Grassy Cove, North America's Largest Sinkhole

Grassy Cove Community Center, Cumberland County, Tennessee

Nestled in the southeastern part of Cumberland County, Tennessee lies a quite community known as Grassy Cove. Its flat fertile land is home to scenic farms and farmland which are still owned by the descendants of the first European inhabitants of the area.

Without looking more closely at the landscape, it might escape the casual viewer that one was sitting inside a giant bowl. But with the aid of maps, it becomes readily apparent that Grassy Cove isn't like a regular hollow.

My first rule as a geographer is "Ask a local". Locals know a lot. For example, it's long been known that if it rains too much in Grassy Cove, then the flat farmland turns into a lake. It's also known from the days of yore that discarded corn shucks in the cove will appear in the Sequatchie River a few miles away, a few days later. What's going on here?

Grassy Cove flooding, Cumberland County, Tennessee 2  

Structurally, there's a lot to be said about Grassy Cove; it's not just a simple sinkhole. Its formation is the middle stage of a grand weathering process, with hints of its beginning and end observable along the axis of Sequatchie Valley Anticline, a giant geomorphological feature that spans two states. The Sequatchie Valley Anticline is a remnant of the same compressional forces which created the ridges and valleys in the physiographic province of the same name east of Grassy Cove. The Sequatchie Valley Anticline is a huge structural deformation where the bedrock was folded. Imagine a carpet getting scrunched up; the spots where it bends upwards are anticlines, downwards are sinclines.

Anticline, Sequatchie Valley, Cumberland County, Tennessee

The anticline has been around for a really long time, slowly being weathered away. The erosion process means that it is retreating northward, so features at its northern extent may offer a glimpse into what happened further south longer ago (that is what is happening now in Grassy Cove). And it also lets us see what the future of Grassy Cove is (to be part of the Sequatchie Valley).

Currently though, Grassy Cove is drained completely through a cave system and the water from there emerges at Devilstep Hollow, at the head of the Sequatchie Valley. 

Devilstep Hollow Cave, Head of Sequatchie Unit, Cumberland Trail State Park, Cumberland County, Tennessee 2

As stated, this is the largest sinkhole in North America (which I am aware of). Back in 2013 Tom Dunigan and I crunched some numbers for sinkholes in Tennessee. I made areal and volumetric estimates based on 10m Digital Elevation Models (DEMs). Now, 2.5' DEMs area available, which are 172 times more detailed than the 10m DEMs. I wanted to make a renewed comparison and try to tease out more detailed information.

Map of Grassy Cove, Cumberland County, Tennessee

Below are the updated figures for the region based on the 3DEP lidar.

Region Measure                 Value Unit
Watershed Area 33.17 Kilometers2
Watershed Perimeter 42.15 Kilometers
Sinkhole Area 13.31 Kilometers2
Sinkhole Perimeter 35.71 Kilometers
Sinkhole Major Axis 7.67 Kilometers
Sinkhole Minor Axis 4.36 Kilometers
Sinkhole Axial Declination 30.80 Degrees
Sinkhole Minimum Elevation 467.27 MASL
Sinkhole Maximum Elevation 507.24 MASL
Sinkhole Depth 39.97 Meters

Overlook, Grassy Cove, Cumberland County, Tennessee 1


Sandstone Boxwork

Sandstone boxwork is a unusual phenomenon that we see on the margins of the Cumberland Plateau.

There are three things that must happen first to produce boxwork.

Photo credit: Brian Solomon / @waterfall_hillbilly


Produced in a parent rock of sandstone, this type of boxwork originates as small joints which run in a box shape. The origins of these joints could be from frost wedging - where water in the rock freezes and cracks the rock, or it could unloading pressure from the rock de-watering, or it could be from some other physical process. The origins of the boxwork joints are still a bit of a mystery to me.


The joints become a place of preferential mineralization. Groundwater loaded with iron minerals moves through the pores of the sandstone. Where it finds a void, like in the joints, it tends to drop minerals out of solution. The rust colored minerals you find on your water heater at the house are chemically quite similar to this boxwork. 


The joint surface must first be exposed to weathering, whether by cliff failure, slump, or some other wasting process. The exposed surface's sandstone weathers out first, leaving in place the mineralized boxwork. Because the sandstone around it is removed, the boxwork tends to protrude after its been well developed.

Similar Features

Boxwork is a type of Liesegang ring. Liesegang rings lack obvious structural control, as in they do not have joint influenced development.

Painted Rock, Painted Rock Cave, Big South Fork NRRA, Fenress County, Tennessee Liesegang rings, Rockcastle conglomerate, Eagle shelter, White County, Tennessee 2 Liesegang rings, Claire Murphy, Fentress Formation, Big South Fork NRRA, Scott Co, TN


KTAG - Cave Density Map

Caves of Kentucky, Tennessee, Alabama, and Georgia (KTAG)
This is a map showing the density and distribution of caves in Kentucky, Tennessee, Alabama, and Georgia (KTAG). The strong line of density extending north-east from the intersection of Tennessee, Alabama, and Georgia shows the western escarpment of the Cumberland Plateau where Mississippian aged limestones outcrop and are exposed to weathering processes.

At the southern end of the Cumberland Plateau it becomes more heavily dissected, with islands of Plateau present amidst the Highland Rim. Along the margins of these islands caves can be exposed, which explains the dispersion at the southern extent of the Plateau.

Central and Western Kentucky show a secondary area of cave density; a ring of Mississippian aged carbonates has been exposed from the weathering of the Cincinnati Arch. Mammoth Cave resides in the densest part of this region.

This map was created using data from the following sources:

  • Alabama Cave Survey (ACS)
  • Georgia Speleological Society (GSS)
  • Kentucky Speleological Society (KSS)
  • Tennessee Cave Survey (TCS)

ArcMap 10.3 by ESRI was used to create the map. Workflow follows.

· For Tennessee, Alabama, and Georgia

  • Tabular data joins in Access, output to Excel table
  • Format Excel tables by converted DMS to DD
  • Reference E1 (only one entrance per cave; the first documented entrance) points
  • Spatially join points to county feature class and display count as label
  • Kernel Density

· For Kentucky

  • Summarize county table and tabular join to county feature class and display count as label
  • Summarize 7.5' quadrangle table and tabular join to 7.5' quadrangle feature class
  • Feature to point (create centroid) of quadrangle feature class; preserve count attribute
  • Kernel density with new quad centroid feature class with count as Z attribute

cave, location, density, distribution, records, cave location, cave density, cave distribution, cave records, National Speleological Society, NSS, Alabama Cave Survey, ACS, Georgia Speleological Society, GSS, Kentucky Speleological Society, KSS ,Tennessee Cave Survey ,TCS, Kentucky, Tennessee, Alabama, Georgia, KY, TN, AL, GA, KTAG ,TAG, ArcMap ,ESRI, Spatial Analyst, cartography, map, analysis, heat map, Geographic Information Systems, GIS,  Southern Sandstone Ridges, Southern Dissected Ridges and Knobs, Southern Shale Valleys, Limestone Valleys and Coves, Cumberland Plateau, Plateau Escarpment, Southern Limestone/Dolomite Valleys and Low Rolling Hills, Buhrstone/Lime Hills, Coastal Plain Red Uplands, Dougherty Plain, Southeastern Floodplains and Low Terraces, Southern Hilly Gulf Coastal Plain, Southern Sedimentary Ridges, Southern Table Plateaus, Southern Pine Plains and Hills, Atlantic Southern Loam Plains, Carter Hills, Caseyville Hills, Crawford-Mammoth Cave Uplands, Cumberland Mountain Thrust Block, Dissected Appalachian Plateau, Dissected Plateau, Eastern Highland Rim, Fall Line Hills, Green River-Southern Wabash Lowlands, Hills of the Bluegrass, Inner Bluegrass, Inner Nashville Basin, Knobs-Lower Scioto Dissected Plateau, Knobs-Norman Upland, Little Mountain, Mitchell Plain, Northern Forested Plateau Escarpment, Northern Hilly Gulf Coastal Plain, Ohio/Kentucky Carboniferous Plateau, Outer Bluegrass, Outer Nashville Basin, Sequatchie Valley, Shale Hills, Southern Crystalline Ridges and Mountains, Southern Inner Piedmont, Southern Metasedimentary Mountains, Southern Outer Piedmont, Talladega Upland, Tallahasee Hills/Valdosta Limesink, Tifton Upland, Transition Hills, Western Highland Rim, Western Pennyroyal Karst Plain


Kentucky Cave Distribution Map, 2017

Kentucky Cave Distribution including Karst Geology, 2017 This map represents the generalized distribution of cave entrances in Kentucky based on data from the 2017 release of the Kentucky Speleological Society (the data is generalized by 7.5 minute quadrangle). A few easily observable patterns exist and are worth description.

The line of density which extends northeast / southwest shows the margin of the Cumberland Plateau where Mississippian aged limestones outcrop.

The ring of density west of there is the Inner Bluegrass Karst region where erosion of the Cincinnati Arch has exposed Middle Ordovician carbonates. The dense area centered at Edmonson, Hart, and Barren counties represents the Mammoth cave region and the caves of the Pennyroyal plateau. Mammoth Cave is still the world's longest cave, with a newly announced length of 412 miles.

The most cave entrance dense quadrangle recorded is Johnetta in Rockcastle County, but this is likely only sampling bias. This bias could occur because caves are easier to get to, or more likely because cavers are active in that area. Another possible example of sampling bias is in Carter County, which has been the focus of cavers and of much documentation. This may account for its high totals versus other limestone rich areas. To this end, this map serves as wonderful tool to generate the next areas of exploration, as we see several limestone areas that seem to have few caves.

Overall, Kentucky has 4,939 documented caves, and 5,547 documented cave entrances (some caves have more than one entrance). Data from the Kentucky Speleological Survey (KSS) comes from cavers and scientists, and has been collected and compiled since 2000.

Thanks Howard Kalnitz for the assist on this write up!


Tennessee Caves 2017

Tennessee Caves 2017
With the Fall data release of the Tennessee Cave Survey (TCS) comes the updated Tennessee cave distribution map. I've been producing these maps yearly since the TCS began digitally distributing their data in 2008. In that time my maps and my skill as a map maker have changed dramatically.

This year 153 new caves were added, and 42 caves removed. Caves get removed either because they have been re-measured since discovery and shown to fail to qualify for the survey (50' horizontal, 30' vertical), or the cave is connected to another known cave and two or more caves become one cave.

In other news White County is still way out in the lead with 1,250 unique caves. The majority of caves in Tennessee occur at a specific geologic contact, and that particular contact is well exposed in White County.


Public Lands of the Upper Cumberland

Welch Point, Bridgestone Firestone WMA, White County, Tennessee 13 Below you'll find a reasonably complete listing of state and federal public lands in the Upper Cumberland region. Some locations may be closed to the general public, so please refer to their website prior to visiting to ensure a safe and enjoyable trip.

US Army Corps of Engineers Reservoirs Website
Center Hill Lake Website Photos
Cordell Hull Lake Website Photos
Dale Hollow Lake Website Photos
Old Hickory Lake Website
National Park Service Lands Website
Big South Fork National River & Recreation Area Website Photos
Obed Wild and Scenic River Website Photos
State Forests Website
Bledsoe State Forest Website Photos
Pickett State Forest Website Photos
Scott State Forest Website
Standing Stone State Forest Website Photos
TDEC Managed Lands Website
Big Bone Cave Designated State Natural Area & National Natural Landmark Website Photos
Burgess Falls Designated State Natural Area Website Photos Map
Colditz Cove Designated State Natural Area Website Photos Map
Cordell Hull State Historic Park Website Photos
Cumberland Mountain State Park Website Photos
Cumberland Trail State Park Website Photos
Cummins Falls State Park Website Photos
Edgar Evins State Park Website Photos
Fall Creek Falls Designated State Natural Area Website Photos
Fall Creek Falls State Park Website Photos
Hubbard's Cave TNC Preserve Designated State Natural Area Website Photos
Lost Creek Designated State Natural Area Website Photos Map
Morrison Meadow Designated State Natural Area Website
Ozone Falls Designated State Natural Area Website Photos
Pickett State Park Website Photos Map
Pogue Creek Designated State Natural Area Website Photos
Rock Island State Park Website Photos Map
Sgt. Alvin C York State Historic Park Website Photos
Short Mountain Designated State Natural Area Website Photos
Standing Stone State Park Website Photos
Twin Arches Designated State Natural Area Website Photos
Virgin Falls Designated State Natural Area Website Photos
Washmorgan Hollow Designated State Natural Area Website Photos
Window Cliffs Designated State Natural Area Website Photos Map
Tennessee Valley Authority Reservoirs Website
Great Falls Lake Website Photos
TWRA Managed Lands
Alpine Mountain Website Photos
Anderson Pond Website
Big South Fork N.R.R.A. Website Photos
Blackburn Fork Website Photos
Bridgestone/Firestone WMA Website Photos
Catoosa WMA Website Photos
Centennial Wilderness Website Photos
Cordell Hull Refuge Website
Cordell Hull WMA Website
Edgar Evins SP & WMA Website
Fall Creek Falls SP & WMA Website Photos
Hampton Crossroads Website
Headwaters WMA Website
Jackson Swamp Website
Keyes-Harrison WMA Website
Luper Mountain Website
Mt. Roosevelt WMA Website Photos
Pea Ridge WMA Website
Pickett SF & WMA Website Photos
Scotts Pinnacle Website Photos
Skinner Mountain Website
Standing Stone SF & WMA Website Photos
The Boils Website Photos


Upper Cumberland Tourism Maps

Alexandria, Algood, Allardt, Auburntown, Baxter, Byrdstown, Carthage, Celina, Centertown, Cookeville, Crab Orchard, Crossville, Dowelltown, Doyle, Gainesboro, Gordonsville, Jamestown, Lafayette, Liberty, Livingston, McMinnville, Monterey, Morrison, Pleasant Hill, Red Boiling Springs, Smithville, South Carthage, Sparta, Spencer, Viola, Woodbury, Bowman, Clarkrange, Dodson Branch, Fairfield Glade, Grimsley, Lake Tansi, Lakewood Park, Cannon, Warren, Van Buren, DeKalb, White, Smith, Macon, Cumberland, Putnam, Jackson, Overton, Clay, Fentress, Pickett, water, stream, river, creek, kayak, canoe, put in, put-in, take out, take-out, access


Upper Cumberland - Chocolate Lover's Wine Trail

Ruth Dyal and the excellent folks at the Upper Cumberland Tourism Association have put together a wine and chocolate tasting for this weekend; February 10, 11th & 12th, 2017.

I happen to love both wine and sweets, so this seems like a great way to get out, explore, and enjoy local delicacies. Below you'll find an interactive Google map, links to an interactive ArcMap Online map, and a static map. Feel free to use or share any of these links to help us get the word out!

Event Web Page

Event Facebook Page

ArcMap Online Interactive Map

Google Map Interactive Map


Lidar Map of Sparta, Tennessee

I wanted to make a map of my hometown of Sparta, Tennessee using the newly available lidar from the State of Tennessee. Compared with previous maps, like the USGS 1:24,000 topographic map, the new lidar data is approximately 700 times more accurate. You'll see what I mean below.

After I produced it there were just too many odd little things that were worth explaining. The first half of odd things are natural processes on the Earth that produce a specific form, or in more scientific terms, the features are of a geomorphic nature. The remainder of forms I want to share with you are human formed, or anthropogenic. 

First, let's look at the map. Your web browser likely won't let you see it in full hi-res glory (the image is 4'x3' at 300 DPI, it's big at 77 megabytes), so here is a hard link to the hi-res image.
Sparta, White County, Tennessee
Below is the stuff I thought was interesting. Feel free to follow along if you've downloaded the hi-res map.

Monteagle Limestone Boulder Field
In the north-west of the map, here are two small green colored mountains that rise around 300’ above the surrounding landscape. Look closely at the hills and you will see that they are quite rough. If you’re looking at this in real life, you’ll see a bunch of grey rocks jutting up through a thin layer of soil. This landscape should be familiar to many locals, but it’s unlikely that most have considered why it is this way.

The Monteagle limestone is our most prolific producer of caves, with 34 percent of documented caves in Tennessee being formed entirely or at least partly within this strata (TCS 2016). That’s the same reason the landscape is boulder-strewn with little to no soil. The caves exist because the rock dissolves and leaves little insoluble material behind. What remains is a thin soil that doesn’t completely cover the bedrock. This is exposed in many places across White County. I have taken to calling those areas Monteagle boulder field. Also noteworthy, but not surprising, is that there are few structures on this type of landscape.

Monteagle Limestone Sinkholes
On the eastern border of the map, at the half-way point of north and south, one can see the foothills of Dibrell Mountain, where Highway 70 winds its way up to the Cumberland Plateau. In those foothills there are large sinkholes that look like pockmarks on this map. Contrast those with the St. Louis sinkholes mentioned below. Moneagle sinkholes are more circular, and in this particular location are around 60’ deep. Some of the larger sinkholes are not exactly circular, which is likely result of two or more sinkholes combining.

St. Louis Limestone Sinkholes and the Sinkhole Plain
East of the center of the map (and in many other places), a line of shallow sinkholes runs north to south down to the Calfkiller River. Comparing these sinkholes with those of the Monteagle limestone a few things should be apparent. They tend to be less circular, some have visible drainage channels within them, and many of them have distinct swallets (looks like a steeper sinkhole inside the sinkhole). They rarely occur isolated and are more often found in large irregularly connected networks. The networks are remnants of collapsed stream caves. It also may be evident that there are few smaller surface streams on this map. This is due largely to the St. Louis limestone’s high solubility forcing streams underground.

Karst Spring
If the streams are underground, then they must surface at some point. By following Town Spring to its source you can find this location on the map. Karst springs are places where the rivers emerge. Often what they look like is a low shelf of boulders with water coming from in between them. As a caver, I would love to be able to get into all the caves behind them, but few are accessible.

Row Crop Agriculture vs Pasture and Hay
You know where Walmart is (that’s not a question). Look west from there on the other side of County Highway 1161. There you’ll find a distorted rectangular property with an interesting pattern on the soil. Those are overlapping rows, possibly where the farmer dug rows at a different direction than a previous year. Several other examples of rows can be seen if one looks closely.

Thompson Weiman Quarry
Many Spartans are familiar with this large abandoned limestone quarry, but if you need help finding it on the map, look to where Highway 70 is on the east side of the map and go a little north.

The entrance to the quarry is quite obvious on the map. For those who have been there, they will recognize the towering cliffs. For those who haven’t been there, the quarry runs underground some distance.

Abandoned Railroad
Following the obvious linear railroad bed south-west from the Thompson Weiman Quarry, one can follow what was once the route of the railroad. The line traces Nent Flatt St, turns southward at Baker St, and almost disappears until the south side of Gaines St. The railroad would have started westward, crossing South Young St and the Calfkiller River (using the same bridge as the current pedestrian walkway), and heads north towards the White County Sheriff’s Department where it would have joined a spur rail to the north which is, surprise, on Depot St. The modern rail line begins at roughly the intersection of these two historic lines.

Waste Water Treatment Plant
Almost dead center of this map is where our sewer waste goes to be treated. Two large pools can be seen.

Low head dam on the Calfkiller River
Downstream from the waste water plant, just before where Calfkiller River goes under Highway 111, there is a low head dam. Many of us have seen this as we have driven past. Have you ever wondered wh it’s there? Read on.

T & A Hydroelectric Plant
Following the north shore downstream of the previously mentioned low head dam, there is a linear feature which runs parallel to the river. This is a causeway along which water from the river was once diverted to run a powerhouse. The causeway, and much of the poured cement structure are still in place, as well as the powerhouse itself.

Clarks Mill
On Town Creek, just north of where the railroad crosses overhead, I found curious evidence of a large dam that failed. Until the creation of this map and the integration of the GNIS dataset, I had not been able to identify the structure. Looking at this map, one can now only see two lines perpendicular to the stream which represent parts of the dam structure of the mill.

Drive in Theater
Just outside the city limits of Sparta, on the north side and east of Highway 111, one can see moved earth in a radial pattern. If you’ve enjoyed watching movies at the Sparta Drive-In, then you’ll immediately recognize that those are the low rises for each successive row for cars to park in.


Making Cartographically Accurate Rivers using High Resolution Elevation Data in ArcMap

This begins as a standard stream delineation process beginning with a Digital Elevation Model (DEM). I'll work it step by step and show you what I've got so you can work alongside me. I'll be using a DEM from my part of the world, near Sparta, Tennessee. You are free to use a DEM from wherever you like. Generally, the same process applies.

I'm using the new lidar derrived 2.5' DEMs and I'm looking at a segment of the Calfkiller River for this example. I have docked the toolbar menu and expanded Spatial Analyst > Hydrology. We'll be using a bunch of commands from there. I also have raster calculator added to my menu buttons on top, that will get some use as well. All the tools we use will be on their default settings unless otherwise stated. I follow the general naming conventions provided by the tools. You may find it useful to duplicate my naming strategy until you get a feel for it.

One final note about this technique is that so far I have only used it to work on small areas. It may not be ready to by employed large scale. Also, I developed this only as a cartographic technique. While it tends to match up with water in hi resolution aerial images I'm looking at, I would certainly hesitate before applying this to more meaningful work like flood modeling.

The first thing to do is the Fill function on your DEM to smooth it out and let water flow across it in a way that makes sense to a computer (by the way, this really doesn't work well in a karst basin like the Calfkiller, as you'll see momentarily).

Once that is done, you'll run the command flow direction on the filled DEM. The flow direction output should only have numbers which are powers of 2. If you numbers which aren't, you've run flow direction on your original DEM, not the filled DEM.

When you have the flow direction, you're ready to create run flow accumulation. This will begin to develop your stream networks. If this tool takes longer than the others to run, don't worry, that is normal. When it is done, I often go into the layer's properties under the symbology tab and change the stretch type to equalize. This will let you better see the stream networks.

A few more steps and you'll be able to pull out and vectorize your stream networks. You have to make a decision about where streams begin in terms of value along your particular flow accumulation raster. There are lots of ways of going about determining this. Some of those methods are more empirical and academic than others. My method for cartography is simply eyeballing it. I use the identify tool to get the value of a pixel at a particular location, and I copy that value to the next step in Raster Calculator. What that pixel will represent is where the stream network will begin. No streams will appear upstream of it, in other words. In the picture below you can see that the expression I use is Con("FlowAcc" > 1788574,1). The odd value is where I used the identify tool to get the value of the Flow Accumulation raster at the point where I wanted the stream to begin.

Now that you've got a raster stream layer, you should convert it to a vector stream layer. Use the Stream to Feature tool as shown below.

Your streams are vectors now. In my part of the world there are lots of rivers that flow underground because of karst. I'm going to do a reality check and remove those.

Using the editor tool I've deleted all the line segments that I know aren't flowing on the surface. There, that looks better.

The next part of this workflow will require that we buffer the line slightly. We want it to be a polygon so that we can later union it with the next batch of data we produce. I'm going to buffer the line by ten feet, which is again, a completely arbitrary number. After you've worked through this once with your data, you can see how it looks and adjust that number accordingly.

Go ahead and remove Stream03, Stream02, Stream01, FlowAcc, and FlowDir from the data frame, we will no longer be needing those.

Run the Slope tool on the Fill layer, and then we want to get at the areas of the slope_fill that are less than 2%. This will help us identify where water could be pooling on the landscape. Again, my area is karst heavy, and the results of these steps will identify lots of false positives, while yours may not. See the next few pictures on how to do these steps.

Time for another reality check. Those sinkhole aren't actually lakes. I am only interested in the river channel. So we'll need to turn this into a vector layer so we can grab only what we need. The next few images will show how I do that.

Get rid of everything but Streams06 and Elev. You'll want to do a hillshade to see how the streams look against a more realistic landscape.

I think that looks pretty good. I hope that you found it helpful. If you have any tweaks, secrets, comments, or questions I'm happy to talk with you.