Working with Multiple Offsets in GRSAT

Video Transcript
Download Video
Right-click and save to download

Detailed Description

This video demonstrates how to develop multiple offset surface-water rating curves using GRSAT.


Date Taken:

Length: 00:37:20

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


In this video we're going to discuss how to
develop multiple offset ratings using GRSAT

and we're going to be building upon a lot
of the information and methods and skills

that were presented in earlier videos in this
video series on GRSAT.

As you can see here I have GRSAT open and
I am going to open the file, the XML file

of the primer that I'm going to show you here.

So I'm going to click on From File and I would
navigate to where it's at.

This is the Multiple Offset Primer.

I'm going to click on it.

I'm going to click Open and then I'm going
to Read Data and what you see here is the

last or the existing rating for this specific
site the Noxubee River near Geiger, Alabama.

What you can see is if you go into the Offset
Manager that this is a single offset rating

but it's essentially a curve that was developed
using a number of rating input points to define

that curve as shown here.

So as we showed in all the earlier videos
when we are going to be developing a new rating

or changing an existing rating the first step
in the process is to essentially clone that

existing rating so that you have one that
you can work with.

So I'm going to click on the Clone Rating
Tool, the little sheep icon shown here.

I'm going to renumber this to rating number

I'm going to put in a remark just a really
quick and simple remark here indicating we're

developing a multiple offset rating.

Now you would put in a more specific set of
remarks or comments to describe why you did

that sort of thing, something we definitely

I'm going to remove the Selected Shifts.

I don't need those coming over.

I don't need any Offsets coming over.

I'm going to be developing my own.

I can leave the measurement settings.

Those seem just fine.

And I'm going to click Okay.

And what you see here is essentially the exact
same rating that was shown earlier.

The offset changed so you see somewhat of
a different plotting going on here between

23.0 which is shown back here and 24.0.

But this is the one we're going to work in
right here.

And the first step because we're building
a new rating is we want to pretty much get

rid of the existing rating points that were
here from the previous rating and there's

a number of ways you can do this.

You can go into the rating table here and
you can select different rows.

You can select multiple rows, click the X.

It deletes them as shown here.

You can utilize some of the tools or the graphical
buttons that are shown above here.

Here's is Delete Rating Points.

I can draw a box around the area I'm interested
in deleting.

They go away and you can see the rating table
updated dynamically and then I can continue

drawing a box around here to delete those
rating points.

And you'll see in the rating table pane here
there are no inputted rating points.

Similarly, there are no triangles connected
by lines and there are no lines shown here.

So we now know that all we're looking at right
now is a plot of the measurements for this


The next step that is discussed in some of
the earlier videos is to remove or filter

out measurements that you don't want to include
in the development of the rating.

Measurements you'd want to filter out is would
be older types of measurements, measurements

associated with controlled conditions other
than clear conditions.

Essentially you'll want to filter out measurements
that you don't want to put any weight into

drawing the line for the rating curve.

And so here we would go into the Site Visit
data pane.

I can click on the tab, drag it out.

Make it a little bit larger because I'm going
to be working a bit in here in the Filtering

of Measurements and unselecting them.

There are a number of different ways in GRSAT
to unselect measurements or to re-select measurements.

You can utilize the table here in the Site
Visit data pane by sorting different columns,

having the measurements sorted in different
areas, selecting them, clicking the X to unselect

them or the check box to check them to select

There are options to drag multiple rows to
unselect them or to re-select them.

You can take advantage of Windows shortcuts
by selecting one row.

Going down, skipping a few rows holding the
shift button down clicking on a row somewhere

down below and all those in between will be

Again you can unselect them or you can re-select
them in that same fashion.

And there's the opportunity to utilize the
Time Series View to select or deselect measurements

based upon where they fall in time.

Now one of the first filters I want to apply
to these data are to get rid of older measurements.

I want to build my rating curve to be representative
of current stage discharge relation conditions.

So in taking that approach I can select the
Time Series View right here and you have time

on the X axis and measure discharge on the
Y axis and I can paint a pretty broad brush

here and deselect a number of measurements.

For this rating curve that I'm building I
really don't want to develop a rating with

data that's older than say about water year
2007, so I can go in the table over here and

deselect a whole swath but I can also do it
here in the Time Series.

So I'm going to click on this Unselect Measurement
Points and I can draw a box around different

measurements in the Time Series and it unselects

They showed up here unselected here in the
rating table, the Site Visit table; excuse

me, not the rating table.

And they became gray on the plot of the rating
you see right here.

I can keep drawing a box in the time period
I'm interested in, de-selecting them.

It takes a little bit sometimes and they go
away, so that's how you do it graphically

in the Time Series View.

You can also do it in the rating curve itself.

I can go into the rating curve and unselect
the measurement if I like.

Select here Unselect Rating Measurement Point
and I could remove this point right here if

I wanted to and now it's gray and it will
also be gray in the Site Visit pane.

So to make it a little bit simpler I'm going
to spend the rest of my time filtering measurements

out in the Sight Visit Data pane.

And so the first thing I want to do like I
said before is remove older measurements,

so I am sorting this.

It's sorted by measurement, date that it was
made and I want to get rid of everything prior

to October 1, 2006.

So I can select and drag, click the X. I can
go down here and find the next ones that haven't

been selected.

Select one and then scroll down and to Water
Year 2007 or just before that.

And then I can hold down shift.

Select that and those measurements are now

To see where we're at I'm going to click on
Default Rating Shift Layout to stick is the

Site Visit Data pane back in there.

We can readily see which measurements we have
filtered out just simply based upon our time

filter of those that occurred prior to the
Water Year 2007.

So we need to do a little bit more filtering.

We want to make sure we include only measurements
that are associated with clear control conditions

so we're going to pull out the site visit
data pane once again and I'm going to sort

this by control condition and what you can
see here if I go all the way to the top it's

going to filter it alphabetically or sort
it alphabetically so I see Clear Conditions

first which are the ones I'm after and I'm
going to scroll down until I see Debris type

of or Non- Clear Control Conditions and the
first one I come across is LGDB that is a

light debris flag.

Let's see where this measurement is.

If I click on it I get a halo around the measurement
itself and from what I can tell this light

debris is likely not affecting the stage discharge
relation dramatically because there's no other

measurements really plotted nearby other than
that pink one.

I am going to keep this measurement in because
it's something I want to use when I build

my rating.

So I'm going to keep that in and keep scrolling
down to the other control conditions that

are not clear.

Here I see a submerged one.

I'm going to get rid of that.

I'm going to get rid of that submerged one
and then I start coming in to what's known

as Unspecified Control Conditions.

And these are basically control conditions
that have not been indicated in the notes

by whoever was in the field so they could
be Clear Controlled Conditions.

It can be something other than clear.

So for now we're going to keep those in.

We have filtered out older measurements and
undesirable controlled conditions however,

what we can see now is that we have filtered
out a lot of higher flow measurements that

were made in the past and it's time to bring
some of those back.

I'm going to grab to Site Visit Data pane
and I'm going to say to myself I would like

at least to see all measurements that have
ever been made above 10,000 cubic feet per


So I'm going to sort by discharge, go to the
top of my pane here.

Click on it again and I'm going to include
a re-select discharge measurement greater

than 10,000 cubic feet per second.

So I'm just going to drag this down because
there's not all that many of them.

And so I'm going to put a checkbox next to

And we see that they show up in our rating

Now there's quite a gap here in this segment
of what we would expect to be the channel

control part of the rating curve and we're
going to want to add some of those in as well.

I am going to go back to my default rating
shift layout and I'm going to select or re-input

these graphically using the plotting window
right here.

So I'm going to select Measure at Rating Points
this curved red arrow and I'm going to draw

a box of the area I'm interested in right
here and all those measurements come back.

Now notice there's quite a bit of scatter
or kind of a larger cloud that's occurring

somewhere in here right around, just short
of 30 feet of stage.

I'm going to zoom in there and unselect some
of these measurements, some of those that

don't seem to plot on that same relation that
is starting to show up.

So I can use the zoom window at the top or
the indicator where the zoom is and then deselect

them here in the zoom window.

I'm going to click on the unselect measurement
points and if I look up at this top pane I

see that these two red ones and these three
pink ones are kind of in an area that I don't

think is related to the stage discharge relation
currently so I'm going to draw some boxes

around them, deselect them.

They go away and similarly there are these
two blue ones up here and right about here

they don't seem to fit in this general trend.

There's also a blue one just next to where
I zoomed so I'm going to grab this little

hand for the panning and I can grab this rating
zoom box and move it so that what is in this

box now is down here and there's essentially
this one measurement, measurement 674 that

I would like to get rid of coming down here.

I'm going to unselect it.

I'll make sure it's that one, 674 and I'm
good to go and it's gone.

At this point we've done a pretty good job
of filtering out the measurements that we

don't want to give any weight to as we develop
our multiple offset rating.

I'm going to back this out to the main view
shown for the rating curve and at this point

we have these gray measurements that are still
showing up on the plot but we probably don't

want to see them or view them as we develop
our rating utilizing the colored measurement


I can click on the show/hide unselected measurements
button shown up here at the top and you'll

see all the gray measurements that we've unselected,
that we don't want to use don't show up anymore.

It's pretty handy.

We are now ready to start defining our rating

From the plot of measurements here it is obvious
that we have probably three unique hydraulic

controls that control the stage discharge
relationship at this site.

Each measurement should have notes on it that
indicate what hydraulic control the flow is

under at the time of the measurement.

So I would have notes that help me out identifying
where or about what range of stages different

controls are active.

I have this graphical plot here that certainly
shows three distinct slopes or different angles

and I would like to have a cross-section if
it was available to help me define certain

important hydraulic points in the cross-section
that indicate where controls may be transitioning.

I'm going to present here my approach to defining
and developing a multiple offset rating.

You'll come to a comfortable method that you'll
end up using.

Some folks like to change the colors of the
Site Visit Data to indicate which measurements

they want to include or which segments of
measurements they want to include.

I typically don't change the colors or mess
with the Site Visit Data at this point because

I've already gone through the filtering aspect
and I kind of want to leave it alone.

It's set to where the measurements I want
to use to build this rating are activated.

The approach I take is fairly step wise.

I work in a fashion from the lowest hydraulic
control segment into the next highest control

segment and go from there and the process
I take is to essentially zoom into the segment

I want to work on first and then work into
the next segment from there.

So I block out the other segments by zooming
in and focusing on the measurements that are

related to that hydraulic control.

I'm going to start at the lower section control
that probably goes from somewhere of about

seven, eight feet of stage up to somewhere
around 10, 11, 12 feet of stage.

I'm going to zoom in clicking my little zoom
window here or zoom button to highlight that

area of interest right here.

Again, I would have notes that indicated where
or what measures were under what hydraulic


And you can see again what area that is zoomed
in is shown by this blue box and the area

I'm working on is down here in the zoom window.

At this point I'm going to make this a little
bit larger, a little bit easier to deal with.

We're now ready to start defining this section
control of our rating curve with some inputs

points and an offset.

Recall that when we cloned this rating we
deleted all the offsets so there aren't any

offsets identified at this point.

So let's go to the Offset Manager, turn on
Offset 1 and make sure that we're plotting

everything to Offset 1.

This will ensure that our rating points show
up when we add them graphically and that when

we start adjusting the offset that we're plotting
it properly and it's looking like we would

expect it to look.

So as was the case earlier in some of these
videos we discussed how it's good to use,

to input two rating points.

They define a straight line and use that as
your guide as you adjust the offset to try

to get the measurements to plot on that straight

So we're going to add a point down here at
the bottom somewhere near these lower points,

these lower measurements and then from my
notes and from the plot of these measurements

I have essentially accepted the fact that
the section control goes at least up to this

measurements right here.

So I'm going to try to stick the rating curve
point right on that measurement.

I start looking at the variable shift diagram
shown on the right and you start seeing the

cluster of the measurements that are associated
with this lower segment.

You won't see any other measurement show up
because the rating curve and rating input

points don't go any higher than this measurement.

So we can zoom in here and as was the case
before it would be nice to use this as a guide

for adjusting the offset and removing some
of the curve nature of the rating curve itself.

And you'll notice the measurement that I've
highlighted in the halo is not showing up

in the variable shift diagram.

That's because this rating point is slightly
below the gauge height of that measurement.

So I want to move that just a little bit,
get that rating to go straight through that


Use it as a guide so that's anchoring the
top of the rating which is shown by this zero

line in the variable shift diagram and I will
add, I will move this lower point just a little


It seems like it could be splitting those
lower couple of measurements just a little

bit better and okay.

That's works out pretty good.

It's time now to start adjusting the offset
in an effort to get these measurements to

plot more linearly closer to our guide that
we are using as defined by these two input

points and the straight line.

As we notice here there is a curve with the
plotting under this current offset.

As we adjust the offset because we raised
the offset because it's concave we should

see these measurements start moving towards
the left in more of a linear type of fashion

up and down along this rating line as defined
by the zero.

So let's toggle the offset and see the measurements
start to move over.

You will notice that in the zoom window the
rating zoom window I'm starting to lose the

view of my rating curve and that's because
I do not have the automatic rating scale option


If I had it on this rating curve would keep
zooming out to the entire plot and that's

not something I want to use.

Recall earlier that we said there's times
when you want this automatic rating scale

to be on and times when you want it to be

When dealing with a multiple offset rating
it's typically best to have it off.

And so to get that rating curve back I will
go in here, grab the hand tool and I will

move this down.

I will actually set this scale to be on a
scale back.

Ooh, it moved them all on me.

That's okay.

I can zoom in, zoom back in to my rating curve
and go from there.

I can always move this around but I can't
hit that scale.

It will bump everything back up.

So let's go back to where we were.

We are at our offset and we are increasing
this until we think we see the measurements

start to get in a straight line and it's usually
readily obvious when you've gone too far as

the curve becomes convex and goes more off
to the left and that obviously happened there.

So somewhere in here about 7.something, 7.4,
5, 6, it looks pretty darn good.

Again I probably want to make some adjustments
of the little bottom rating curve point here

to get it more in line with these bottom measurements,
split them a little bit better.

And it looks likes that's probably about as
good as I am going to get plus or minus a

tenth here and there.

Yeah, and I'm going to go right about there.

I think that looks pretty good for my section

My first rating curve segment of this multiple
offset rating.

Let's unzoom the rating so we can see all
of our measurements.

I'm now ready to start working on the Channel
Control segment of this rating curve so I'm

going to zoom in into the segment of the rating
curve that I believe to be associated with

Channel Control.

Now if I had a cross-section of this channel
I would have an elevation of about the top

of the bank which is about the point in which
the control changes from channel control to

over bank control essentially the highest,
about the highest rating input point you would

want to have for this segment of the rating
curve itself.

And so if I had that information which ideally
I would I would know about where the put my

next highest rating input point.

I don't that information here for this example
but looking at the curve it has to be somewhere

right around in here at about, I don't know
11,000 cubic feet per second or so, so I'm

going to stick a rating point in there.

Now I'm going to go into the Offset Manager
and I'm going to indicate that I want to add

a second offset by clicking on this box.

You'll notice that I'm still plotting to Offset
1 which is associated with the lower section

of this rating curve and if I'm going to change
the Offset 2 I'd like to see it plotted.

So I'm going to click on Offset 2 as the plotting

You'll notice when I toggled the fact that
I wanted a second offset all of a sudden a

break point value was populated down here
in the Offset Manager.

The break point indicates the location or
the gauge height in which the offset changes

and essentially it marks the location where
the extrapolation in log, log space changes

from one offset to the next offset.

By default, when you click on an offset it
selects the lowest rating input point as the

break point.

So you'll see, if you look at this top screen,
it selected my lowest rating input and if

you look at the rating table you'll notice
that the 8.63 stage value for the rating input

is gray.

The proper location for this break point is
at about this stage of 11.42 feet so I'm going

to toggle up the break point by clicking on
this arrow and by default it's going to go

to the next rating input point.

Now you notice that this rating input point
is grayed out.

I cannot edit it.

I can't click on any of these boxes and change
it and I cannot come down into is the zoom

window and move this rating input point.

It gives me an error.

That's because it's an important point in
the rating and it is very important because

of the way the extrapolation changes.

The fact that I cannot move this point creates
somewhat of an inconvenience when I'm trying

to fit the straight line segment to the measurements
down hereby changing the offset.

You remember how we adjusted the rating input
points for the last segment we worked on to

get things centered on the variable shift
diagram and to get a better fit for the measurements?

Well, I've got a little trick.

I add an input point by clicking on this add
rating point somewhere just above this break


So now I have the ability to move this point
around as I see fit, as I'm trying to fit

this line right here.

So just as we did before as far as changing
the offset and moving rating input points

we're at that stage of the game for this segment
of the rating curve.

So I can move this point around just a little
bit try to get things worked out and then

I want to zoom out of the variable shift diagram
and do my best to fit the upper portion of

this segment with this rating point to somewhere
over here so that this top point is near the

zero line of the rating curve itself so it's
got to move just a slight little bit, not


Oops, went the wrong way.

Come down here.

And that's pretty close so I could probably
go just a little bit that way and now I'm

pretty close to the zero line and as you can
see before the offset that we're plotting

this to which is zero right now is causing
the measurements to plot curve linearly, concave

somewhat off of the rating curve and it's
a little more obvious here when you look at

the zoom window.

So I'm just going to toggle this offset and
watch the measurements move, hopefully get

closer towards that zero line in the variable
shift diagram and get a little bit closer

to the line on the rating curve plot, the
zoom window by toggling this up.

And you'll see I did pretty good for the segment
up here and I'm falling off a little bit here

and these aren't moving so much and it's likely
due to the way I put this input point in here,

so I can move it.

Move things around just slightly, just a little
bit see how we're doing and that moves them

over a little bit, gets things a little bit

Again this is an iterative process like I
explained before and there's a chance that

there's some clustering of measurements here
that might not be that great to have there

at this point in time so I'm at about the
same offset point I was before but I think

I went too far looking at this bend here in
the convex shape of the curve that is starting

to show up so I'm going to back this down
just a little bit to see if there's something

I can do, see if I can get these measurements
to plot a little better.

Now they look really good in here but I am
struggling a little bit down here so I'm going

to move these around a little bit and see
what I can do.

Try to get them to fit.

It's getting a little bit better so I'm going
to keep moving around this lower rating point

and see if I can get something a little bit
better a little bit closer.

Seems like I'm having some trouble getting
the measurements in this area here to kind

of plot that way.

That looks a little bit better.

I can back this down somewhat to see I can't
get these things to get more in a straight


That's not too bad and I'm going to move this
one just a little bit more.

What you're noticing is there's an area in
here right around the break point that we

call the transition zone and typically there's
a lot of noise in the measurements here.

Things are changing.

One controls drowning out, the other control's
becoming active again and so as we talked

about earlier you'll learn about as you move
along with rating curves is that our goal

is straight line segments for hydraulic controls,

We are on the Channel Control segment here
and we dealt with the Section Control down

here so these are straight line segments defined
by two points but what we have to do in between

is have a smooth transition of a curve to
blend in these two straight line segments

and there's a chance that this area of the
rating curve here is a little bit noisy.

We're not going to get it on a straight line
segment as we are for these measurements here.

So I think we're doing pretty good and for
this demonstration I think we'll call it good.

So at this point we think we're close enough
and we're ready to move on to our next rating

control segment which is the over bank flow.

And so I'm going to auto scale, zoom back
out, give a look at our rating.

As we did for the Channel Control segment
of this rating curve now we're going to move

in and zoom into the Over Bank Control for
this rating segment.

So I'm going to draw the zoom box and there
we go.

Again, ideally we would have the elevation
of the over bank where this break point or

this rating input point would be.

Now we're going to add a rating input point
at the high end about where we think the top

of the rating curve is and it's pretty obvious
here this is the highest measurement we have

for this site.

So we're going to add a rating point smack
dab in the middle of that rating input point.

I'm going to auto scale this until we start
seeing all the measurements that we're dealing

with and I'm going to the zoom in to about
this segment which is above 35 feet up until

about 48 or something like that.

That's the part I'm interested in.

Now you'll notice that this part of a rating
curve is typically quite flat.

There's a lot of change in discharge for a
little bit of change in gauge height.

And in a lot of instances unless you have
a stream that experiences over bank flow somewhat

frequently or it's a very, very long period
record you rarely have much definition in

the upper end here, typically one or two maybe
three measurements at most and so it is difficult

to resolve a good offset in this upper end
for that reason.

Luckily for us for this example we have a
number of measurements up here and there is

some scatter going on which would be expected
at that type of flow.

So we're going to turn on the third offset,
click on it and we would like it to plot to

that offset.

Now you notice when I toggle the third offset
I, by default get a break point at the highest

rating point which doesn't make sense, right?

We should have the break point located between
the control segments so I can just radio button

this down one or up one, down one and it will
go to that last input point that I had which

ideally would be about the top of the bank.

I'm going to use that same trick I mentioned
earlier so that I have some flexibility in

moving rating input points and fitting the
measurements as best I can to this top end

so I'm going to input a point somewhere just
above that rating break point that we have

right there.

And I talked about smoothing between segments.

This is definitely a location where you're
going to have to input a series of points

that nicely curve this straight line segment
associated with Channel Control into the Over

Bank Section Control to bend it over nice
and smooth so you don't have a hard angle

or rapid change in the discharge per gauge

So you're going to have a nice smooth curved
elbow that's going to go in here so I'm going

to put another point in here just for fun,
just to kind of illustrate that.

But typically what you have to do when the
rating is all done is put a nice smooth curve

in here with a series of input points.

So I mentioned earlier that typically you
don't have much definition at this upper high

end over bank section of the rating and it's
difficult to kind of resolve a true offset

for it because there's not much curve.

There's not enough points.

There's a lot of scatter similar to what we're
seeing here and so we'll do our best here

and try to increase the offset and see if
we can have any success or decrease the offset

and see if we have any success bringing these
values any closer to each other and getting

it more lined up here on the zero line which
represents the rating curve.

Oh, and I've noticed this measurement is not
being shown up here because the rating curve

isn't far enough out so let's move that just
a wee bit.

There we go.

And it's about on the zero line so you'll
see we have some convex action going on right


It could be something that we need to just
kind of increase this slightly make it a little

bit smoother going through to start ourselves
out, see if we can get this to, this relationship

to work itself out somewhat.

So, I'm going to toggle this up see if we
can get this measurement to move and like

I was expecting we're not seeing much difference
here, much change going on and part of it

is because it's a flat relationship at this
upper end not to mention the fact that there's

a bunch of scatter even though it and look
like we have some kind of curvature going

on around here which could be a function of
our scattering measurements, so again it's

an iterative process.

You can try a number of things see if you
have any luck perhaps the break point needs

to be lowered slightly or you can work something
out by moving around get things to show up

a little bit better.

But it could be a tough challenge at this
upper end to get things to match up.

So I'm going to keep moving this see if I
can get something to happen.

See if I can watch and see if there's any
inflection that happens up here.

I'm not seeing too much and when that's the
case and I'm really not doing myself any good

as far as getting an offset that's working
for me I will typically just assign the same

offset that I had in the previous section
for this upper section and go from there.

I definitely want to make sure that the rating
curve goes through this top point and I want

to make sure I end up with a nice smooth type
of transition down here something we're going

to have to work on with our rating input points.

It would be worth looking at these measurement
which one's the most recent of them all.

For example, that's from 1969.

This one here is from 1983 and filter these
out a little bit.

Go for the most recent higher water measurements
when there's a cluster like this and try to

draw your rating from them.

So that about wraps it up for going through
the process of developing a multiple offset


Remember this is just one recommended approach
to developing multiple offset ratings using


It was step wise.

It works from one rating segment to the next
rating segment and onwards and understand

that after you've resolved these straight
line segments for the different controls you

need to go back and smooth the transitions
around the break points and that requires

looking at the expanded rating table, making
sure discharges and changing too fast per

unit of gauge height and inputting some points
to make a nice smooth curve that connects

these straight line segments that you've worked
so hard on.