Linear Referencing

NHDPlus and Flow Network= s

Prepared by Tyler L. Jantzen, Nishesh Meh= ta and David R. Maidment

GIS in Water Resources

Fall 2007=

Goal<= /span>= = 1

Computer and Data Requirements= . = 1

Data description: = 1

Exploring NHDPlus. = 2

NHDPlus Flow Attributes= . = 6

Building a Stream Network= . = 9

Network Tracing. = 12

Identifying Catchments using the USGS gages= . = 14

Longest Flowpath. = 22

Using the Base Flow Index= . = 24

Land Use Statistics= . = 28

Land Cover Cumulative Characteristics= . = 28

Summary of items to be turned in:<= span style=3D'mso-bookmark:_Toc479668056'>= 33=

Goal

The goal of this exercise is to explore NHDPlus and so= me concepts of linear referencing on network flow lines. In the next exercise (Exercise 6) = in the course, we’ll use these inputs to create an Arc Hydro with Time Series geodatabase for these data. <= /p>

= Computer and Data Requirements

ArcInfo version of ArcGIS. The d= ata are provided in the accompanying zip file, http:= //www.ce.utexas.edu/prof/maidment/giswr2007/ex5/ex5.zip =

<= span style=3D'mso-bookmark:_Toc179617223'>Data description:

SanM= arcos_NHDPlus_raw.mdb, which is the entire San Marcos= basin cut from NHDPlus Region 12. Its structure is shown below. The N= HDPlus is based on the National Hydrography Dataset Medium Resolution (1:100,000 scale), which was originally developed by the USGS. More information about the National Hydrography Dataset can be found at http:/= /nhd.usgs.gov/. More information about NHDPlus can= be found at http://www.hor= izon-systems.com/nhdplus/. Elevation-based flow direction gri= ds, flow accumulation grids, and catchments have been developed. The National Land Cover Dataset ha= s been applied to these catchments, and information regarding catchment and waters= hed land cover linked to flowlines. Mean precipitation has been calculated for each catchment, and basic streamflow modeling has been conducted for each reach. Two methods have been used to dete= rmine mean annual flow and mean annual velocity.= Thus, it is possible, without any additional calculations, to estima= te stream flow and velocity at any given reach. Value Added Attributes, which assi= st with tracing throughout the stream network, have also been added. Lastly, USGS Streamflow Gages have= been snapped to the network, allowing nearly seamless integration between NHDPlus and the National Water Information System (http://waterdata.usgs.gov/nwis/r= t). The snapped USGS Streamflow Gages = are available through the USGS at http://water.usgs.gov/GIS/dsdl/USGS_Streamgages-NHD_Locations_GE= ODB.zip.

Exp= loring NHDPlus

Now lets hav= e some fun and explore the delights of NHDPlus.

<= /o:p>

Open ArcMap and from the SanMarcos_NHDPlus_Raw.mdb geodata= base, add the entire Hydrography fea= ture dataset and the Catchments fea= ture class from the Hydrologic Units feature dataset. Recolor the symbology display and = lets take a look at what we’ve got here.&= nbsp; Save your map display as SanMarcos.mxd. Here is th= e map that I created:

<= /o:p>

You’ll= notice that I have symbolized the NHDPoin= t feature class using the attribute = FType, which is an NHD classifier for types of water points on maps – in this case, you can see that we have only two types of points: SpringSeep and Well. Think about the s= patial patterns of these points on the map and remember the intersection we did in Exercise 2 of the Edwards aquifer and the Guadalupe basin.

<= /o:p>

To be turned in: What does the pattern = of location of the SpringSeep and Well points tell you about the groundwater system underlying this basin?

<= /o:p>

Now lets tak= e a look at the USGSGageEvents feature class. Open its attribu= te table. You’ll see a set= of stream gage descriptions highlighted, select the gage for the Blanco River at Wimberley:

<= /o:p>

If you go hit Selected and scroll along this record, you’ll see some statistics of = the presently available daily streamflow information from this gage<= /span>

<= /o:p>

This shows t= hat we have data from 1 September 1924 to 30 September 2004 available from the NWIS archive, a total of 28368 days of streamflow record. Those USGS people do cool stuff fo= r us! Now lets scroll back along the att= ribute table a bit and whoa, what do we find? Our old friends the percenti= le values for USGS water data that we worked with in Exercise 2! For example, P1 =3D 5.= 8 means that on 1% of the 28368 days in the record, the flow is less than or equal = to 5.8 cfs.

<= /o:p>

Now, lets tr= ansfer these data into Excel and make a plot of them. I found it easiest just to manuall= y type in the data rather than messing around with table exports and the like thou= gh you could that too.

<= /o:p>

Does it look= like these flows are normally distributed? I don’t think so! So, lets right click on the Discharge axis and change the axis type to Logarithmic. Now what do we get:

<= /o:p>

Oh, this loo= ks a lot more like a nice normal distribution, so this tells us that these flow data= may follow the lognormal distribution, though it would take a considerable amou= nt of statistical characterization beyond what we have done here to confirm th= at.

Now, lets ge= t some index information that tells us about drainage area (DA_SQ_MILE) and mean annual flow (AVE). If you go = along to the attribute table, you can extract their values as 355 mile² and 142.183 cfs, respectively for the Blanco River at Wimberly.

<= /o:p>

Lets modify = our plot header to add this information:

<= /o:p>

To be turned in: Prepare a plot of the cumulative frequency versus log of the discharge for the San Marcos River at San Marcos with drainage area and ave flow annotated on it. Compare this plot and the correspo= nding data with the one in the exercise for the Blanco River at Wimberley. The avera= ge flows are similar but the drainage areas and flow variability are vastly different. What does this tel= l you about the source of water flow at the two gaging sites?

<= /o:p>

NHD= Plus Flow Attributes

<= /o:p>

A design goa= l of NHDPlus is to be able to provide enough land surface attributes to permit estimation of the mean annual flow for each NHDFlowline, rather than just at gages.   To find out what= those values area, go to your ArcMap display again and add the Table flowlineattributesflow If you open this table, you’= ll see a number of attributes for each flow line, among them MAFLOWU and MAFLOWV.<= span style=3D'mso-spacerun:yes'> These are mean annual flows derive= d from (1) taking a mean annual runoff map in mm/year and integrating it over the drainage area, and (2) estima= ting the mean annual flow using some regression equations developed by Rich Vogel and collaborators at Tuft= s University.   These methods are described = in the NHDPlus documentation at = htt= p://www.horizon-systems.com/nhdplus/data/NHDPLUS_UserGuide.pdf   We’ll use the MAFLOWU v= alues because they are available for all lines (the Vogel equations have a restri= cted range of drainage areas to which they can be applied and the -9999 values in the table below are missing values outside of this range).   But the difference between t= he two values (e.g. 41.0 cfs and 30.8cfs) gives you a sense of the imprecision of = the science of spatial flow estimation at this time. The corresponding estimates of mean annual flow velocity are much closer together (e.g. 1.24 ft/s vs 1.22 ft/s)= . This information is important for = water quality computations because it tells us how long organisms and chemicals w= ill take to travel through each stream reach.

<= /o:p>

<= /o:p>

The key fiel= d that links the attribute tables with the features is COMID, which behaves for NHDPlus as HydroID does for Arc Hydro (and in fact the idea for HydroID came from COMID), so lets join the attribute table to the NHDFlowline feature class.   Right click on NHDFlowline feature class and sel= ect Joins and Relates/Join, fill in t= he fields as shown below, and say Yes= when you are asked to do Indexing to support the relationship.

<= /o:p>

<= /o:p>

Now, lets sy= mbolize our flow lines using the MAFLOWU attribute. Use Quantities/Graduated Symbols and the default classification method (Natural Breaks (Jenks)).=

<= /o:p>

And now we s= ee this marvelous map of streams mapped by flow size. Oh happy day!! Those of us who have spent years looking at blue line maps where all the lines are the same thickness regardless of the size of the river can appreciate the marvel of being able to see instantly the hydrologic flow pattern of the landscape!

<= /o:p>

<= /o:p>

Now, lets se= e how these flows compare to what the gages tell us.   Select the USGSGageEvents fo= r the Blanco River at Wimberley and zoom in there. If you use the Identify tool in Arc Map and click on the NHDFlowline each ther= e, you’ll see that its attributes are MAFLOWU =3D 213 cfs, and MAFLOWV = =3D 121 cfs, while the stream gage at that location tells us that the actual mean annual flow is 142 cfs. You can see that Vogel’s regression equations give a much more accurate value than simply integrating over a mean annual runoff map at this location. = But, of course, the comparison you’ve just made with the San Marcos River at= San Marcos also s= hows that local factors can be crucial in defining the mean annual flow of a riv= er!

<= /o:p>

To be turned in: Make a map of the San Marcos basin NHDFlowlines attributed with mean annual flow, and compare the flow estimate from the mapped flow line with that at the stream gage for the Blanco River= at Kyle.

<= /o:p>

Bui= lding a Stream Network

<= /o:p>

There is lot= s of other fun stuff we can do to explore NHDPlus, but for now, lets move on to = creating a geometric network of the NHDFlowlines.   Save your SanMarcos.mxd ArcMap display file, close Arc Map an open Arc Ca= talog. You have to have the ArcInfo version of ArcGIS not the ArcView version to do the next step. Right click on= the Hydrography feature dataset and s= elect New/Geometric Network.

<= /o:p>

<= /o:p>

Click Next to the first Panel that disp= lays, in the second one, click on NHDFlowline to be included in the network, say Yes to preserving existing attrib= ute values, and No to do you want complex edges in your network, No<= /b> to Do your features need to be snapped, and No to Do y= ou want to assign weights, then hit F= inish.

<= /o:p>

You’ll= see a computation box emerge and then go away as the connectivity table is built,= and then, you’re done. Now = if you look at your Hydrography feature dataset, you’ll see you’ve got some new things in it, a Hydrograp= hy_Net network and a point feature class of Hydrography_Net_Junctions that were created during the network flow creation.

<= /o:p>

Close Arc Ca= talog and Open a new ArcMap display, and add the Hydrography_Net network to the map display and appropriately symbolize it.   Save your ArcMap display as = SanMarcosNetwork.mxd

<= /o:p>

<= /o:p>

You can see = all these new junctions that have appeared as part of the network building process. In Arc Map, use View/Toolbars/Utility Network Analyst= to add the Network Analyst toolbar to your ArcMap display. Use Flow/Display Arrows to symbolize your flow directions. Oops! Looks bad. A lot of black dots.

<= /o:p>

<= /o:p>

The NHDFlowL= ines have an attribute FlowDir that indicates what direction the flow proceeds on them. In this case, FlowDir =3D 1 means = that the flow direction is the same as the direction of digitization of the lines in= the NHDFlowline feature dataset.

<= /o:p>

Use View/Toobars/Arc Hydro 9 tools to= bring up the Arc Hydro toolbar (if you don’t have this toolbar, see the instructions in Exercise 4 for acquiring these tools). Then in the Arc Hydro tools Network Tools/Set Flow Direction = . Click on NHDFlowline in the Layers window to select the feature class. Click on the Assign based on attribute option and choose FlowDir as the attribute. Click OK to initiate the assignment of flow direction and pretty soo= n, you’ll see your bad black blobs turn to nice black arrow heads and fl= ow direction is assigned.   = What this means is that each NHDFlowline had a numerical attribute (0, 1, 2, 3) = for that line whose values correspond to different types of flow direction, as shown below, and this direction is now assigned internally within the geome= try of the network so that the Utility Network Analyst can correctly interpret = it.

<= /o:p>

<= /o:p>

<= /o:p>

Net= work Tracing

<= /o:p>

In the Utility Network Analyst toolbar, = use Flow/Display Arrows again to togg= le off the flow direction arrows and also click off the display of the generic junctions so you just leave the network itself displayed in Arc Map.   Use the Utility Network Analyst Add Edge Flag Tool to add an edge flag near the outlet= of the network.

<= /o:p>

And then sel= ect Trace Upstream to identify all the upstream edges. Hit the symbol on the end of the toolbar to initiate the trace.

<= /o:p>

And Voila!! = You get this marvelously selected network. Use Analysis/Clear Results<= /b> to get a clear network again.

<= /o:p>

If you use <= !--[if gte vml 1]> , you’ll just get a “bong” noise that indicates that you ha= ve no disconnected edges. Ok, Houston, we have liftoff!! Disconnected edges = are not good in networks because they require editing to be fixed or they indic= ate isolated streams whose linkage to the stream network is unknown.=

<= /o:p>

Use Analysis/Clear Flags to get rid of the Flag that you laid down, and then Analysis/Options to switch the Results from Drawing to Selection. This h= as the effect of allowing us to actually select records with network traces, rather than just see a red graphic network as we just did. Click Apply and Ok, to c= lose out this window.

<= /o:p>

        =

<= /o:p>

Ide= ntifying Catchments that drain to a USGS gage

Using the NHDFlowlines and Catchments feature classes we can find out which catchments are associated with a given USGS gage. To use this capability we first build a relationship between catchments and flowline feature classes. Save your map. Open ArcCatalog. Browse to the folder you have stored the exercise data to the SanMarcos_NHDPlus_raw geodatabase. Right click on the geodatabase and scroll to new relationship class.

Click to open the following window. Let’s call t= he new relationship class CatchmenthasFlo= wline. For origin feature class = browse scroll to Hydrographic units and associate it with Catchment feature class. In destination feature class under hydrography to NHDFLowline.. C= lick Next.

In the next window select Simple (peer to peer) relationship and click next.

The next screen inquires the direction of message propagation within the database. Click Both. This enables queries from either feature class to return data from the rela= ted feature class. Leave the labels as they are.

Click next. In the next window we must specify the nature of relationship. Since sometimes= a single catchment may be associated with more than one flowline we choose 1-M. Select that and click next.<= /p>

Next, choose not to add any attributes to the relation= ship class. Click next.

In the next window we must specify which field will co= nnect the two feature classes. Like previously, we use COMID. In both drop down menus pick COMID and click next.

Click Next and then Finish.

Now you have anew relationship class in your database.= Our data is ready to find associated catchments. Open ArcMap and add the feature classes Catchment, USGSGagEvents and the geometric network to it. Zoom to a USGS station of your choice.

Place an Edge= Flag on the USGS gage.

Trace upstrea= m. Click the Solve button (Make sure that you have set Analysis/Options to Se= lection.)

This gives you the selection of the flowlines flowing = into the USGS gage. To find the catchments that are associated to the USGS gage = we will now invoke our relationship class we created previously.

Right click the NHDFlowline feature class and open the attribute table. You can see the selected featur= es. Click Options and browse to Re= lated Tables. You will see the relationship we created shown as CatchmenthasFlowline. Click the table.

This opens the Attribute Table of the Catchment feature class. You can see that the Catchments relate= d to the selected flowlines are automatically selected.

Using this table we can now find the number and the ar= ea of the catchments associated with any USGS gage.

To be turned i= n: Make a layout of the related catchments and flowlines to the USGS gage near the = Blanco River at Kyle. Find the number and the total area of the catchments associated wi= th gaging station. What percent of the total San Marcos basin does this constitute? Compare it with= the area given in the USGS gage feature class.

Lon= gest Flowpath

Use Selection/Clear Selected Features= in the ArcMap toolbar to clear all the selections you made in the last section= . Use Analysis/Clear Flags in the Network Utility Analyst Toolbar to clear the Edge Flag you’ve been using.

<= /o:p>

Zoom into th= e top of the network, place an edge flag there and execute a TraceDownstrea= m

And if you z= oom back to the extent of the layer, open its attribute table and hit Selected for t= he records, you’ll see that 93 of the 557 flowlines in the network have = been selected by this process.

<= /o:p>

If you move = along to the LengthKm attribute, right = click on the field and select Statistics. You can get a table of statistics of the selected records, from which you can find the average line length in km and the total length along the f= low path that you’ll need shortly.

<= /o:p>

<= /o:p>

Go through t= he same steps that you used earlier to add the flowlineattributesflow table to the ArcMap display and to Join it to the NHDFlowline feature = class using COMID as the common key feature. Right click on the Slope attribute (last one at the = end) and use Statistics to get the average slope of the lines along the flow path. The Slope value is give= as % slope, so don’t be alarmed if the value seems large to you. Do the same on the MAXELEVSMO attribute (elevation in meters above mean sea level)= and use the Maximum and Minimum of the Statistics values to tell you the elevat= ion drop over this flow path.

<= /o:p>

Save again y= our ArcMap display as SanMarcosNetwork= .mxd so you can come back and get this information if you need it later.=

<= /o:p>

<= /o:p>

To be turned in: What is the total flow length from top to bottom of the San Marcos Basin (km). What is the average len= gth of the 93 NHDFlowLines on this flow path (km). What is the average slope of these= lines (%). What is the elevation dr= op along the flow path (m).   If you divide the elevation drop by the flow path length, do you get a result consistent with that from the mean slope computation?   If it is different is that o= k? Think carefully about this last question before you answer it.   The average of a set of rati= os (a/b) is not the same thing as the= ratio of the average values of a and= b.

Usi= ng the Base Flow Index

Open the attribute table of the USGSGagEvent feature c= lass. One of the fields present is called BFI_Ave. This refers to the average Ann= ual Base Flow Index (BFI). BFI refers to the percentage of the flow which is Ba= se flow i.e. that part of the stream<= /span> discharge that is not attributable to direct = runoff from precipi= tation or melting snow; it is usually sustained by groundw= ater. The Base Flow Index (BFI) is used as a measure of the base flow characteris= tics of catchments. It provides a systematic way of assessing the proportion of = base flow in the total runoff of a catchment.¹ Let’s try and explore it further. We will symbolize the Basin using BFI on the gaging stations. Right click the USGSGagEvents feature class and go to properties. Under the Symbology tab go to = Quantities and pick graduated sym= bols. In values select BFI_Ave from = the drop down menu.

Click OK.= Go to Properties/Labels and set the Label Field to be BFI_Ave<= span style=3D'mso-spacerun:yes'>   Right click on the USGSGageE= vents feature class and Label Features You can produce a display like the one shown below.

These labels have three trailing 0’s that are unnecessary. To eliminate the= m, right click on the USGSGageEvents feature class, and select Properties/Labels and the box for= Expression. Enter the following in the resulti= ng Expression window:

“[BFI_A= ve]:” & round ([BFI_AVE], 5), as shown below:

What this does is create a label that includes the att= ribute BFI_Ave as text and then conconcatenates onto this the rounded version of t= he attribute value. The result t= hen appears in ArcMap as shown below. If you don’t want the [BFI_AVE]: prefix, then just use the expression round ([BFI_AVE], 5)

This shows that at some gages 95% of the flow is Base Flow while at other location= s only 5% is base flow. For more information on how to determine base flow, see http://www.usbr= .gov/pmts/hydraulics_lab/twahl/bfi/ from which the following picture h= as been extracted. Base flow is = the flow beneath the red line in this graph and surface runoff is the excess fl= ow defined by the difference between the blue line and the red line.

Why should there be so much variation in the base flow= in the San Marcos Basin? The stations on the San Marcos river at San Marcos and Luling are right above the Edwards Aqui= fer and draw significant amounts of water from it to fulfill their needs. Let&#= 8217;s visualize the Edwards aquifer to get a better understanding of this phenome= non.

Add the Aquif= er feature class from the Geomorpholo= gy dataset in the geodatabase. This is the Edwards Aquifer.   I have symbolized this as we = did in Exercise 2, and also sized the rivers according to their flows. Cool!

To be turned i= n: Make a layout of the Edwards Aquifer overlayed on the San Marcos basin. What is the river on the right tha= t has low base flow. What is the ri= ver on the left that has high base flow. Make a base flow balance between these two rivers to see if you can reasonably predict the base flow at the most downstream station (the San Marcos River at Ottine).
Lan= d Use Statistics

NHDPlus provides a host of tables that provide detailed information about Land Use patterns. A description of the data available in= the NHDPlus dataset may be found on the NHDPlus user guide

= http://www.horizon-systems.com/nhdplus/data/NHDPLUS_UserGuide.pdf

on pages 33-34, 38-39, appendix C1 pages 118-120. Let = us use it to find something interesting.

Lan= d Cover Cumulative Characteristics

The National Land Cover Dataset is described at: http://landcover.usgs.gov/cl= asses.php

In the NHDPlus, in the Catchme= ntattributesnlcd table, the NLCD % areas are computed for each catchment and in the Flowlineattributesnlcd the corresponding attributes are calculated for the total drainage area (including the catchment of the flowline itself):

_{=

(1)}

Where CUMNLCD_n =3D % of cumulative drainage area of flowline n classif= ied in a particular NLCD Class, NLCD_i =3D % of area of catchment i in t= his land cover class, and i =3D 1, 2, …, n-1 are the set of catchments up= stream of catchment n. Areas characterized by at least 30 percent of constructed materials (e.g., asphal= t, concrete, buildings, etc) are in NLCD class 22:

High Intensity Residential – Highly developed areas where large numbers of people reside, such as apartment complexes and row houses. Vegetation occupies less than 20 percent of the area, and constructed mater= ials cover 80 to100 percent. Here = are the Catchmentattributesnlcd an= d Flowlineattributesnlcd tables:

We will visualize the San Marcos basin characterized by % of drainage area classified as High Intensity Residential.

First, add the table Catchmentattributesnlcd to the map. Right click Catchment<= /b> and click create Join.

Match COMID of the Catchment Flowline feature class to= the COMID of Catchmentattributesnlcd table.

Click OK. Now the join has been created.

Let us visualize the Catchments using High Intensity Residential. Right click on Catchment and click properties. Under the Symbo= logy tab select quantities as graduated colors. In Values select catchmentattributesnlcd.NLCD_22. = Pick an appropriate color scheme.

Click Apply and OK.

You can now see the high intensity residential areas. = Which areas do you think might face water supply shortage in the near future?   Now lets explore the upstream drainage area properties. Join the NHDFlowline feature class to the Flowlineattributesnlcd table using COMID as the relate attribute (notice that since this is the second table you’ve joined to NHDFlowl= ine the nomenclature is a little bit more complex this time).

Symbolize the lines with the % in CUMNLCD_22. You can see how different parts of= the stream system are affected by upstream land cover.

To be turned i= n: Prepare a layout of the San Ma= rcos basin characterized by intensity of residential development. What is the total area (km^{2)
has a local land cover that is at least 30% high intensity residential? For
each of the stream gage locations, what is the % of upstream land cover tha=
t is
high intensity residential?}

Sum= mary of items to be turned in:

(1) What does the pattern of location o= f the SpringSeep and Well points tell you about the groundwater system underlying this basin?

(2) Prepare a plot of the cumulative fr= equency versus log of the discharge for the San Marcos River at = San Marcos with drainage area and ave f= low annotated on it. Compare this= plot and the corresponding data with the one in the exercise for the Blanco River at Wimberley.   The avera= ge flows are similar but the drainage areas and flow variability are vastly different. What does this tel= l you about the source of water flow at the two gaging sites?

(3)&nb= sp; Make a map of the San Marcos basin NHDFlowlines attributed with mean annual flow, and compare the flow estimate from the mapped flow line with that at the stream gage for the Blanco River at Kyle.

(4) To be turned in: Make a layo= ut of the related catchments and flowlines to the USGS gage near the Blanco River at Kyle. Find the number and the total area of the catchments associated wi= th gaging station. What percent of the total San Marcos basin does this constitute? Compare it with= the area given in the USGS gage feature class.

(5) What is the total flow length from = top to bottom of the San Marcos<= /st1:PlaceName> Basin (km). What is the average length of the = 93 NHDFlowlines on this flow path (km). What is the average slope of these lines (%). What is the elevation drop along t= he flow path (m).   If you d= ivide the elevation drop by the flow path length, do you get a result consistent = with that from the mean slope computation?   If it is different is that o= k? Think carefully about this last question before you answer it.   The average of a set of rati= os (a/b) is not the same thing as the= ratio of the average values of a and= b.

(6) To be turn= ed in: Make a layout of the Edwards Aquifer overlayed on the San Marcos basin. What is the river on the right tha= t has low base flow? What is the ri= ver on the left that has high base flow? Make a base flow balance between these two rivers to see if you can reasonably predict the base flow at the most downstream station (the San Marcos River at Ottine).

(7) To be turned in: Prepare a l= ayout of the San Marcos basin characterized by intensity of residential area. What area (km^{2)
has an upstream land cover that is at least 30% high intensity residential?=
For
each of the stream gage locations, what is the % of upstream land cover tha=
t is
high intensity residential?}