Water pressure pushes the crack open. Debris can temporarily seal it. Groundwater outside the pool can push back. Three separate forces -- any one of which can make the same crack behave differently on different days.
Learn the H.U.N.T.E.R. Method →A gunite crack may only leak when the pool is completely full for three reasons. First, hydrostatic head pressure -- a full pool creates more outward force on the crack than a lower pool, which can push water through a crack that barely leaks at lower levels. Second, debris stuck in the crack can temporarily slow or stop the leak -- a crack that tested negative for dye may still be the source if debris is plugging it. Third, hydrostatic groundwater pressure from outside the pool can push back inward, closing or reducing a crack's behavior depending on the groundwater level relative to pool water level. All three forces must be considered when a crack appears inconsistent.
Water pressure at any point in a pool depends on the depth of water above it. A crack near the bottom of the pool has more water pressure pushing outward than a crack near the waterline. When the pool is completely full, every crack in the shell is under its maximum outward pressure for that pool's depth.
A crack that is marginal -- fine, partially sealed by old plaster, or bridged by debris -- may have just enough resistance to slow or stop leakage when the pool is partially full. When the pool reaches capacity, the increased hydrostatic head pressure overcomes that resistance and forces water through.
This is also why some cracks located near the top of the pool wall only leak when the water level is at its highest -- the crack is barely submerged at full level and has almost no outward pressure when the pool is even slightly below full.
Fine debris -- algae, calcium scale, silt, plaster particles, organic material -- can accumulate in a crack and temporarily slow or stop water movement through it. The crack may be a real, active leak path, but at the moment of testing the debris is acting as a partial plug.
A crack that tests negative for dye may still be the leak source. Debris stuck in the crack at the time of testing can temporarily block the flow, making the dye test appear negative. The same crack may flow freely once the debris shifts, dissolves, or gets pushed through by increased pressure when the pool fills up.
When a crack matches the water-level behavior -- pool loses water when full, stops when it drops to that crack's elevation -- that crack is still the primary suspect even if it passed a dye test. Test again with the pool at full level and under still-water conditions before ruling it out.
The inverse is also true: increased head pressure when the pool is full may push debris through the crack, opening the leak path further. The crack that barely leaked at lower levels can become the primary leak source once fully submerged under pressure.
Gunite pools are surrounded by soil that may contain groundwater. When the groundwater level outside the pool is high -- after heavy rain, in areas with a high water table, or in poorly drained soil -- groundwater creates hydrostatic pressure pushing inward against the pool shell.
That inward pressure from outside the pool can partially or fully counteract the outward pressure from pool water inside. A crack that would normally leak when the pool is full may show reduced or no leakage when groundwater pressure is elevated, because the net pressure across the crack is reduced or reversed.
This creates confusing and inconsistent crack behavior -- the pool seems to hold water better after rain than during a dry period, or the leak appears worse in summer when the water table is lower. The crack itself has not changed. The balance of forces on either side of it has changed.
This also means that draining a pool in high-groundwater conditions risks the shell floating or cracking under the inward pressure from outside -- a separate but related risk the technician should be aware of before recommending a drain.
When a pool loses water only when completely full and the loss stops at a specific elevation, that elevation is the target inspection zone. The leak source -- whether a crack, fitting, joint, or skimmer throat -- is at or near the water level where the loss stops.
If an autofill is present, it continuously replaces water as the pool loses it. The pool stays full. The owner sees no level drop. The only clues are an elevated water bill, a constantly running autofill valve, or a wet area outside the pool.
For any full-level leak investigation, the autofill must be shut off or bypassed during testing. Otherwise the water level never drops and the level-dependent behavior that identifies the source zone is never visible.
Turn off or bypass the autofill so water level can drop naturally. Without this step, the level-dependent behavior is masked and the test zone cannot be identified.
Mark the starting water level. Watch where the pool stops dropping. The elevation where loss stops identifies the crack, joint, or fitting that is the leak source zone. Note the elevation carefully before inspecting.
Dye testing a crack that is dry or only slightly submerged will miss the leak. Refill the pool to capacity and test while the crack is under full hydrostatic head pressure -- the same condition under which it leaks. If debris is suspected of temporarily sealing the crack, increasing the head pressure by filling to full may push the debris through and reveal the active leak path.
Wind, returns, water features, and surface movement create false dye readings at the waterline. Turn off the pump and wait for the water surface to be completely calm before applying dye near cracks, tile lines, joints, or fittings at the waterline elevation.
If a crack tests inconsistently -- leaks on dry days, seems to seal after rain -- consider groundwater pressure as the variable. A crack that behaves better after rainfall may be closing under inward groundwater pressure. Test during a period of low groundwater (after dry weather) to see the crack's behavior under maximum outward differential pressure.
A visible crack is a suspect -- not a confirmed leak. Dye must be observed moving into or through the crack before repair is recommended. A crack that does not pull dye under full-level still-water conditions is cosmetic until proven otherwise.
Not every visible gunite or plaster crack is actively leaking. Gunite pools develop surface crazing, plaster cracks, and shrinkage lines that are entirely cosmetic and do not affect water level. A crack must show active water movement -- confirmed dye pull, water-level behavior that matches the crack elevation, or visible seepage -- before repair is warranted.
Recommending repair on every visible crack is not diagnostic work. It is guessing. The test confirms the crack is active. The behavior pattern confirms the crack matches the reported loss. Both pieces of evidence together support a repair recommendation.
Ask whether the pool only loses water when completely full, whether the autofill runs constantly, and whether loss seems worse in dry periods than after rain.
Know the three forces: inside hydrostatic pressure, debris masking, and outside groundwater pressure. Any one can change crack behavior independently.
Disable the autofill and track where the water level stops dropping. That elevation is the target zone. Do not test below it -- the crack may not even be submerged.
Test at full level under still-water conditions. Apply dye near the suspect crack, joint, or fitting and watch for active movement. If debris is suspected, increased head pressure at full level may open the crack further.
Confirm dye movement before calling the crack active. A visible crack is a suspect only. A crack that pulls dye under full-level still-water conditions is confirmed.
Repair the confirmed active source. Retest at full level after repair. Document whether groundwater conditions could affect future crack behavior.
Sees a crack, applies dye at a low water level, dye does not pull (debris is plugging it), calls the crack cosmetic. Pool keeps losing water. The crack was active all along -- debris had sealed it at the time of testing and the pool was not full enough to stress it.
Tracks where water level stops dropping. Fills pool to full. Tests under still-water conditions with maximum head pressure. Notes that a negative dye result with debris in the crack is not conclusive. Considers groundwater conditions when loss behavior is inconsistent between dry and wet weather periods.
Jeff David is the founder of Leak Business Academy, LLC and Leak and Subsurface Locators, LLC. He is a licensed pool contractor in Florida and an FSPA Palm Beach board member. Jeff teaches the H.U.N.T.E.R. Method -- including the three forces that control gunite crack behavior (hydrostatic head from inside, debris masking, and groundwater back-pressure from outside), why a negative dye test does not rule out a debris-plugged crack, and why the pool must be tested at full level under still-water conditions for any level-dependent leak investigation.
LBA training covers the complete gunite crack diagnostic sequence -- water level tracking, autofill management, full-level dye testing, debris considerations, and groundwater effects -- so every shell investigation is thorough and defensible.
View Training Resources Get the Pool Leak Detection Guide