IRIC Pilot Applications for the Potomac River Using Bathy-Lidar Data

Video Transcript
Download Video
Right-click and save to download

Detailed Description

This short presentation will illustrate how detailed topographic-bathymetric LiDAR data sets can be used within the USGS iRIC software package to compute detailed flow and particle tracking simulations across a range of length and time scales, including capturing the effects of local unsteadiness and turbulent mixing. While results will concentrate on the recent Potomac River data set, other examples will also be included to illustrate a small subset of iRIC capabilities, including a brief discussion of the new iRIC-MI model coupling tools, which allow end users to couple ground water, hillslope runoff, tracking and other subsidiary models with detailed river models globally and at the time step level. iRIC (International River Interface Cooperative) is a river flow and riverbed variation analysis software package which combines the functionality of MD_SWMS (Multi-Dimensional Surface-Water Modeling System, developed by the USGS) and RIC-Nays (developed by the Foundation of Hokkaido River Disaster Prevention Research Center).

Links:

Details

Date Taken:

Length: 01:01:06

Location Taken: MD, US

Transcript

This is the hydrography community

call or USGS hydrography program.

If you notice,

we used to call us the advisory call.

We've changed the name of it,

'cause there's some specific kind

of legal things involved with

what's called an advisory committee,

and this is not one of those.

So we wanted to change the name to make

it clear that this is just started.

Our communication with

the hydrography community.

And not not one of those special types

of calls or special types of committees.

Comma by St committee.

Today we're going to have a talk by

John Nelson in the USGS who he's

going to talk about bathymetry data

and some of the recently collected

bathymetry data that we've done

as sort of a pilot project.

We've been doing.

We've done a few pilot projects

collecting bathymetry data using

LIDAR and in conjunction with that using,

you know.

ACDPs or sonar data basically.

And so we're experimenting with

this technology right now.

We don't have a full blown program

like we do with the.

The 3D Elevation Program 3DEP

or the OR the hydro program,

like the NHD,

but if something that we're considering

and trying to learn more about.

We do have the ability to contract with.

Private vendors to collect this kind of data.

The word gypsy contract GPS see is,

we call it the gypsy contract and

so it is something that we can do,

but we don't have funding at the

moment to to do like matching

funds or anything like that.

So, uh.

Right now we will go ahead

and pass the mic over to John.

Thanks Alan. So I'm Jonathan Nelson.

I'm at the USGS Geo.

Morphology and sediment

transport lab and golden.

And I'm going to talk to you today

about some work we did to sort of

illustrate how bathy LIDAR datasets

can be used with computational modeling

to answer a variety of issues.

So this this work was originally

presented a few weeks ago as

a as a short briefing to the

Potomac River Commission,

so I'm pretty.

I've added a couple of slides to

it 'cause I think some of our oils

bill people are on today as well,

but primarily this is just going

to be stepping through.

The the What the data looks like.

What kind of model like we we can

do and it's sort of a tip of the

iceberg view of what's possible with

these detailed data sets and the

actual computations that I'm going

to show primarily done by myself

and Paul Kinzel in the same group,

and we have many pirate

collaborators around the world.

For those of you that aren't

familiar with what Eric is Irick.

Is a sort of generic interface that

was originally conceived in my group

about 10 years ago to try to make

surface water models and sediment

transport models available to our

users in the Water Science Center.

Since then, it's developed.

And become not only a platform

for USGS models with this now.

Sort of a community platform

that has on the order of 25 one,

two and 3D models for rivers from

scientists around the world.

So although this started at the Jew

morphology instead of transport lab,

but it's bigger than that now.

So I'm speaking to you today,

not from the lab, but from my barn.

Lately repurposes my office and.

The land of forgotten furniture.

So I apologize for any crowing

Roosters in the background,

but that's the state of my

office situation as of now,

so let's just go ahead and

first look at what to pull.

The symmetry data looks like,

so this is just a simple map of the reach

of the Potomac for which we flew Topo bathy,

Lidar.

For those,

I think probably most of you

are familiar with the idea,

but everyone knows how Lidar works.

You shoot a laser beam,

you get.

Reflected light back and from the

timing information you can figure out

elevations that along with GPS and so forth.

So the idea of bathymetric Lidar

is you use visible wavelink

lighter that penetrates the water.

It goes through.

The water is reflected off the

various things in the water column

and also ideally off the bed if

the water is too deep you don't get

a reflection from the bed and you

have to as Alan referred to earlier.

You have to basically amalgamate.

Different kinds of datasets,

and so here we've got,

you know,

the Potomac a couple hundred

kilometers or something on

that order and just in detail,

I'm showing you the point of rocks.

Reach an are each downstream that

actually is is significantly deeper.

And has, as you can see,

voids in it where the lidar

did not penetrate to the bed.

This is a typical bathymetric lidar data set,

so the idea would be that we would

go in and either use existing data.

Or go out and use acoustic devices

to fill in that missing data.

We're working on that now,

but today I'm going to concentrate

on this model of what I call the

model to reach at the point of rocks,

which is about a 15 kilometer reach where

essentially we gotta complete data set.

So we've got the bathymetry and topography

here, and it's quite detailed.

So this this is the ideal situation.

My group first did bathymetric and

topographic lidar just about 20

years ago in collaboration with NASA.

When it was a little more experimental

but the same issues you know arose

back then that are still present now

having to do with the inability to

penetrate to the bottom in certain

situations and various other things.

So it's not a panacea,

but it does offer the capability for us

to get good data sets in certain kinds

of rivers within much less time and well,

depending on how you measure effort,

perhaps less less effort

than would be required.

For example, this reach.

If you were doing acoustics and

ground surveying and whatever

other standard techniques to do

15 kilometers of the reach of the

Potomac at this level of detail would

be a very significant undertaking.

Certainly weeks of work,

so and instead this can be done in

a relatively short amount of time,

hours, maybe a couple of days,

and we've done this on 10 or 15 different

rivers around the country over the years,

and so we kind of know what

the pros and cons are.

So what I'm going to do is

just sort of illustrate.

What that data gives us access to

in terms of computational modeling,

and I'm going to do it with iRIC.

That's not necessarily the

only tool out there,

but it is a tool that we commonly use,

and it has a lot of different kinds

of solvers within it, so we can do it.

Look at 1D or 2D or 3D solutions.

We can do particle tracking for problems

like the the dispersion of contaminants.

We can do sediment transport,

morphologic evolution of the channel bed,

and things like that,

and I'm just going to show you

a few examples.

So the hierarchy of approaches

for the Potomac.

I'm going to step through as I'm

going to start with a quasi steady

2D model called Fastmech over that

point of rocks reach with at

high spatial resolution.

I'm going to then from that go to fully

unsteady model called Nays2D Flood,

which is again an iRIC

over the same reach with

the July 2020 high flow and then.

Use that same model over a shorter reach

with the the 1936 flood of record.

I'll show you some particle tracking's

at both low spatial resolution and

time resolution and much higher

spatial and time resolution,

and I'll mix in some other examples just to

illustrate methods and things like that,

But this is not going to be a technical talk.

It's really just an overview of the

kinds of things we can do with data sets,

such as the one we collected here

on the Potomac.

So this is what the data looks like.

It's that's about a 15 kilometer

reach an it consists of both depths

in the channel or elevations at

the channel bed more accurately,

and elevations of the adjacent.

Land surfaces,

so the nice thing about this is

we get a seamless data set and

it's pretty easy to right away.

Run a variety of different

computational models to route both

standard flows and also floods.

An even huge floods because we have

the land surface presumably to quite

high elevation around the river.

So stepping head the typical first

step in modeling this is we make a

coordinate system and iRIC this is

all done through user interaction just

by drawing the center line down the

River and and expressing the resolution.

This.

This is actually being graphed at a much

lower resolution than the actual model,

and I'll zoom in and show you

some higher resolution results.

So with this coordinate system

fit on to the data,

we can run a variety of models

and get water surface elevations.

And vertically average velocities

from those computations.

So this is just a single example

of water surface elevation.

This is done for the flow when the

lidar was flown in one of the

things that many of you probably know

is when you do a model of a river,

you have to figure out what the roughness is.

Sometimes that involves a lot of

spatial spatial accounting and sort

of spatial mapping of different

areas of vegetation,

different substrate type, things like that.

In this case,

we're going to just use a simple

value single value.

Roughness and calibrate it using

the lidar data so the lidar gives

us not only the bed elevation but

the water surface,

essentially from the first return

of the lidar data,

so we're going to use that to calibrate

roughness for these model and so

here we have a simple quasi steady

2D run with water surface elevation

that we've adjusted the roughness

to try to give a good RMS fit to

the laser measured water surface elevation.

So here I've added the

velocities on to that same plot,

but.

That's really not part of

the roughness calibration.

So these are what the lidar

returns look like for the surface,

and these were developed from the

lidar data by Paul Kinzel just

using a filtering algorithm that

picks the first

return off in the region of the river.

It's not perfect,

but it generally does pretty well.

And sure enough the water runs downhill,

so that's you know that's a good sign.

So we've got this water surface elevation.

Now we can compare that to the model

and iterally adjuster roughness

to try to get the best bit.

Again, this is a very simple sort of standard

procedure when doing a model of this type.

So when we compare water surface elevations

from the model and the measurements

will get things that look like this.

The measurements are shown as points.

You can see there's a bridge about

halfway through where we got reflections

from the surface of the bridge and the

lines are actually the model result.

And because there are results on,

you know throughout the channel and we're

just picking the closest computational

point and essentially filtering it

onto the locations of the measured

points and drawing lines between them,

you get this sort of strange effect,

but if we go to a more conventional view.

Of the predicted and measured roughness,

you get something like this,

which is generally pretty good.

I forget what the RMS value was on

the order of 10 or 15 centimeters.

If you exclude the presence of the.

Of the bridge,

which gives a obviously a quite large error.

Since we're measuring there,

not the water surface,

but the elevation of the road.

So this allows us to calibrate

roughness and then we can go on

to more complicated modeling.

Zooming in on that model.

I'm not zoomed in on the bridge,

we were just speaking about.

We get results like this.

You can see flow around this little

upper channel in the floodplain

and sort of typical flow.

Now this is not the resolution of the model.

Excuse me,

this this is a much lower resolution

depiction where we're just plotting,

you know,

like every I think it's every 10th vector

or something like that in the flow solution.

If we zoom in on this region

downstream of the bridge,

I hope you guys can see my mouse,

but there's a reason why there's

a bunch of rocks and there's sort

of a complicated flow around that

this is a rocky bed.

It's it's got some interesting features.

If we zoom in on that at full resolution,

we we get we can actually see

that the flow is resolved.

We get steering at the flow around these

rocks between the three rocks down here.

So we have a very highly spatial

resolve pattern of flow.

Around even relatively small scale features,

the reason we can do this is that

this is a quasi steady model,

so it to do this with an unsteady model

would be computationally very challenging,

but we have a quasi steady model and

the discharge is not very rapidly.

We can very easily and quickly computational.

I think this model takes like

on the order of

10 minutes 5 to 10 minutes to run it very,

very high spatial resolution

because there is no.

Current condition for those of you

that are in the computational fluid

dynamics so so you can see we can

get very high spatial resolution,

but we can't handle rapidly varied flow,

so that's the advantage of

a quasi steady approach,

and this approach is at this

point completely 2-dimensional.

I'll come back and show you some

3D results later on, so now we're

going to look at a real hydrograph,

an go to an unsteady model.

This is the flood from last year.

I probably should have updated this

because I think there actually was

quite a large flood,

just a few weeks ago,

but I'm too lazy so there's

just the discharge versus time.

back last year. And we'll run that through

an unsteady model. And you'll see a

telltale go along this discharge plot

at the same time that you'll see

the velocity vectors and contours,

and presuming that I can

animate this.

There we go.

OK, so the flow, is now about to come up.

Comes up rapidly, the river gets wider.

Obviously the velocities get higher.

No big surprises.

Kind of what you expect. You

get more flow in these back

floodplain channels for example.

But that's what a typical 15

kilometer river model looks like.

So this is fully unsteady model.

So it does have the effects of the flow

variation in at the slope of discharge.

Slope of the discharge curve an.

You know I should point out that quasi steady

models can actually be pushed pretty hard.

You can have discharges that very

pretty quickly, not up to the dam

break situation or anything like that.

But the rule of thumb is as long as

the slope of the floodplain is small

compared to the slope of a river,

quasi steady model is sufficient

and computationally much faster.

This kind of unsteady model,

instead of taking a few minutes.

for this 15 kilometer reach it has

much lower spatial variability,

but it takes on the order of an over an hour.

I think for this long, longer reach,

maybe a little less than that.

So let's see here.

Go on to the next slide.

I'm going to show you some

particle tracking results.

I said this wouldn't be mathematical,

but well, I lied so.

But the basic idea part of

tracking is very simple.

If you have a passive tracer you know its

position and you know the velocity flow.

You can essentially project where that

particle will be a little bit later.

Now in situations like groundwater

where you don't have

turbulence particle tracking

is super easy cause

it's purely advective.

But all of you know that in surface

water flows water is turbulent and

so if you measure at a point you

get variation about a mean.

Our models typically compute that mean.

Now there are models in iRIC that are

turbulence resolving there are direct

numerical simulations included,

but really those models aren't

they're not really trackable for a

large river simulations like this.

So typically what we do is we use

what we call Reynold's averaged

momentum equations in the models and we

average over the turbulence variability.

Now the problem with that is if

you do that and you use that

flow field to effect something,

you lose the effect of turbulent dispersion.

To bring that back in,

you can assign a random walk characterized by

the diffusivity from the computational model,

whatever that computational model might be.

So then the position of the particle

is it's old position.

Plus it's advection by the flow,

plus any active movement of the particle.

For example,

swimming if it's a fish, plus

turbulent diffusion.

Which is just characterized by picking

values randomly from a Gaussian distribution

characterized by the diffusivity of the flow.

So let's there's a little depiction,

and here I've used sort of a 1D/2D

sort of explanation,

where the particle moves forward and

as we put four particles at this

original red point as they move downstream,

they spread apart because

of the action of turbulence.

But that action of turbulence is

is governed by the correct turbulence

diffusivity from the computational flow model.

So we can look at both

advection and dispersion,

which were the people here

interested in oil spills,

is of course the very

important part of the problem.

So I'm just going to show you

a simple example.

This is for a line source is

actually the Delaware River and

what you get out of this kind of

calculation is you get time of

travel from the initial point of

injection of whatever your tracer is.

You get concentration at every

point throughout the flow for all

times during the concentration,

so you get time series of

concentration as well.

You get spatial and temporal extent of

whatever constituent you're looking at,

whether it's a pollutant or you

know something like larvae that

you're trying to track through the

river as part of a biological study,

and you get spatial and temporal

variability as well as trapping,

and you have the opportunity

because you use the Lagrangian

approach rather than Eulerian one to

add behavior to those particles.

I'm not going to talk about that,

but what we are working now on

having multiple populations and

having interactions both between the

particles in the particles in the flow.

Low and the opportunity to do

geochemistry by having different

particle populations actually interact

using the USGS model Freak See.

So this is an avenue of approach

into a lot of different problems.

so our line source as it moves

down the river,

the particles in the middle or in the

thalweg go little faster than the edges,

so this is sort of a typical 2D particle

tracking many of you have seen

results like this.

OK, so let's go forward.

Yep, sorry,

and then you know one of the problems

with this is that if you go to

a complex reach like this one,

this is on the Snake.

What you find is that if you really

want to understand the spatial

distribution not only in terms of

concentration, but in terms of exposure time,

that is things like.

How, how much time is this

Is this pollutant present at a

given location?

You need a lot of particles and this is

sort of the computationally challenging.

Whoops, sorry,

challenging part of this problem.

So in this plot you've got

particles moving through the reach,

but you've also got the blue and

orange showing you essentially the

integrated particle counts in cells

or what you might call exposure and

you can see even though most of geez,

I'm not.

I'm having problems hear.

You can see that even though most of

those particles zip down the thalweg,

a lot of them end up in eddys

and hang around for a long time.

So even though the highest concentrations

are seeing in the thalweg,

the actual highest exposures are in

things like eddys. so if you're looking

at what water intakes or things

like that this is very important.

This is what you want to know.

How long does this stick around?

How much is being sucked into my

water intake and things like that.

And the downside of that is

you need a lot of particle.

Because so many of them flush

through that to model

these trapping zones,

you need a lot of computational muscle.

So to get around that and I'm just

going to show you a simple example,

we developed an idea over the last

about 18 months with new code that used

what we call cloning and colocation and

cloning essentially allows the user to

to identify a region in the flow where

they need very high resolution information.

So this is. It's easier to show this by

example then to actually explain it.

So here is just a simple flow calculation

for an eddy in the Grand Canyon,

and if we do a

particle tracking simulation.

You can see most of the particles go

downstream if you get into the Eddy.

Eddys aren't actually very

well connected to the main flow,

so if you get in and are slowly bled out,

but in this case and in many other

cases were specifically interested in

what's going on in these relatively

low velocity trapping zones and having

four particles isn't really good enough

to give us any spatial information.

So what we do is we draw a

polygon that identifies a cloning region.

We can also do this with

mathematical relationships,

specifying a minimum density

and things like that.

But let's start simple.

So here we have a polygon where we say

if density is below a certain point,

particles should clone.

That is, they should split apart,

each having half the information

of the original particle.

So for example,

if this were a pollutant,

each would represent half the amount of

the original pollutant put into the river.

Or we could make them split by 1/3,

1/5, 1/12 whatever

we want and we can continue that

through many generations so that

subsequent particles can split again

and we can get whatever degree of

accuracy we want in the region of

interest without having to calculate

millions and millions of particles.

So here's that same calculation.

Now with cloning you see particles

going to eddy,

but that because that's

within our Polygon.

They begin cloning and you can see the

generations quickly get up to 12th generation.

We get very high detail in this specific

region of the polygon we drew,

including vortex shedding and all

kinds of interesting fluid mechanics,

but without having to supply millions

and millions and millions of

particles through the main channel.

So that's the idea.

Now I'll show you one more example

where we have big differences in scale.

We have rapid velocities in the

channel and very low velocities

in a floodplain pond.

Which only connects at the highest flow.

So this is for the Kootenai River.

The hydrographic is shown in the upper right.

You can see water goes in

first at the lower connection,

then at the upper connection,

at which time the pond fills and water

starts going out the lower connection.

So there's a lot going on,

but the velocity in the channel

is orders of magnitude.

Well a couple of orders of magnitude

higher than that in the pond.

If we do particle tracking with cloning

instead of having to compute millions

of particles to get any information.

In the pond we can have a cloning

Polygon in the pond region and get.

These are now moving it like in many cases,

fractions of centimeters per

second in the pond.

It's a very slow process,

but we do get the circulation in

the pond, we can see the dynamics

when water comes in the first the.

upper channel first,

the lower channel,

then the Upper Channel and then the

lower channel reverses and so forth.

So sorry, that probably went

by a little quick backup.

Well, I can't back up.

I'm not sure why.

Anyway, no, I guess I can.

It just. It's laggy, apologize so.

Your forward again.

So let's again, water begins to go

in this lower channel into the pond.

The pond fills, is slowly filling

during this period.

As a pond gets lower,

this is during a hydrograph,

so the river is still coming up.

Ultimately,

water starts going in this upper channel.

The pond fills further,

and water them comes out this lower channel,

so you can see water

starting to go in here.

Then it gets over the sort of

break in the topography and then.

We've got this kind of circulation

and that's that's what's

observed in the field as well.

So let's go back to the Potomac and

look at some particle tracking first.

Just a simple sort of low

spatial resolution load,

temporal resolution particle track.

And again I started with a line source.

It could have been anything I can put

in any source I want wherever I want.

So it moves down the river under

the influence of

the velocity field that's

computed goes around islands.

Things like that or bars.

You know we get a picture like that,

so we're going to zoom in now and

look with a little more detail at some

of those points we looked at before.

So here's that back Bar channel

that I showed before and.

We look at the particle tracking

through their first without cloning,

so we get a few particles back in there.

We don't get a whole lot of detail,

but if we were if,

for example,

there was a water intake in that

back channel or there was some

industrial installation and we

really wanted to know where things

were going and where and how long

they were going to stay there,

we would use cloning and  whoopsie.

Sorry there we go,

so making cloning cloning Polygon as before.

And. Run this and it.

Now we get a lot more.

A lot more details,

a little hard to see in this

resolution,

but we have 10 to 20 times the number

of particles and we can see more

spatial resolution in that c hannel.

So. Moving on.

Book.

This is now the same reach of

Potomac with a 480.

This is the flood of record from 1936.

Hard to believe it's the flood of record.

Yeah, actually its the USGS flood of record.

So again we can look at this.

We get flow is much higher on the

Yeah,

the the particles are now moving

over a wider range relative to

the size of the Channel.

The channel is just a single

depiction from Google Earth and you

know it wasn't taken at high flow.

So we see what looks appears to

be water and particles moving

outside the channel.

That's because there's water

outside the photograph channel.

All the outputs from iRIC: velocities,

stresses, settlement fluxes,

concentrations.

Everything are directly available

in kml. There are tools to

look at them within iRIC.

But you can also bring them

into any kml-based software

like Google Earth for example,

or anything like that.

So that's all I've done to

make these these plots.

OK, so now we're going to zoom

in a little more and we're going

to look at that upstream bend.

And here we have a time

average flow developed by quasi steady

model for that peak flow event.

And you can see at high flows that

during this high flood you actually get

flow separation and eddy near the.

Point Bar or the

Inner side of this bend.

So if we now look at particle

tracking for this more highly

resolve flowfield whoops.

Yeah, I think this is OK.

First without without cloning,

I'm sorry it is with cloning.

What you see is that you leave some

particles behind when you clone,

and those particles then move

upstream in the eddy itself.

So this was a fairly crude

temporal and spatial resolution.

Now we're going to look at that same

problem with higher resolution.

First, we're going to look at.

Well, I guess that something

not quite right there, but.

Let's skip that will look at this flow field.

So now we're going to look at

a time varying flow field with

a high frequency resolution.

So instead of seeing a constant eddy like

we see in the quasi steady flow field,

we're going to see a varying flow.

Still, on the average,

it looks like a lateral separation,

but there's vortex shedding along

the outer margin of that eddy,

and there's any kayaker will tell you,

and there's a lot going on at

the edge of eddys,

and so will now take that more highly

resolved flow field and look at the movement.

Of particles through that flow field.

Now.

Again, we have a line source.

This is a lot of particles,

so it moves a little slow.

But it will create a cloning Polygon

near the apex of the band and you can

see cloning is already started to occur

and now you can see these big vortices.

So on average we have a single eddy,

but what's really going on in

terms of the physics of exchanges,

we get a lot of complex temporal and

spatial structure that for understanding

certain problems is critically important.

So this sort of helps you to see.

I hope that.

There's a hierarchy of approaches of spatial

and temporal resolution that give you.

progressively more accurate

views of the physics of what's

going on and in the river.

So everything I've shown you

so far as had particle sources,

just simple sources upstream,

but sources can be anywhere,

and so I've just brought in one more

example to illustrate that this is

during a measured flood on the Delaware.

Actually an I've just put a pile of

some neutrally buoyant material here,

and will animate that and what you'll

see as water comes up behind it.

As the water comes up,

it pushes that material over

and sort of a linear source,

and over time it's pushed out

to connection with the river

and is then advected downstream.

So you can have multiple sources and you

can do all kinds of things with sources,

including to put them exactly

where you're still occurs and

get concentrations and spatial

patterns from that point forward.

Just wanted to illustrate that.

So you know I'm showing you a lot of pretty

pictures and this is just a very short talk,

so I haven't gone into a lot of detail,

but we have done dye studies.

This the black here is for

three different positions.

These were done a few years

ago on the Kootenai,

using a simpler version of

the particle tracking model.

The black lines are the predictions

of the model.

The dotted lines are measured

concentrations from sondes.

This is rhodamine demo experiment.

and this is at three different

locations down the river,

so these are explicit spatial

locations they're not cross sectionally

average or anything like that.

You can see it's not perfect,

but generally we're doing a really

good job of capturing both the temp.

The spatial variation as the peak drops.

And getting the concentration

at a specific point, correct?

That's not so easy getting the cross

sectionally average concentration

of point at cross sectionally

averaged concentration correct

is quite easy; getting it right

at a point means you've got the

distribution of the movement of

that material in a 2D sense correct.

And the dark shaded area is showing

you what the variability is across

the cross section.

Because we had several sondes,

but so again,

it's clear that we fit this individual

sonde which was void in the river quite

well using this kind of approach.

I'm gonna step to a slightly

different viewpoint,

so so far everything we've

talked about is been just single

calculations in a decoupled mode, but.

So we've done flow calculations

and then subsequent to that of

calculated the movement of material

and we can do that in an oil area

sense to just convey conventional

advection diffusion equation.

But there are situations where

we need to consider what goes on

at time step basis and we also

there are also situations where.

We want to have a coarse view

of the entire system,

but we want a very high resolution

depiction of what's going on

at some specific location.

I've showed you how cloning can

give that to some extent,

but about 2 years ago maybe

2 1/2 years ago

we were working on

trying to figure out how to

do what we call multigrid so,

not multiphysics,

but multigrid calculations where

we could have coarse grids,

couple of fine grids and also 1D

grids coupled with 2D grids and 2D grids

coupled with 3D grids within iRIC.

At the same time Don Cline came to

the USGS and started saying he wanted

us to work on coupling models and

we took a step back and we realized

it by taking a more general view.

We could make models couple

not only for multigrid.

And multi dimensionality,

but for in a much more general sense,

and so these are relatively new developments.

So I'm just going to show

you a couple examples.

I'm almost finished here, but.

This is a a rainfall runoff model

running on the on a 15 minute

time stamp with measured rainfall

actually partially measured,

partially interpolated,

using with Doppler data,

and this is the river in this basin,

so I'm just going to animate this.

We're going to see rainfall on

the left and we're going to see

the flow field on the right,

and these are coupled at the

time state basis and so.

Within the iRIC interface you set up

these models separately using an interface,

so there's no programming involved,

and then you run them and using MPI

with server software we manage both

processes and we run both processes together,

but that same general approach can

be used across any kind of physics,

including habitat models,

fish swimming models.

Anything you can come up

with that's written in Fortran,

C++, or Python.

you can couple together within the iRIC.

It's called iRIC MI framework

in order to have this

couple model capability.

so we should see rainfall on the

left and you'll see there's a lag,

but the flow comes up in the

river as this is running,

so this is a very simple

example of a couple of models,

But then the iRIC MI system.

Now you could have subsidiary coupling

and couple as many things as you want

and you can have disparate time steps,

so I could for example,

have a habitat model.

Running at the same time with

a bioenergetics model for fish

running at the same time,

things like that all within

the same computational frame.

So let's go to a little more

complicated situation again.

Rainfall runoff couple do a

2D unsteady model,

but now it's more realistic.

We've got something like ten

tributary basins we're running.

Ten rainfall runoff simulations

and one river model.

So you can see the inputs coming

down from the separate channels,

and this is an area where we

have detailed bathymetry again.

And you can see the blue dashed lines.

Those are actually lines

that were digitized as

the maximum extent of

flooding in this urban area,

so this is an urban flooding model

driven entirely by rainfall with

a river model all coupled together

running on the same time steps.

So I'm going to zoom in and

show you that inundacion.

As predicted from this approach,

only the only input is to

topography, the river roughness, and

the rainfall.

So you see,

we do a pretty good job of getting maximum,

and in addition we have two

ways of handling buildings.

You can explicitly put them in using

Google Street map digitization,

or you can treat them as porosity in

areas that are too complicated to resolve.

You can give a building porosity and

use that to do models like this,

so this is obviously this is a

very large flood and this city was,

you know, completely inundated.

But the model did a good job of

hindcasting that the important thing

to note is this kind of process.

Can be done in a forecasting

sense using rainfall from,

for example,

the national water model

or any other approach,

and predicting what's going to happen both

within the river and more importantly,

in this case on the flood plain.

So stepping too just very quickly

3D results in a 3D model.

You know you got verticals.

You've got vertical shear in

the primary flow and you've got

all kinds of things going on.

And then what we call secondary flow

flow with different directions over

the depth because of curvature or

bathymetrically driven calculations.

I'm not going to go through

this in great detail,

but we can do those two and then do tracking.

For example near the bed or

near the surface this is.

Please calculate this calculation was done.

It's the upper Kootenai.

There was a spill of diesel oil from

a train that derailed and went into

the Kootenai upstream of Bonners Ferry,

New Year's Day late on New

Year's Day of last year.

And I did these calculations in

support of that so this just shows

where surface material is going

and also more importantly where

it's where it sticks where it

hangs out over this hydrograph.

And you can see that you get

a lot of material along here,

and that was well supported

by observations in the field.

Finally,

you know you can go nuts at all

kinds of things together.

This is just a simulation of hurricane

Harvey and so there's the storm track again.

This is obviously this is hindcasting.

This work was done by.

Yasuhiro Ishida,

PhD student that finished recently.

Working for our group and.
NOTE Confidence: 0.85112363

So it's going to see the

storm proof the storm surge comes

in as a boundary condition.

We're going to kind of jump

around and look at

all the iRIC model results

for that situation.

Again, we can go from large

scale river stuff too.

Individual urban area inundations

through this process.

This isn't a particularly high resolution

model itself, but it's running.

Obviously, over a very large area,

essentially the greater Houston area.

Showing both inundacion

and river flow.

And again, the inputs are the

the sea levels and the rainfall only,

and a very good

Bathymetric topographic depiction.

So you get the idea and

then just in closing.

It's pretty easy to take these kind

of results and generate them into

things that you can look at on.

For example, a cell phone to see

what water inundacion looks like in

an urban setting on the buildings.

So that's sort of a whirlwind

introduction to what

High resolution, bathymetry and

urban area digitization can do

for our capabilities in making predictions.

I would be happy to answer any

questions anyone has or whatever.

So I'm done, thanks.

(Al)
Yeah, thanks John. This is really

been fantastic. Hearing it echo but.

If people have questions for John,

you can unmute yourself and ask,

or you could just type your

question into the chat.

We have a few minutes here so.

Please ask if you have any questions.

Oh, it looks like there's

a message I've been.

If, if you're if you don't have a question,

I ask that you mute yourself, yeah?

I was gonna post I'll I'll post another

link in here to the chat and this is a.

This is an inventory of of bathymetric and

Topabathymetric surveys that have been done.

It's it's probably not exhaustive, but

it's sort of the best list we have so far.

And a little earlier I pointed I

put it in the in the chat link to.

Sort of our really

Our first complete Topobathymetric.

terrain model for an inland river.

Which is on the Kootenai river

up in north Idaho.

And so there's a link to that there,

so that's that's where they've taken

You know bathymetric data like

what John has been showing,

integrated it with the land surface from
NOTE Confidence: 0.8210038

terrestrial lidar and produced a

single surface of the of the land,

including down under the water.

And that would be, I think,

ideal right John,

for for doing this kind of modeling.

We've been experimenting,

we've done a few pilot projects

on that sort of work,

and I can show you in here how

you can find some of that data.

Not all of it is been published yet.

But do people have questions

for John first. See.

(Cynthia)
This is Cynthia can you hear me?

(Al)
Yeah we can Cynthia.
(Cynthia)
I'm sorry I

don't have a question for John so much,

but this would be a good place for me

to insert regarding the inlandbathy/

topobathy inventory. The original

one was developed and put on ScienceBase.

as a data released product.

So the inventory is good through 2020

Surveys collected. but the footprints

and Map packages frankly were my

first attempt at trying to take all

kinds of data input into one format,

so they're pretty wonky,

so I apologize, but new.

Footprints are being developed and hopefully

will be out in another month or so.

So if you look at the map packages

the footprints do give you a

good idea of location

and extent of survey,

but many of them do not really

Show the Shapefile don't show the

boundary or the actual survey,

just so that people know that,

not shocked if they look at the map packages

and wonder what they're looking at.

(AL)
OK, and also I think you you mentioned

earlier to me, Cynthia that you you're

planning an update to that right?

(Cynthia)
And that I'm doing right and the 2020

inventory is going to stay as is.

But the footprints are in really sad shape.

'cause like I said this is my first

attempt so I'm redoing all those but

right now what's in the inventory is good

through surveys collected through 2020,

there's 75 or so more to add, but I

wanted to get the map packages cleaned up.

Before I go and add those on

top of what's already there.

But they are series, collected from the

first ones discovered up through 2020.

USGS inland survey, inland

Coastal zone surveys.

So it's it's pretty exhaustive.

I hope that's what could be found.

(AL)
Yeah, that's that's great.

That's great to have that information.

I'm sharing my screen right now.

Are you seeing it?

It's got the national map viewer on there.

I wanted to show people where they could

find data similar to what John was showing.

You know, it's not a lot of places yet,

but we do have a number of places

where we have this kind of information.

This topo bathymetric data and just

to show you how I got this.

You know, there's this layer list.

On the National map viewer.

And I'll put a link to the viewer

in the chat here in a second.

And you have to go in and find the list down.

Scroll down to the 3DEP elevation index.

Turn off these other things and just turn

on the topobathymetric data index.

That way you'll see these areas and

There's there's been quite a bit of

work done in some coastal areas on,

you know,

integrating the terrain and the

bathymetry together for coastal models.

So in that those go a ways inland as well.

So there's there's those,

and I know there's a new one being

developed in the Northern Gulf of Mexico,

and I just saw recently there was one

developed along the Southeast Coast

here in the Carolinas and Georgia Coast,

so there's quite a bit of that

along the coast.

And then there are these few inland areas

where we've done the Klamath River,

the Kootnai River,

part of the Colorado,

and I think this is the Niobrara.

In Nebraska.

So and then there's I don't even

think the the the

Potomac and the Delaware are showing up here.

They may be obscured,

but you can also if you click on this.

If you turn this query

layer on under the DEM,

The 3DEP elevation index and

turn off all of the other layers and

just leave these last three checked.

Then you can click on these and you can see,

for example,

here's.

This tells us about that Klamath

River bathymetry data set.

And it's got the bathymetric.

It says this is a bathymetric lidar

data set and you can actually get

a link to go actually download

it. And there are some other records

there with some additional other.

You know there's a that was

the bathmetry.

This is a total bathymetric

lidar and then I think.

Or maybe some other data

sets are too so anyway,

just to let you know how to find

This kind of data that we have.

Right now you know,

just mention again.

Like I said earlier,

we don't have a program yet to develop or

collect this kind of data everywhere,

but we're talking about it and

we are actually hopeful that we'll

be able to push that forward and

actually collect this kind of

data on a much more routine basis.

Something like what's been done

with the 3DEP program.

Hopefully that's in the future.

(Jim)
AL.
(AL) Yeah.
(Jim) This Jim Mitchell.

I'm just wondering everybody loves

you know out of the box data.

For these 3D applications you need to

have everybody on the same sort of datum.

And I'm wondering how you get you know datum,

corrected gauge heights and water levels,

that kind of stuff.

(AL)
Yes, that's that's obviously

critical for doing this kind of work.

I didn't see whether we have is Jeff

Danielson on the call. (attendee) No, he's not.

Yeah, he's our expert in doing that,

so he works at the EROS data center.

He would be the best person to answer

your specific question about that Jim, but.

It's definitely something that you

have to consider.
(Jim) Yeah, there's a.

There's a new data coming out in 22.

I guess I'll be released in 24.

Seems to me like we have an opportunity

to get everything on the same Dang datum.

Now we ought to do it.

(Al) yeah, and that that datum

shifts as time goes on to right?

so it's not going to be...

Yeah, exactly exactly.

Thanks, good question.

Definitely something that has

to be taken into account here.

Was there anything further I didn't see?

(Vanessa)
Matt Mercurio had a question earlier and

he says Cool AR on the phone. Anymore

info on that? I guess that's for John.

(John)
I'm sorry I didn't

understand the question AR.

(Vanessa)
Augmented reality on the phone.

That stuff you showed on the phone,

he says. Yeah, it's yeah.

(John)
It's basically those graphics, images

are developer right out of iRIC,

so it's part of the graphics package.

You know, putting them on the cell

phone like that is once you have that

image is pretty straightforward,

but I have to admit in that slide

that was done just by putting the

image on over a cellphone frame.

If you see what I mean.

So there's no, there's no

fancy AR software involved.

The realizations are 3D within iRIC and then.

So we're just drawing drawing the surface,

especially within Google Maps.

Or excuse me, Google Street

map to get those images.

Does that make sense?

I'll assume it does, unless someone

sends me another question. To me

it does. That's Matt's question.

I notice that someone asked about,

you know why do we need river bathymetry

and that that's a really good question,

and the problem is that in gigantic floods

where the conveyance in the channel is

a tiny fraction of the total amount of

water going out on the land surface,

bathymetry is not very important,

but there's a whole lot of floods.

In between normal river flows and

those sort of NOAA's flood events

where knowing the conveyance in

the channel and knowing where water

comes out of the channel is extremely

important for understanding flooding

and especially urban flooding.

Because unfortunately Rivers don't just sort

of overflow their banks and move out,

sort of rectilinear under the

floodplain like we like to think.

I mean they come out as channels and

(inaudible) and they shoot water like a

fire hose onto the flood plain so knowing

the details of the bathymetry in the

channel I would say for all but the

very largest flood is really critically

important for making good predictions.

(AL)
John, we're getting short on time,

but I'm able to stay longer.

If you are, there are some more questions

that are showing up in the chat.

One is basically can outside

scientists or groups access and

use software modeling software.

That iRIC software I think is yes.

Absolutely now all in the public domain.

It's available at i-ric.org.

It's all free.

There's 20-some solvers,

they all have tutorials.

There's a developer's manual,

an one of the things I like to emphasize.

You know, we we taught like 150

classes over the last eight years, but.

You know people learn this from

the tutorials on the website all the time,

all over the world and people.

There's also a developer's guide showing

you how to put your own solver into iRIC.

So this is not a static piece of

software that USGS puts out its

something that's not changing.

It's changing all the time.

People are putting new solvers

in and making them available.

It's a community effort.

We welcome other people.

There is a developer's guide that

in explicit detail tells you how

to put your own solver into iRIC,

and it's it's quite straightforward.

The models aren't really

Decoupled from the interface itself.

There just introduced to the

interface through XML,

which is just a markup language

that tells the interface how

to interact with the model,

what parameters are needed,

and what tools are needed

graphically to build coordinate

systems and things like that.

We have something like 20 coordinate systems,

well structured and unstructured,

so you know, but by all means,

if you have a different

coordinate system come on down,

you know we'll put it in there for you,

and again,

we welcome any collaborators

to build this out.

For everyone to use,

it's widely used,

especially in Asia because everybody lives

on the floodplain and rivers are just

generally thought to be more important.

But it is used around the world

and we have lots of collaborators.

Again, we welcome anyone else to join us.

Yeah, along those lines,

(AL)
Could you explain the overlap and

differences between iRIC and HEC-Ras?

(John)
So iRIC, who was originally conceived.

To support Morphodynamic models and it has

since involved to do a lot of other things,

but one of the things that we still

do is we predict with bold 1, 2 and 3D

models how the bed changes overtime

with most of the models in iRIC.

So we have sentiment transport,

well bedload suspended load and we use

that to modify the bed over a hydrograph

to predict deposition on the floodplain,

for example. And generally we're strongly 2D/3D.

More on the hydrodynamic side,

we believe HEC-ras does a

great job for 1D models of really.

We haven't put much effort into that.

We're trying to compliment what the Army Corp

has done with a more 2D and 3D approach,

including turbulence resolved models.

There are models in iRIC.

You can start with a flatbed and have

a turbulence resolve DNS model and

predict the formation of very complex

bed forms that are extremely realistic.

So this is really a state of the art tool.

For bed change,

as well as just a tool for what

I've shown you today.

In fact,

really what I've shown you

today is was not the original

center focus of the iRIC work.

It was really about morphologic evolution

and to some to some degree still is.

Although we do have a lot of

other tools.

we have one simple groundwater model

and I'm working with Chris L

to incorporate at least modflow 6.

Especially because we want to use

it in some of our coupled models

to look at bank erosion.

We do have some bank erosion now,

but we want a dynamic groundwater

problem coupled to the river

problem because we believe it's

important in a lot of cases.

(AL)
OK, great, well we are a couple

minutes over and I don't see any more

real burning questions in the chat,

so I think we'll wrap it up for today.

John, I really want to thank

you for the presentation.

I think this is really opened my eyes

to all the possibilities that are.

Out there for working with

data like the bathymetry data

that we've been talking about,

so really appreciate your taking

the time to show us this and.

Will try and keep in touch and

see see how things develop.

(JOHN)
Yeah my my pleasure on the

presentation an you know I can't.

I can't really emphasize enough.

How important the bathymetric effort

is to guiding future development,

especially for real societal problems like

urban flooding and dispersal of contaminants.

Things like that.

I mean, having good topography is

the key to accurate prediction.

There is no question about it.

So I hope that over the next decade or

two we move away from a sort of river

averaged approach or cross sectionally

averaging approach to a more spatially

explicit one,

and I think the new techniques

give us the opportunity to do that.

So I'm excited about what you guys are doing.

(AL)
Excellent, thanks again.

We'll call it a day for for now.