Weather Sandbox lets you grow forests, start fires, and flood the landscape. These are officially named environmental outcomes, but the public description does not explain their exact triggers, growth speed, spread rules, water levels, recovery, or relationship to each atmospheric control. Treat the landscape as a second observation layer: weather describes the event above you, while nature records what changes on the ground.
Use a clean baseline and change one condition at a time. A forest, fire, or flood that appears after a storm may be caused by the named weather, one underlying input, a tool, a biome rule, or a combination. Repeating the test is how you separate those possibilities.
Environmental reaction matrix
Leave unsupported cells blank in your notes. “Unknown” is a valid result when the game has not shown a value or label. It is better than filling the matrix with real-world assumptions that may not match the simulation.
Prepare a clean landscape baseline
Before testing, record the biome, visible vegetation, standing water, burned areas, sky, and all four variable values. If the landscape already contains trees, define what counts as new growth: number, area, height, density, or an explicit game indicator. If water is already present, mark its shoreline or choose a stable camera reference.
Check for equipped tools and active contracts. A tool might change the landscape directly, while a contract may guide you toward a condition. Do not attribute the result to temperature, wind, humidity, or pressure if you also used an unknown tool during the trial.
Confirm the reset behavior. A control reset may restore atmospheric inputs without removing the forest, extinguishing fire, or draining water. Record each layer separately. If only a fresh server restores the landscape, note that as the recovery method for the current build.
Test forest growth
Choose a patch of terrain you can recognize from a fixed viewpoint. Record the starting vegetation and time. Change one atmospheric input by a repeatable amount, wait, and note the first visible addition or transformation. Continue long enough to see whether growth stops, spreads, or accelerates.
Return to baseline and test the same change again. If growth does not repeat, compare biome, existing plant density, landscape reset, tool state, and delay. A maximum density or persistent world state can make a valid trigger appear to fail on the second attempt.
Once a single-input response is repeatable, test a second input while holding the first at baseline. Combine them only afterward. Avoid describing the game as using realistic plant biology unless the developer says so; report the actual current response.
Test fire and spread
Start with an unburned, clearly visible area. Record whether vegetation is present and whether a tool is selected. If fire begins, note the ignition location, delay, spread direction, affected material, duration, and what stops it. Watch wind motion and precipitation, but do not claim either controls fire until a comparison supports it.
For a spread test, keep the ignition route fixed and change only one suspected influence. Compare distance and direction from the same camera reference. Do not stand inside an effect merely to measure it if the game presents a hazard; preserve your session and progression first.
Test recovery separately. Reverse the input, produce precipitation, use an explicitly labeled tool, or reset only when the game makes the action clear. Record whether flames end, burned terrain restores, and vegetation returns. Extinguishing and recovery are different outcomes.
Test flooding and retreat
Use a landmark as a waterline reference. Record existing water and low terrain before the trial. Change one input, wait, and note where water first appears, how far it spreads, whether it rises continuously, and whether the game labels the result or advances a contract.
If a storm occurs at the same time, separate the weather record from the flood record. Repeat the atmospheric route without changing the landscape in another way. Then test whether reversing the input or resetting causes water to retreat. A flooded scene may persist after the weather ends.
Do not assign a numerical depth unless the UI provides one. Visual coverage, landmark height, and elapsed time are enough for a useful comparison.
Compare input effects without a wide table
Temperature
Record the displayed value and compare one increase or decrease. Do not assume a real ignition or freezing point.
Wind
Watch movement and spread direction. Distinguish visible association from a repeated cause.
Humidity
Compare sky, precipitation, vegetation, fire, and water response, but preserve every other input.
Pressure
Record the game value exactly. Test it directly instead of borrowing a real-world weather rule.
Biome
Change it only after the same trial is repeatable. A different landscape is a new experiment context.
Tools
Use one only after recording its exact current name and purpose. Keep direct actions separate from atmospheric effects.
The Mechanics hub explains the baseline and single-variable method. The Weather index helps identify the event above the landscape, while this page records the response below it.
Avoid false conclusions
Timing can create a misleading link. A forest may begin growing after you changed pressure even though an earlier humidity change had a delay. A fire may spread with the visible wind while an equipped tool or persistent setting is responsible. A flood may remain after the trigger ends, making later values look like its cause.
Use reset, repetition, and reversed tests. If an increase produces the reaction twice and a return to baseline stops it twice, confidence improves. If the result varies, label it inconsistent and preserve the context rather than inventing a hidden probability.
Do not generalize from one biome. After a route works reliably, repeat it in a second biome with the same values and tools. Record whether the outcome is unavailable, visually different, faster, slower, or unchanged.
Connect nature to progress and Collection
A contract may ask for an environmental result, and a weather collection entry may unlock during the same experiment. Record the exact task counter and Collection name before and after the trial. That shows what the game recognized without assuming the landscape itself has a separate catalog entry.
If a tool or biome unlocks, log the displayed requirement and reward in Progression. Do not turn a one-time unlock sequence into a universal order until the current UI confirms it for another player or session.
Nature questions
How do I grow a forest?
The feature is official, but no verified public source provides a complete current trigger. Establish a landscape baseline, test one input, wait for new growth, reset, and repeat before sharing the route.
How do I start or stop a fire?
Record whether ignition came from atmospheric controls, a tool, a contract, or an existing event. Test extinguishing and terrain recovery separately; no exact public formula is confirmed here.
How do I flood the landscape?
Use a visible waterline reference, test one input, and record both rise and retreat. Do not treat rain, standing water, and a recognized flood as automatically identical.
Does wind spread fire?
That is a reasonable test question, not a confirmed game rule. Keep ignition fixed, change wind only, and compare repeated spread from the same baseline.
Evidence note
Forest growth, fire, and flooding are named in the official Roblox Experience description checked September 26, 2026. Their input relationships, rates, limits, spread, persistence, tools, and biome rules remain current-build observations.