TalonLab / Carp
CP-01● Research active05 SEP 2026
CP-01 // Species-first decision research

Carp

Turn common-carp biology, lake physics, venue geometry, sonar and field outcomes into decisions that can be tested — and disproved. Rosemellyn Fisheries is the first proving ground, not the limit of the model.

Cyprinus carpioSpecies firstLake physicsSonar / bathymetryBehavioural stateEvidence graded
MAP-01 // Interactive Rosemellyn field model

Rotate into the swim. Then decide.

Open the 12-swim Rosemellyn intelligence map, select a bank position, rotate into its point of view, inspect live wind geometry and compare evidence-graded strategies without converting unknown water into fake certainty.

Open interactive swim map →
SPECIES SCIENCEPeer-reviewed physiology, behaviour, sensory biology and field studies.
VENUE EVIDENCEFishery records, rules, historic reports, planning and local environmental context.
FIELD MEASUREDSonar, temperature, oxygen, conductivity, captures and repeated transects.
HYPOTHESISUseful ideas stay labelled until observation survives an attempt to falsify them.
Decision architecture

From fish biology to one cast.

Live observation should outrank folklore. Venue history should modify broad carp priors. Every outcome becomes new evidence instead of a story written after the event.

01Carp biologyPhysiology, senses, learning, feeding, social behaviour.
02Water physicsTemperature, oxygen, light, wind, stratification, chemistry.
03HabitatDepth, substrate, structure, food, inflows, vegetation.
04Behavioural stateFeeding, travelling, basking, refuge, spawning, avoidance.
05Venue memoryRecent catches, pressure, seasonal patterns, individual fish.
06Live evidenceSonar marks, visible fish, water profiles and current conditions.
Current species findings

What survives the first evidence pass.

Strong

Carp are three-dimensional fish.

They are often benthic feeders, but can exploit middle and upper water. A fish at 12 ft over 30 ft is a different tactical problem from a fish on the bottom in the same GPS position. Fish depth and total water depth must be logged separately.

Strong

Survival is not preference.

Common carp tolerate extreme temperature and low dissolved oxygen better than many species. That does not mean marginal water is where they prefer to feed. Temperature and oxygen profiles define opportunity through the water column.

Strong

Detection ≠ attraction ≠ acceptance.

Carp chemical senses are extremely sensitive, but a detectable molecule is not automatically a long-range attractant or a palatable food. Bait needs separate scores for dispersion, orientation, oral acceptance and nutrition.

Strong

The mouth is a sorting machine.

Carp suck in food with water and sediment, manipulate it internally, chemically inspect it, retain edible particles and reject unwanted material. Soft silt is therefore not automatically “bad bottom”.

Strong

Carp learn angling risk.

Direct hooking reduces subsequent vulnerability and some caution can spread socially. The effect is real but not permanent. Recent pressure belongs in the model as a decaying variable, not a myth about lifetime memory.

Strong

Individuals matter.

Wild populations contain disproportionately active feeders and socially connected fish. A productive spot may be sampling a behavioural type, not revealing a universal preference of every large carp.

Field + science

Bait meaning can change.

Field experiments show bait performance changing over weeks. Population food history and learned familiarity can matter alongside chemistry. “Best bait” is therefore partly a moving local variable.

Mechanistic

Wind moves water, not magic.

Wind can mix the surface layer, move food, alter oxygenation and tilt thermal structure. Its fishing value depends on lake geometry and fish depth. “Follow the wind” is a hypothesis, not a universal command.

First proving ground

Rosemellyn Fisheries.

A deep former china-clay working near Roche, Cornwall. The public venue facts provide the outer frame; the field model and repeated measurements provide the internal decision space.

Verified public venue facts

Known before inference.

  • Former clay mine / pit now used as a carp fishery.
  • Around 6 acres with water reported to roughly 36 ft.
  • 12 named designated fishing swims.
  • Bait boats are permitted under the published rules.
  • 2025 planning documentation anchors the site at grid reference SX 009 598 and provides a north-oriented site plan.
Working field model

Two basins, one constriction.

  • Large main basin and smaller secondary basin in the current bathymetric working model.
  • A pronounced central neck / funnel links them.
  • Swim positions are treated as fishing-origin points around the perimeter.
  • Exact underwater features remain evidence-coded: measured where validated, hypotheses where not.
  • The neck is a priority candidate for movement, exchange and repeated sonar testing.
Key discipline: a beautiful contour does not become truth because it looks plausible. The system distinguishes verified public facts, measured field data, reported information and inference.
First field protocol

Measure change, not just location.

1

Observe first

Surface activity, bubbles, wind lanes, light, disturbance and visible fish before introducing bait or repeated boat passes.

2

Repeat transects

Run the same basin / neck / basin sonar routes at several times rather than wandering until a mark appears.

3

Record the Z-axis

Store fish depth, total depth and distance above bottom separately for every meaningful return.

4

Profile the water

Temperature first; dissolved oxygen and conductivity when available. Compare the two basins and the connecting neck.

5

Fish contrasting ideas

One presentation can test bottom feeding, another suspended fish, another a control. Change one major variable at a time.

6

Log every outcome

Takes, liners, hook holds, rejects, no-action periods and moves all become useful data — not just landed carp.

Falsifiable hypotheses

The model has to be allowed to lose.

HypothesisWhy it mattersWhat supports itWhat weakens it
Preferred vertical layerDeep water can be present without being productive.Repeated marks cluster at similar depths over different bottom depths.Marks consistently track the bottom instead.
Neck as movement corridorTwo basins may exchange fish through a constrained route.Repeatable time-linked passage through the constriction.Basins change independently while the neck stays quiet.
Wind shifts usable waterWind acts through water movement, mixing and geometry.Fish/water layers shift with sustained changes in wind forcing.Distribution remains stable despite meaningful wind change.
Basins behave differentlyDifferent shape/depth/exposure can produce different water columns.Temperature or oxygen profiles differ consistently between basins.Profiles are effectively identical across repeated sessions.
Inflow / seep zones matterGroundwater or pipes can alter temperature, oxygen and chemistry locally.Persistent spatial anomaly in temperature, DO or conductivity.Anomaly disappears or merely follows transient weather.
Pressure modifies catchabilityCarp learn recent angling danger.Catchability declines after repeated captures at the same spot/presentation.No relationship after enough controlled observations.
Bait model

Retire the single word “attractor”.

PropertyQuestion
DispersionDoes a meaningful chemical signal actually leave the bait in this water temperature?
DetectionCan the carp sense the released compounds?
OrientationDoes detection cause movement toward the source?
ExplorationDoes searching behaviour increase around the source?
PalatabilityDoes oral contact encourage retention rather than rejection?
AcceptanceIs the item swallowed under the current behavioural state?
NutritionDoes ingestion provide a worthwhile reward?
FamiliarityHas this population learned that the signal predicts safe food?
Risk associationHas the same signal or presentation recently been paired with hooking?
MechanicsDoes the hookbait enter, move, reset and hook reliably during real carp feeding?
Data contract

What one useful field record eventually contains.

timestampswimx / ywater_depthfish_depthabove_bottombottom_classwater_tempdissolved_oxygenconductivitywind_vectorlightpressure_trendbaitpresentationangling_pressurefish_idoutcomehook_locationconfidence
Unknown is a valid value. Unscanned water should remain UNKNOWN, not be painted as poor. A recent field measurement should outrank an old anecdote. One historical capture should not become a permanent hotspot.
Not claimed

Things the project refuses to pretend it already knows.

Exact fish locations
until live observation or repeated field evidence supports them.
A universal “best swim”
independent of water state, pressure, season and current fish position.
A universal best bait
that ignores chemistry, familiarity, nutritional state and recent angling.
An exact thermocline
from a fuzzy sonar band without an independent temperature profile.
Magic pressure/moon rules
without venue-scale statistical support and plausible mechanisms.
Perfect bathymetry from appearance
unless the contour has actually been measured or independently validated.
Selected research ledger

Sources currently shaping the model.

This is a working ledger, not a final bibliography. Species-wide evidence is deliberately kept separate from venue-specific material.

RESEARCH STATE: ACTIVENext layers include bait release chemistry, rig mechanics, growth/condition energetics, seasonal natural-food cycles, disease/parasites, turbidity, vegetation, predators, navigation and winter aggregation.