Garden Waterfeatures


Long requested by Sandy and long resisted by Dale. Here is the stream, waterfall, and bridge.

Although this page presents construction by blocks of tasks, the edges of the blocks overlap considerably. Sometimes the blocks were performed simultaneously. For instance, the walkway to the bridge was in progress while the pond-tank hole was being dug. The electrical lines went in after the pond tank was put in place, but the walkway was still not finished. So, photos showing one block may show another block in various states of completion.


Just some baseline pictures of the area before any digging started.

Drainage ditch picture
Drainage ditch picture

Back to Top

Month One - The Arched Bridge

The arched bridge was built and installed, then taken apart and rebuilt and re-installed. The first version looked good on paper, and looked good after it was built. However. it was just too big for the yard. Rebuilding it involved moving the buried foundation in addition to the re-cutting of structural members. The rebuild was more work than building it originally. Sometimes you just can't tell what will look right until you see it.

What a memory! I forgot to take pictures of the bridge construction. All I have are the drawings and the photos that were taken when I began putting in the stone walkway to the bridge.

Bridge drawing

Bridge photo Bridge photo Bridge photo
Back to Top

Month Two - Pondless Pond

To make the pondless waterfall stream flow continuously, a 100-gallon Rubbermaid stock tank was buried where the stream would normally enter into a pond. A 3200 gph submersible pump was plumbed in the tank to move the water from the holding tank back up to the waterfall spillway. The size of the holding tank was calculated using an average stream width of 24 inches, an average depth of 3 inches, and a length of 15 feet.
Stream = 24" x 3" x 180" = 12960 cu. in.
12960 cu. in. / 1726 cu. in./cu. ft. = 7.5 cu. ft.
7.5 cu. ft. x 7.48 gal./cu. ft. = 56.25 gal.
With 56 gallons of water in motion in the stream, the maximum in an empty 100-gallon tank would be 34 gallons. The water level in the tank will be several inches below the tank lip, there is some space taken up by the concrete blocks, and evaporation will slowly drop the water level in the tank between fillings. Thirty-four gallons will likely be reduced to less than 17 gallons, and this will be needed to keep the pump under water.

Hole photo Tank in hole photo Tank in hole photo Tank in hole photo
Eight concrete blocks, and 8 ceramic tiles left over from a previous project, cheaply provide a pedestal for a covering grate without wasting much of the tank volume. The cover grate was constructed from two plastic pallets. The bottom one was selected because of its straight edges, dense matrix for high strength after cutting, and its low cost. The top pallet was selected for high strength and a solid surface at the lowest price. The solid surface allowed me to drill 1/2-inch-diameter holes that were large enough for handling the pumped water flow, while being small enough to prevent rocks from entering the tank. The grate has to hold a layer of rock plus the weight of a person that might step on it. My design minimum load was 300 pounds. The static load capacity of either pallet is over 1200 pounds. Cutting the pallets and removing the original supports may decrease the load rating of the pallets, but with the edge support and concrete block support, 300 pounds carried on the two pallets should be well under their capacity.
2 pallets photo
Plastic pallets before cutting

Cut pallet photo
Cardboard template

Cut pallet photo
Bottom pallet cut

4 blocks photo
Top pallet marked for cut

4 blocks photo
Concrete block pedestal

1/2 pallet photo
One-half bottom pallet in place

2/2 pallet photo
Completed bottom pallet

top pallet photo
Top pallet in place

The bottom pallet was cut length-wise to get two long sections. The edges of this pallet are strong and after a little grinding provide a good suface to set on the ledge of the holding tank. A gauge block was made that mirrored the pallet-to-tank contact point. I used a drill with a rotary rasp to grind the pallet edges to fit the gauge blockand therefore fit the tank edge.

Gauge block photo
Gauge block

Gauge block in use photo
Gauge block in use

I agonized over whether to cut a hole in the grates for removing the pump without removing all the rock supported by the grate. At this point, I have chosen not to cut the hole. The hole would greatly weaken the grates. Hopefully, I won't have to remove the stone more than once per year. The rock probably should be removed and cleaned of debris once per year, I reasoned. The flat surface of the top pallet should make the job easier. I can cut the hole later if I so choose.
Back to Top

Month Three - Electrical

Power for the submerged water pump was 150 feet from the pump. To get from the circuit breaker to the pump, and add an outlet to power one of Sandy's other projects, a fountain, a 180-foot long trench was needed. A GFCI (Ground Fault Circuit Interrupter) protected circuit was needed if the trench was to be 13-inches deep. If not, the trench would have to be 19-inches deep. Keep in mind that the code requires 12-inches or 18-inches of fill ABOVE the wire. I have wanted a GFCI outlet between my garage doors for years to plug in a sweeper for cleaning cars. So, I added an outlet with a GFCI receptacle and tapped off the load side of the receptacle to run the line out to the pump. One hundred eighty feet of wire in the trench, plus 30 feet of wire in the garage to get to the receptacle made a run of 210 feet. This is a VERY long run for a GFCI circuit (about 30 feet over the recommended MAXIMUM lenth), but so far it has not had any nuisance trips.

Treanching is not an easy task. After Miss Utility did her job and located the existing wires, hand digging was required to find them or avoid them. A rented trencher for a 19-inch deep trench was $180 per day, plus a trailer rental and a a hitch attachment. That trencher weighs 1270 pounds, is a bear to move around, and doesn't fit in small spaces in a garden. This is another reason why a GFCI-fed circuit is desireable; with a GFCI-fed circuit, a small trencher can be used.

The small trencher does a 14-inch deep trench, 3-inches wide. It weighs 270 pounds, can be put in a van by two men, and is somewhat manuverable between plants in a garden. It rents for $120 per day and there is no trailer to rent.
Trenching took one full day of HARD work. It took another full day to put in the UF wire where the trench was clear of obstructions, and another full day to chop out obstructions, such as roots the trencher could not get through, and hand dig areas like trench right turns. Yet another day was needed to hand dig under roots of a tree that could not be cut without hurting the tree, and chiseling through an old stump that had been ground below the ground surface but not 13-inches deep. Finally, it took 2 days to dig post holes, mount conduit and boxes, connect wiring to the receptacles, and fill and tamp the dirt in th trenches.

Can you tell where the trenches are? In a year, I won't be able to tell either, that is why there are so many pictures here.
 Garage to fountain photo  Fountain from garage photo  Fountain photo Fountain to walkway photo Walkway from fountain photo
 Walkway right angle to tent branch photo  Tent branch from walkway right angle photo  Tent branch photo Tent branch to pump branch photo Under stream to pump photo
Back to Top

Month Four - Stream Digout

I dug the hole for the holding tank about 1-inch to 1-1/2-inch larger than the tank around all sides, and about 1-inch to 1-1/2-inch deeper than the final desired height.

Making the floor of the hole as level as possible was done to ensure it stayed level as the ground settled. Next I poured an inch of sand on the floor, placed the tank in the hole, and wiggled it. The bottom of the tank is ribbed, and wiggling it works the ribs down in the sand. I removed and replaced the tank several times, each time adding, subtracting, or moving sand to get the height of the tank just right.

When the tank was at the right height and was level, I placed two concrete blocks in the tank and then added some water to the sand. The water together with some gentle tank wiggling let the tank settle down into the sand and not be resting on only the ribs. Overnight the water soaked into the soil and the tank remained at the correct height and was still level.

After the bottom was settled, I filled the tank with water and poured sand around the sides to a level about two inches below the desired final ground level. Again, I used water to settle the sand. A layer of top soil on top of the sand was added to allow ground water to flow around the tank without too much of it getting into the sand, The sand absorbs any moisture that seeps through the top soil and lets it slowly soak into the soil.

Hole photo Tank in hole photo Tank in hole photo Tank in hole photo

The plan for the stream bed was to create a pond-stone-filled drainage channel beside the stream to carry away ground water. The drainage channel would go around the outside edge of the holding tank and empty into the real rainwater drainage ditch.

This was an impractical plan. It was difficult to sculpt the bottom, made putting the carpet in the trench very hard, used up valuable water depth, and didn't work. I had to rework the bottom when I started building the stream in Month Six.
Stream layout drawing

Project Delayed

Month five is behind schedule. Other priorities, like digging a trench and putting in a drain for a downspout, some work related travel, and family commitments have come into play. Cold weather and shorter daylight hours are making things difficult. It is beginning to look like this project will not get completed until next spring.

Back to Top

Month Five - Plumbing

The plan for plumbing was to put a removeable joint at the tank grates, another one about 4 feet from the tank, and another one at the waterfall spillway. Gradual turns of the pipe would be done by heating and bending the PVC pipe, and sharp turns would be done using long Tees.

Then reality hit and the plan was history. Of couse the pump outlet was the wrong size and in the wrong position, so the first chore was to make an adapter to get the connection from the pump to a 1-1/4" threaded fitting where the backflow valve would go.The turns needed to get from the pump to the the removeable joint were too tight for bending pipe. And, during the fitting process, a flimsy adjustable joint on the pump broke. Thanks to Lowes large selection of DWV fittings and a large can of PVC cement, the pump got connected to the backflow valve.

The backflow valve that threads into the fitting at the tank grate has a 2-inch diameter housing at the outlet end, which is just right for attaching one end of a 2"-to-1-1/2" rubber coupling. A rubber coupling here lets me disassemble the pump plumbing to remove the pump. Another rubber coupling in the pipe 4-feet from the pump lets me rotate the pipe so that I can pump out the water in the tank if I need to.

One last rubber coupling at the waterfall was planned to attach the spillway to the pipe from the pump. I left it out. It may be added if I need to remove the spillway someday. The spillway is a 9-inch wide spreader that I made by casting concrete around a vinyl form. The water from the pipe is focused into a 9"-by-3/16" flow that goes under a piece of flagsone on top of a boulder. [outlet area = 9 X 3/16 = 1.7 = approx. 3.14 X (1-1/2 / 2) SQ. = pipe area]

Exit pipe detail drawing Tank exit pipes drawing

Pass-thru coupler photo Pump plumbing photo System plumbing photo

Here is how the spillway was made.
Spillway armature photo
Sheet-vinyl housing glued
together and pop riveted
to down pipe.

Spillway armature photo
Curve in housing at top
distributes water laterally.

Spillway armature photo
3/16" x 9" gap equals area
of 1-1/2" diameter pipe.

Spillway concrete photo
Saran covered boulder with
concrete covered spillway
housing in place
passes inspection.

Spillway minus flagstone photo
Concrete covered spillway
housing in place
with flagstone removed.

Spillway with water flowing photo
Spillway in operation
with water flowing.

Back to Top

Month Six - The Stream

Month six has drug out way beyond what was expected so I have broken it into five sub sections.

First Try

Now that the basic elements were in place, it was time to start building the actual stream.

I had all winter to visualize what path the stream would take, so as soon as weather permitted, I began digging an oversized ditch. The ditch was about 6-inches deep to accomodate the average 2-deep stream. Why so deep? Underneath the liner is a layer of sand, a layer of old carpet, and a 2-inch layer of 3/4-inch river stone. The stone is a mole barrier, the carpet is a root barrier, and the sand is a surface smoothing medium.

In my mind this seemed like a formitable task. In actuallity it was worse than I imagined. Estimating how deep to dig to allow for the gravel and underlayment carpet was difficult. Visualizing where rocks would go and what their effect would be, then planning for keeping ground water out of the stream, made this more of a hope-and-pray exercise than a build-to-drawing operation. The first test was encouraging, unless you think water coming from under my plastic-sheet test liner wasn't planned. Sealing the spillway boulder to keep water from following the boulder wall underneith the boulder and then underneath the liner was not a task I comfortably felt would be a permanent fix. Sore by the end of the day, I was wondering how much it would cost to rent a bulldozer to level the whole thing.

Stream mole guard gravel photo
2" gravel mole guard.

Stream test plastic sheeting photo
Plastic sheeting flow test.

Stream underlayment photo
Carpet underlayment over gravel.

Stream liner photo
Mounds formed under sides.

Stream gravel photo
Gravel in center of liner.

Stream  photo
Liner tucked over mounds.

Stream  photo
Water test.

1st bag photo
Sandbag falls.

Waterfall Leak

Now, before I could start adjusting the water flow to get a gurgling stream, I needed to make the water come off of the big rock correctly. The curve of the rock caused the water to flow back under the rock and under the liner. As I saw it, the only solution was to glue on a deflector strip of liner to catch the mis-directed water and redirect it into the stream. Since I would be attaching liner to the rock, I decided to attach another piece to the side of the rock so I could allow some water to flow gently over the side for astetics. This would increase the evaporation, but I thought it looked nice.

I looked at epoxy, polysulfide, polyether, and polyurethane adhesives to bond the liner to the rock. All the adhesives would stick the liner in place (I thought), but only a marine polyurethane from 3M was rated for water immersion. The others are waterproof, but not rated for immersion. I use PL Premium polyurethane for many projects and it is great stuff at a good price, but it clearly states that it is not intended for water immersion. Some of the other brands are not so forthcoming in declaring their lack of immersion survival.

Clamping the glued liner to the wire brushed, water scrubbed rock was a challenge (modern euphonism for what I always called a problem). I decided to use sand bags. It doesn't take a lot of force as the liner is not stretched, but the shape is irregular and the sandbags conform to the rocks shape. After a dry run, I found the sand bags were not enough. The rubber liner wanted to pull away from the rock at the very edge of the liner. This I solved by using every man's favorite tool, duck tape. It stuck to the rock and the rubber long enough for the adhesive to cure, and was strong enough to hold the edge of the liner flush with the rock.

I made a discovery while waiting the 7 days for the adhesive to cure; 3M5200 doesn't stick well to EPDM rubber. This illustrates the danger of Internet research, you get a lot of mis-information. The bond strength of 3M5200 to EPDM is 10 psi compared to 44 for vinyl, 101 for steel, and 243 for fiberglass. Since the added piece of EPDM is a flap to get the water redirected into the stream, and not part if the stream liner, the solution might have been to use a different rubber, maybe neoprene or nitrile, instead of EPDM for the flap. But, as with the EPDM there is no information available on how well the 3M5200 sticks to the other rubbers. If, or when , the current flap starts to tear away I may replace it with another rubber type, or I may try putting many perforations at the edge of the flap so the 3M5200 can form a mechanical bond to bolster the chemical bond. Another possibility would be to replace it with a fabric-backed, or fiberglass-backed, vinyl-faced material; the adhesive would stick to the fabric and the vinyl would repell the water. Yet another possibility has presented itself. I have found a silyl terminated polyether hybrid adhesive, Chem Link M-1, that is immersible and specifically lists EPDM and stone as two of the substrates it bonds to. Due to my mistrust of Internet research, I will have to check it out further.

 Flap taped photo
Liner glued and taped to rock.

Flap sandbagged photo
Glued, taped, and sandbagged.

 Flap w/o sandbags photo
Cured liner flap.

Rock covered flap photo
Camoflaged and covered .

Pump Problems and Water Loss

After a weekend in New York, I returned to find a dead pump. The motor was running but it was not pumping water. The impeller had wobbled off the shaft, but even after it was replaced, the pump had low pressure and went into thermal shutdown after a very short time.

So what killed it? Was it that the water level had dropped below the pump inlet because the pump was not on the bottom of the tank. Yes. Was the water loss excessive, which dropped the level in the tank to less than one half? Yes. Shouldn't there have been protection designed into the plumbing? Yes. Shouldn't the pump been able to survive the water loss? I'd like it, but none are able to in this flow class.

Solution: buy new pump, lower pump, correct water loss, add automatic water fill, add low-water pump shutdown.

Lowering the new pump was easy. By substituting a modified 12-inch block for one 8-inch block and one 4-inch block in the pallet support pedestal, I made space on the bottom of the tank for the pump. Using a carbide tipped drill bit , I cut the side and center out of the 12-inch block to give me a U-shapped block that goes over the pump. Re-plumbing the pipe to get the pump connected to the backflow valve at exactly the same point where it was previously connected turned out to be easy, as the pump output port was nearly lined up after the move. I didn't put the cost of this change into the cost breakdown because if I had done it right the first time it would have been zero cost.

I found a sump pump piggyback float switch on E-Bay for $27.50 including shipping. This provided me a quick means of protection to stop pump damage when the water drops too low. If I ever get this project finished enough to consider putting in an automatic filling system, I will look at integrating the low-water-shutdown and automatic-fill functions into a single water level sensor assembly.

The water loss is still being investigated. There is a tremendous amount of splash at the water fall, and this could be the source. I don't think any water is leaking behind the skirt on the rock, but I will recheck that area. It looks like I will be removing some gravel and doing some repositioning of the liner.

New block and tank drawing Modified block
Pump in modified block

Second Try

In the first trial I used the sand filled socks (sandbags) from the flap glueing to act as temporary rocks. This let me make easy changes to the stream bed and see how the water would move before I wrestled a heavy rock in place. What I discovered was the drainage channels didn't work, I didn't have enough terracing behind my rapids, and I lost too much water from waterfall splash.

All the stream rock was removed, and I started over.

To keep out groundwater run-off from the dirt mound next to the stream, a drainage channel was to be dug beside the stream. The drainage channel would be about 6-inches wide and 6-inches deep and be filled with river gravel. This was a impractical plan. Just a lot of work preparing the shape of the bottom surface and a lot of work trying to get the carpet to make sharp compound curves. After trying this and having to dig it out, I can say positively it is just better to dig the trench with vertical walls and use the gravel to smooth the edges. Lay the carpet up the sides as high as the waterline then bring the liner over the edge of the carpet.

The steps were kind of like this:
(1) put in the mole guard layer,
(2) put in the carpet and the liner and put gravel over them in the center of the stream,
(3) lift up the edges of the carpet and liner and stuff gravel between the vertical trench walls and the carpet,
(4) lay the carpet and liner over the added gravel to form the sides,
(5) tuck the ends of the liner down over the carpet and cover with more gravel.

Liner install diagram

The water in the stream moved way too fast and pushed my river pebbles down stream. I fixed this by increasing the terracing of the bed behind my rapids falls until they were level. and in some cases reverse sloping. This created a small puddle behind the rapids fall and slowed the water movement.

Splash from the waterfall sent drops 3- to 4-feet beyond the falls. I couldn't tame the falls, so I had to make the liner catch the water and return it. I didn't want the stream to be wider at that point, so I layed the liner above the waterline out as far as the falls water splashed and covered it with gravel and rocks. The water that splashes out drips through the gravel and eventually drains down into the stream.

 Reconstruction dig out photo
Re-dig for terraces.

Main falls pool photo
Anti-splash pool.

Stream steps photo
Slower flow.

Stream rocks photo
Muddy water falls.

Rock Arrangement

The final stage looks like a never ending saga - arranging the rocks and gravel. Several large rocks were found on the many drives sightseeng through Pennsylvania, and these form the major chunks of the stream. Other rocks are being added to get the desired flow. As I did before, I am using the sand filled socks (sandbags) from the flap glueing to act as temporary rocks and see how the water will move. A sandbag under a rock, or rocks, can also give the rocks a solid base to hold them in place, distribute their load over a larger area to protect the liner, and form a seal between the rocks and liner to keep water from leaking under the rocks.

Near the edge of the stream I couldn't bring the gravel up to the dirt without the black liner showing through. My solution was to use a cement stabilized soil over the liner there. The stabilized soil holds its position, doesn't dissolve easily in the stream water, and looks like a dirt stream bank. The stabilized soil is made by mixing (by volume) 1 part type-N masonary cement to 3 parts dirt. It is better to use dirt with a minimum of vegatative matter in it, but is not critical. I mixed in just enough water to make the thick mud spreadable over the liner, then patted some un-stabilized garden soil on top of the mud to make it look more realistic.

 Rock arrangement  photo
Stabilized dirt bank.

 Rock arrangement  photo
Other side bank.

 Rock arrangement  photo
Bottom half done.

 Rock arrangement  photo
Top half at falls.

 Rock arrangement  photo
Top half done.

 Rock arrangement  photo
Rock placement done..

Before and After

Pictures for comparison to the baseline pictures of the area before any digging started.

Drainage ditch picture
Drainage ditch picture

Drainage ditch after picture 1
Drainage ditch after picture 2

Drainage ditch picture
Drainage ditch picture

Drainage ditch picture
Drainage ditch picture

Back to Top


What does a little project like this cost, excluding labor.


Stream Tank 1 69.99 69.99 TSC
8x8x16 block 7 1.31 9.17 Lowes
4x8x16 block 3 1.06 3.18 Lowes
12x8x16 block 1 2.40 2.40 Home Depot
CPP110ACM pallet 1 10.60 10.60 Nelson
CPP340 pallet 1 16.50 16.50 Nelson
Sand 4 2.50 10.00 Lowes
Sand 4 2.96 11.84 Lowes
Very Big Rocks 4 994.00 994.00 Vinci Stone
Small stones 5 3.26 16.30 Home Depot
River pebbles 34 3.38 114.92 Home Depot
Gravel, drainage 1 2.96 2.96 Ace
Type-N masonary cement 1 8.88 8.88 Home Depot
Pulverized lime 1 3.88 3.88 Lowes
EDPM liner, 6'x20',45mil 1 78.38 78.38 Harrison Pond
4-mil plastic test liner 1 8.28 8.28 Home Depot
Underlayment 1 0.00 0.00 Carpet
3M 5200 Polyurethane Adhesive 1 10.94 10.94 Home depot
3200 GPH Pump 1 119.99 119.99 Harbor Freight
Float Switch 1 27.50 27.50 E-Bay
1-1/4" / 1-1/4" FSL-NPT coupler 1 .58 .58 Lowes
1-1/4" / 1-1/2" FSL-MSL adapter 1 .99 .99 Lowes
1-1/4 NPT/1-1/2 SLP backflow valve 1 9.98 9.98 Ace
1-1/2" / 1-1/2" FSL coupler 3 .85 2.55 Lowes
1-1/2" / 1-1/2" rubber coupler 1 4.37 4.37 Lowes
1-1/2" / 2" rubber coupler 1 3.91 3.91 Home Depot
1-1/2" / 1-1/2" FNPT-MSL coupler 1 1.03 1.03 Lowes
1-1/2" DWV long elbow 2 1.86 3.72 Home Depot
1-1/2" DWV short elbow 1 .61 .61 Home Depot
1-1/2" DWV short elbow 3 .66 1.98 Lowes
1-1/2" DWV 45 elbow 1 .65 .65 Lowes
1-1/2" DWV 60 elbow 1 1.67 1.67 Lowes
1-1/2" DWV M-F elbow 1 1.73 1.73 Lowes
1-1/2 DWV PVC pipe 2 3.34 6.68 Home Depot
Subtotal   1566.50    
Electrical GCFI receptacle 1 7.97 7.97 Harbor Freight
Weatherproof cover 1 3.37 3.37 Home Depot
Siding mini-J-cover 1 6.87 6.87 Home Depot
PVC electrical box 1 .47 .47 Home Depot
250 uf 12-2 wire 1 84.97 84.97 Lowes
Outside PVC electrical box 2 4.99 9.98 Lowes
Outside PVC electrical box 1 4.30 4.30 Lowes
Outside PVC elect box cover 1 1.97 1.97 Lowes
Outside in-use outlet cover 2 11.97 23.94 Lowes
PVC elect conduit elbow 1 2.19 2.19 Lowes
PVC conduit 3 1.37 4.11 Lowes
Timer 1 12.97 12.97 Home Depot
4" x 4" x 4 press treat post 3 2.49 / 8 4.98 Lowes
Trencher rental 1 127.60 127.60 Home Depot
Subtotal   295.69    
2x12x8' ACQ joist 6 10.97 65.82 Lowes
2x4x8' ACQ stud 20 2.82 56.40 Lowes
3/8-16 carriage bolts, washers, nuts 24 .99 23.76 Lowes
Maze spiral hardboard siding nails 2 5.47 10.94 Lowes
8x8x16 block 2 1.31 2.62 Lowes
Gravel, drainage 4 2.96 11.84 Ace
Flagstone 1 122.75 122.75 Vinci Stone
River pebbles 12 3.38 40.18 Home Depot
Metal edge, 8 2 9.98 19.96 Lowes
Metal edge, 4 1 5.98 5.98 Lowes
Metal edge stakes 8 1.07 8.56 Lowes
Subtotal   368.81    
Total Stream + Electrical + Bridge/Walk/Circle   2231.00    

Pump is a Harbor Freight / Pacific Hydrostar 95965-4VGA,
3200 GPH@ minimum head / 18 ft. head @ minimum flow,
same as other brands, but they hide the flow and head data.
Output coupling is flimsy, but there are 1-1/2" pipe thread
input and output ports on the pump housing.
Back to Top

Copyright Dale Thompson,
September 16, 2009 through
last revision on September 25, 2010.