clone.

Yes, we want to clone it.

And it's going to assign that next day.

Let's make this the 15th just to spread it
out a little bit.

And we will say held shift from measurement
667 to beginning of recession.

It's going to be pink, which we're fine with.

And we'll say apply.

Looks good.

Oh, wow, that's really close to where we're
probably going to carry it to.

So, again, you'd want to look at the unit
values and decide when that recession started.

Just for the sake of illustration, let's just
keep that where it's at right now.

I mean we could go in here and adjust the
start date for it if we wanted to, right.

So let's just move it just a little bit right

Okay, that works.

So what we've got going on here is we have
a 0 shift here being held steady to a 0 shift


Now, I'm going to blame this rapid drop.

So I would probably want to hold this, the
shift that we defined for 668, to about this

little point here, the end of the recession.

So I'm going to take the shift that we defined
here on February 28th.

I'm going to clone that one.

I would like to clone it, yes.

And this one I want to make it a little bit
sooner in time, right.

So I'm going to just briefly just make this
02-20 just to get a starting point.

And, again, we're going to hold our carried
held shift from measurement 668 back to end

of recession.

It'll be purple; that's fine.

We'll say apply it, boom.

So that this one right here, right.

See it's highlighted.

And I'm going to drag this to about right
where I think it goes.

Again, you'd want to look at the unit values.

But for the sake of illustrating here, graphically
that seems like a really good place to hold

it back to.

So what's going on here is we have a 0 shift
up until the beginning of the recession.

And then that 0 shift quickly prorates into
that minus .12 shift that's input at the bottom

of the recession itself.

And why don't we look at the time slider.

So let's zoom in a little bit.

Click on the time slider and see what we can

And we can see how that purple line is working
its way over to the time in which we've input


And now it becomes that shift of minus .12

It's time to look at our next measurement,
measurement number 669.

Let's turn it on.

And looking at the overview, let's undo that

Looking at the overview, it is a measurement
that was made at a slightly lower gauge height

than measurement 668.

So let's look on the variable shift diagram.

See what we need to do, if there's anything
we need to do.

And you can see that the shift curve that
we developed for measurement 668 of a minus

.12 does hit the error bars for measurement

For this example, I am going to just let it
ride and keep that shift active through measurement


Or at least utilize the same shift curve.

So I don't have to develop a new shift curve.

But I'm going to clone it.

Clone shift curve that was associated with
the measurement made on February 28th.

And apply it on the start date for measurement

Clone it, yep.

And the date is April 27th at 1315.

Little more reasonable than 4:30 in the morning.

And we're just going to change this to measurement

Hit apply.

And you can see it's the same location, just
a new color.

Let's see what kind of hydrograph happened
between those measurements.

So we'll pan over.

Oh, boy.

We had quite a bit of higher flow occur between
these measurements.

Unfortunately we don't have any measurements
at that upper end.

However, we do have some faith in the stage
discharge relation at the upper end because

our last higher measurement made previous
to measurement 668 -- which was measurement

667 -- plotted on the rating curve.

So we're not too worried about that upper
end at this point changing or being unstable.

At least that's what we can tell ourselves.

So we will hold the same shift curve from
668 at least to 669.

And we're good to go.

We cloned it and we started it on 669.

So we're okay there.

Moving right along, let's see what measurement
670 has in store with us.

So at first glance, we can see that measurement
670 was made under channel control conditions.

And at a fairly higher stage and discharge
than our previous measurements.

Let's look on the variable shift diagram to
see what we're dealing with here.

And it looks like measurement 670 plotted
off the rating by about plus 0.5 feet.

And the error bars are not crossing over the
rating 0 line shown here.

So we are going to have to develop a shift
for this channel control measurement.

When a measurement made under channel control
conditions after you've had some lower flow

measurements that plot to one side or the
other of the rating curve creates some questions

and creates some potential issues that have
to be resolved as more data gets collected.

Now, right now all we have is measurement
670 plotting off the rating about half of

a foot.

And we don't know what to expect as far as
unit values of discharge or flow moving into

the future.

Now, we have a couple of choices to make.

There are two generally accepted shift curve
forms that we would likely be able to apply

to this situation.

The first one being a half house.

Which is what we've been developing for the
section control shifts where we have a merge

point, a hinge point.

And then from the hinge point straight down
we have a steady shift amount.

Now, the other type is known as the truss
shift, which I've mentioned here a little

bit back and forth.

In which you have the shift curve is coming
off of the rating.

So it starts at a 0 shift.

Comes out to a certain amount at the hinge

And then merges back into the rating again
at a 0 point.

And typically such types of shifts are something
that can be identified with your eyes when

you're in the field.

A perfect example -- possibly a very simple
example -- would be a log over the channel,


As flow came up, the log was not affecting
the stage discharge relation.

And then as the water surface elevation reached
the bottom of the log, you'd have the beginning

of some water being backed up.

And as flow got higher on the log, that amount
of backwater, or that effect of the log, would

be at a maximum until the stage got above
the top of the log.

And then the effect would be minimized until
at some point it's completely drowned out.

So you would have a truss type of a shape.

Another typical visual evidence of the need
for a truss shift would be if you had vegetation

on the banks.

And what was actually happening was vegetation
was causing either backwater.

Or there was less vegetation on the banks
than when the rating was developed.

And it would cause, you know, lower stage
values for the same discharge.

And would have a truss shift towards the right

Which is similar to what we're seeing this
measurement 670 plot as.

Let me quickly show you these two different
types of shift curves that we could apply

or develop for measurement 670.

So we're going to want to insert a shift curve
below the last shift curve that's available.

So I'll highlight the bottom one.

And then click on the insert below.

And I do want to insert one.

And let's set this for the measurement time
just to get us started.

May 31st, 2009, at 0900.

And we'll say measurement 670.

And it's black, which is fine.

We'll hit apply, okay.

So let's input 3 points here.

And for the top point I know that I want this
thing to merge at the top breakpoint.

So I'm going to set this to 36.1 and 0.

And I am going to set this for -- I don't
know, I'm just going to take a stab at it.

Make it about 28 feet.

And I know it's about .5.

And if I were to be applied developing a half
house shift, this would also be .5.

And you can see the type of shift curve that
I'm dealing with here.

Now, this works really well if you don't have
to deal with flow dropping too far below the

measurement 670.

And without any evidence of where the stage
discharge relation is below here, what ends

up happening is you start adding in quite
a bit of discharge different from what the

existing rating is.

For example, we have a half of a foot shift
occurring at the lower end.

Which might work.

It might actually be the case.

Ideally, we'd get a measurement somewhat after
this high flow event to define where the stage

discharge relation is after the flow came

But unfortunately, when we're working operationally
working in continuous records, we don't have

that convenience or that luxury.

So we have to be careful for how we develop
this shift curve.

With the understanding that we might be overestimating
in this case discharge.

So this is what a half house type of shift
would look like for this measurement made

under channel control conditions.

Now, let's play with it a little bit and see
what a truss shift would look like for this

particular measurement.

Again, when we were developing that half house
shift, I kind of arbitrarily selected a hinge


That's something you'd want to raise up as
high as you could again.

So that most shifts would smoothly merge with
the rating at that breakpoint of 36.1.

Now, when developing or determining what your
hinge point is going to be for a truss type

of shift, you have to think about what is
causing the truss shift.

And come up with an idea of where about is
the middle of that effect or where that effect

is the greatest.

And that doesn't necessarily mean that you
draw your truss shift out to the measurement

gauge height.

The measurement gauge height is not a hydraulic
point of importance.

Rather, it's something that's arbitrarily
observed because you were there.

So if I were to develop a truss shift for
this particular measurement, first thing I

would do is I would have my input point at
the low end at the breakpoint of the transition

from the lower control to the higher control.

So I'd set this at 11.4 feet.

And because it's a truss it would be merged.

So it would be like 0 right there.

And so you can kind of see what we're dealing
with here.

But at this point I have not selected a good
hinge point.

At least not in my opinion where it's representing
the middle of where the effect of all this

stuff is.

And so at this point I'd probably grab the
move shift point.

And I would pull this out to be somewhere
in the middle of this channel control area.

And whatever would be affecting the stage
discharge relation in this area likely happens

at a maximum somewhere in the middle of the
channel control section.

So if I were to draw a series of truss shifts
to represent other measurements that might

plot off, I would keep the same hinge point
and the same merge points.

I would just have a different shift amount.

So I would have a truss that kind of came
down to a maximum point here and then in.

And then once again down here in the middle
or down to the merge point again.

Now, there are limitations to the types of
shift curves that we are able to develop and


Keep in mind these are very simplistic corrections
to a fairly complicated problem associated

with the stage discharge relation.

And so there are certain rules we kind of
have to follow.

And we are limited by only 3 input points,

And we want to draw shift curves that are
hydraulically responsible to what we think

is causing the problem.

But also remain true to the base rating curve
as best they can.

As well as remain true to fellow shift curve
forms that could be placed in there as well.

So at this point I'm going to assume that
whatever happened in the field, whatever was

observed in the field indicated that a truss
shift was likely to be the best approach for

this measurement.

And the correction to the stage discharge
relation for this measurement.

Now, it's time to take a look at the time
series and decide how we want to apply this

developed truss shift hydrologically.

The last shift we had input for measurement
669 likely should be held to about the beginning

of this increase in flow.

And so what we'll want to do is we would probably
want to take this shift for measurement 669.

And we'll just clone it.

And we'll say yes.

And we're going to do this again.

We're going to say held shift for measurement
669 until rise.

We'll hit apply.

And we'll graphically move this one to be
somewhere right about there.

And so the shift curve for 669 is held steady
from this input point to this point here.

And then we have this one for 670, which we
will probably want to clone as well.

And I want to clone it.

And I want to start it a little bit sooner.

So let's make this 5-28 -- May 28th.

And for now we're going to say shift for measurement
670 held back to peak.

Hit apply.

Now, I can take the tool again and make some
brash assumptions here because I don't have

too much to go by.

But maybe put it on the peak here.

Or you could potentially put it somewhere
right here.

Blame some of the change on the rapid rise.

But what's going to actually happen when this
shift curve prorates into this one is you're

going to be getting a lesser shift down here
and an increasing shift in time over here

in the upper portion right there.

And so we can zoom in and give it a look.

See how things look by grabbing the time slider.

And putting it.

So we know what that shift curve looks like.

And so here's where we start seeing things
starting to move their way out, right.

Until this point here where we think this
is fully active, if that's what we think.

And then it remains fully active throughout
the range right here.

And so if we think that's what's happening
as this rise is occurring, that's when this

change is happening.

Then we think we're applying this properly.

Now, if this truss shift is related to something
more stable that's not becoming more of an

effect, it would probably be more responsible
to apply the truss shift fairly quickly after

the initial rise.

So that it's in full effect for the entire
hydrograph here.

And, again, that's something that your notes
are going to tell you about.

It's going to be something that you will know
characteristically of this site because you

visit it all the time.

You understand what's causing these types
of shifts.

But when you start dealing with shifts, the
take-home message in the channel control section

of a multiple offset rating is that things
get fairly complicated when you start having

a change in the stage discharge relation.

A temporary change in the upper channel control

And so understand the limitations.

Understand how it applies.

There's no silver bullet or best way to do

But you do need to apply it so that it's at
least following your reasoning and your story

that you're going to tell in the narration
of the station analysis.

Now, moving forward in time, this inputted
shift on measurement 670 is going to have

a 0 shift, right.

'Cause we merged with the rating at the breakpoint
of 11.4.

And so there's a 0 shift that's going to be
applied to any stage values below that breakpoint


Just something to keep in mind.

And you're going to come across this as you
are inputting shifts and the future is unknown.

And flows are going to happen as they're going
to happen.

So let's move on and see what the future brings.

And click on measurement 671.

And you can see we are now back down.

Flow has come down to some degree.

Let's look at our time series a little bit
and you'll see.

So flow has come down considerably since our
last measurement.

And we have a new measurement made under section

Measurement 671.

And if you look on the variable shift diagram,
let's zoom in down this lower portion.

You'll notice that it is plotting to the right
of the rating curve.

Something we hadn't seen earlier.

But obviously this large event in which 670
was a measurement that was made -- and also

plotted towards the right -- has changed something
in the configuration of the section control.

Now, having hindsight available to some degree,
we can possibly reformulate the truss shift

curve to match or to take advantage of the
information we gained from having measurement

671 now.

As you've noticed, as we've developed these
shift curves for the different measurements

made under the two different hydraulic controls
that we had measurements made, we stayed true

to the hydraulic control.

And that is, our shift curves were only active
throughout the region specific to that control.

For example, if you look at our section control
shifts -- the ones we made for measurement

665, 666, 668, 669 -- we drew shift curves
that merged back to the rating at the breakpoint.

The transition zone between the two controls.

And similarly, when we developed this truss
shift for measurement 670, we made some assumptions

that this was specific to only the channel
control and was not having an effect on the

section control.

And this is the general mindset or the general
operating procedures that we'd like to see

everybody using.

In which, you know, the understanding is as
flows come up, the effects to one control

are drowned out.

And they're not noticed on the next control
that's higher up.

Cases in which you're coming off of a shift
made to say this channel control coming down

into the section control are one of the few
instances where it's almost okay to combine

the shifts to some degree.

Because of the limitations we have in applying
three-point shifts and remaining true to the

control conditions.

And so what I was saying, we might be able
to refine or improve the developed truss shift.

It's possible here that we could take the
truss shift that we initially had and come

down off of it.

And then come down through measurement 671.

So I'll show you what I'm talking about here.

I want to clone this last shift.

So I'll clone it.

Yep, I want to clone it.

And for the time being I will start again
on the measurement.

So measurement 671 was made in August 10th.

And we'll say this is for measurement 671.

And we'll just put a time in there -- my bad
-- 5:30 a.m.

Holy cow.

Okay, and we will apply.

And now let's do what I was saying.

So if I zoom in, I'm going to keep the hinge
point that existed.

And I'm basically just going to be adjusting
this merge point.

And so I like that it's at the transition
zone between the controls, right, at 11.4.

So I just want to move this horizontally.

I want to move it out just a little bit like

And now you'll see when I zoom out that I've
got a truss shift that kind of merges into

what would've been a half house for that lower
section control.

Now, we are violating that rule of thumb that
we want to keep our compartmentalized shift

curves specific to their control.

But this is one of those examples where it's
difficult to draw shift curves that are allowed

to cross through that transition zone right

And so what we've developed is essentially
a compound shift, one that goes beyond one

control into the other control.

And so now we want to look at this time series
and start thinking about how we're going to

apply this developed shift curve.

Now, a rule of thumb or something to think
about is when you're working records up continuously

and you come across an instance in which you
have a shift at a location in the rating which

you wouldn't expect it to be fairly unstable.

Like we're dealing with here in this channel
control section.

And you want to be able to have some flexibility
to adjust the shift curves over time after

that measurement.

And so you might consider not finalizing record
back, you know, on that measurement.

And give yourself a little bit of leeway to
develop some or refine some shift curves if

it's needed.

Or refine the stage discharge relation after
such an event.

So what we're dealing with here is, remember,
we have that main -- that full truss shift

going into effect right here on the peak,

And so we can leave the one that was initially
put in on that peak.

Because the stage isn't dropping low enough
between then and this little dip right here.

To which we would have to deal with the lower
part of that compound shift.

And so we can also assume that maybe some
of the extra scour that happened over here

at the lower end might've happened on this

Or you come up with some sort of a reasoning
to how you want to apply those shifts.

But I'd be happy keeping the full truss in

And I would have no problem implementing the
compound truss shift somewhere around over

here by measurement 670.

So with that reasoning, I need to adjust the
truss shift that's inputted here on measurement


So that's the shift that started on May 31st.


And so I am going to adjust this just like
I did the other one.

And I'm going to come down here and move it
horizontally right over there.

And I can put this shift curve on just to
look at it.

Make sure they're overlapping one another.

It looks like I need to move this one over
just a little bit further.

And there we go.

And now I need to adjust the start time for
that refined truss shift.

And remember, I decided that this quick change
is going to be what I'm assuming made my section

control possibly move over.

Again, these are things I have to come up
with based on the evidence I've seen or what

I know about the site.

Obviously nobody was there.

There wasn't a measurement made.

It's hard to know exactly what happened.

But that's the reasoning I'm going to give.

So I'm going to move this start time over
to here.

And I would like to adjust the remark for
this one.

And so refined truss -- or actually compound
truss request developed using measurement

670 and 671.

Held back to rapid rise during high flow event.

Hit apply.

It's just good to keep those notes going.

Moving forward, I've got another measurement,
672, to see where it plots.

And it looks like it's made during this lower
flow period following measurement 671.

And let's look on the variable shift diagram,
which is getting a little bit busy.

Let's turn off some of these older shifts
that I kept on a little while ago.

And deal with the current one that we're looking

And it looks like measurement 672 plots almost
exactly the same place as measurement 671.

So this is a pretty simple solution here.

Let's just clone that last shift curve associated
with measurement 671.

And give it a start date of September 9th,
2009, at 10:44.

we'll call this measurement 672.

Hit apply.

Now, we basically have the same shift curve
entered right here, right here, and right


So it's the same.

Nothing is prorating over time during this

Everything is staying the same during this
low period and coming off of this recession.

Now, it looks like I've got a higher flow
event coming up soon after measurement 672.

Instead of just keeping this on the start
time and the measurement, I'm going to adjust

the start time back just a little bit to the
beginning of this event.

Thinking that I know for sure that I can go
that far ahead in time before I load up my

next measurement.

So let's load up our next measurement, 673.

See where it plots.

And it's another high flow measurement made
under channel control.

Very close to the transition to the overbank.

Slightly lower than that measurement 677.

And if you look in the variable shift diagram,
measurement 673 is crossing the rating -- the

base stage discharge rating.

And so we could just 0 out our shift curve
or apply a 0 shift curve at this stage.

And so let's move the time series over just
to look at it before we develop the shift


And it looks like we have an event that comes
up typical of all the events of this site.

That remain high for quite some time before
dropping down.

So I would be inclined at this point to develop
a shift curve of a 0 for this upper event.

However, understanding that I don't know what's
going to happen at the lower end after this

event recedes, right.

I don't have the time series in front of me.

If I was working this operationally, I would
not know what the future held.

But I am going to develop a shift curve.

I'm just going to insert one at the very bottom
of my list here.

I do, and I want to start on this measurement,
10-20-2009, at 12:12.

Measurement 673.

And hit apply.

And as you can see in the shift manager, it
defaults to a 0 shift unless I want to change


I don't think I want to change it.

I'm going to keep it at 0.

I'm going to assume that there's probably
something's going to happen to the lower section

control after this event and go from there.

So what I'm going to do is I am going to adjust
the start date for this shift 

to the top of the rise basically.

And assume that things changed on the rise
and that everything is good.

And so I go from this compound truss shift
over to this rise back to a 0 shift through

the length of the rating.

Now, there's different reasoning one could
apply to possibly keep the truss shift in

place, right.

Because the truss shift actually goes through
measurement 673 as well.

And one could provide some sort of an argument
possibly to say that things necessarily maybe

didn't change on this rise.

Well, this is the type of interpretation and
argument you're going to have to lay out in

your station analysis.

I mean, you know the site.

You know how it typically shifts.

You were there and saw possible differences
from one visit to the next.

So those are the types of things you need
to keep in mind as you apply your shifts hydrologically

in time.

Provide some sort of an argument as to why
you did it one way or another.

But let's just look real quick on the time
slider how that compound truss shift is going

to change into the 0 shift, right.

And so we're going to see that it's slowly
on this rise becoming less of an effect until

finally we are on the base rating throughout
the entire stage discharge relation.

This demonstration stepped through a series
of measurements made throughout on entire

water year.

But they were presented in a fashion similar
to working records in a continuous basis.

As you can see that there are decisions that
have to be made as you develop and apply shift

curves throughout a water year without knowing
what the future holds as far as flows and

shifts and that sort of thing.

What's interesting, if you look at the shift
manager, we can turn on all these shifts.

And we can have an idea of what kind of range
in shifts we saw at this station throughout

the entire water year.

And you can look at your overview plot and
you can see a series of curves happening over


And if you zoom in on your variable shift
diagram, you'll see a variety of different

shifts that have shown up that you've applied
throughout the year.

What's worth pointing out when you look at
all the shifts over the entire year for this

station is that they all remain true to the
base rating curve, right.

They have similar smooth slopes that merge
at certain locations.

And gracefully come off of the rating curve
and merge back in in other places.

We don't see any shift curves that are crossing
the rating curve.

We don't see any shift curves that are crossing
each other.

And we don't see any dramatic changes in slope
in these shift curves.

And that's the take-home message.

We want to make sure that our shift curves
-- which are very simplistic when compared

to the rating curve itself -- remain true
to that base rating.

And that pretty much wraps up our demonstration
on using GRSAT to develop shift curves.

And to apply them in time.

I hope it wasn't too long and that you learned

Thank you.