What an average of readings represents
An arithmetic mean adds the readings and divides by their count, as defined in the NIST statistics handbook. For it to represent an area without weighting, each reading must stand for the same area on the same measurement plane. Several readings clustered in a bright corner would otherwise give that corner extra influence.
Use readings in lux under the same conditions: the same lighting setting, measurement height and relevant daylight state. Converting units first with the illuminance converter makes mixed-unit records comparable, but does not make differently collected samples interchangeable. This tool accepts an existing sample set; it does not specify a measurement grid or verify meter performance.
Read the minimum alongside the mean
For an artificial set of 100, 200, 300 and 400 lx, the total is 1,000 lx and the arithmetic mean is 250.00 lx. The minimum is 100.00 lx, the maximum is 400.00 lx, and minimum divided by mean is 0.40.
That ratio is dimensionless. A value of 0.40 says the lowest submitted reading is 40% of the submitted mean. It does not mean 40% of the area is illuminated, or that the lighting satisfies a uniformity requirement. A standard or design method may require a particular grid, area boundary and threshold before a uniformity criterion can be assessed.
The same average can conceal a different low point
The mean alone hides the low sample. Even a ratio of 1.00 says only that these submitted readings match; it does not prove there are no darker points between them. The maximum helps identify the observed range but is not a glare measure.
Weight readings by the area they represent
When representative areas differ, select supplied area weighting and use mean = Σ(EᵢAᵢ)/ΣAᵢ. Each positive area is the region represented by that reading, not the physical size of the sensor. Regions must not overlap or leave the intended boundary misrepresented. The formula extends the arithmetic mean by the relative area shares; it does not determine which areas a measurement can legitimately represent.
For an artificial 100 lx reading representing 1 m² and 300 lx representing 3 m², the weighted mean is (100 × 1 + 300 × 3)/(1 + 3) = 250.00 lx. Averaging the two readings without weights would give 200 lx and give the smaller region too much influence. Minimum and maximum remain 100 and 300 lx; minimum divided by the weighted mean is 0.40. Multiplying every represented area by the same factor leaves the weighted result unchanged.
Keep the original readings and area basis with either result. Meter specification range interprets a compatible instrument tolerance separately; averaging does not remove shared calibration or spectral errors.
Compare samples with your own reference
An optional reference separates the submitted readings into below, exactly at and above that illuminance. Record where the reference came from, the measurement plane and the purpose of the comparison. A declared assumption is suitable for a learning comparison; it does not become an Australian recommended target. The calculator supplies no target or compliance criterion.
Each category reports both the number of readings and its share of the declared represented area. Equal-area mode gives every reading the same weight. Supplied-area mode adds the represented areas within each category and divides by the total submitted area. A category’s reading count and area share can therefore tell different stories.
For the artificial pair of 100 lx representing 1 m² and 300 lx representing 3 m²:
At 200 lx, one of two readings is below the reference, but it represents one of the four submitted square metres: 1 ÷ 4 × 100 = 25%. The other reading represents 75%. These are shares of the submitted, declared represented areas. They do not establish how much of a whole room meets a target, nor account for unmeasured points or measurement uncertainty.
Comparisons use the original entered decimal values. For example, 100.00000000000000001 lx is above a reference of 100 lx even if both display as 100.00 lx elsewhere. A measured zero is exactly at a zero reference. Area percentages are rounded to two decimal places after classification, so rounded categories may sum to 99.99% or 100.01%.
Leave the reference blank to omit the comparison. An invalid reference or missing reference basis makes the comparison unavailable while valid basic statistics remain visible; correct those fields to obtain a comparison for the current samples.
A zero reading is different from a missing reading
A measured zero belongs in the sample set. If every submitted reading is zero, the mean, minimum and maximum are zero, but minimum/mean is undefined because it would divide zero by zero. If just the minimum is zero and the mean is positive, the ratio is valid and equals zero.
An empty field or missing comma-separated entry is not a measured zero. Correct incomplete records rather than filling them with zero. Values and the ratio display to two decimal places; display rounding is not used to calculate the mean or ratio. The output describes the supplied samples, without assigning an Australian compliance result.
Apply the same spatial arithmetic to measured PPFD
PPFD counts 400–700 nm photons arriving at the receiving plane, in µmol·m⁻²·s⁻¹. Choosing PPFD changes the quantity being summarised; it does not convert lux. Apogee explains why a lux/PPFD relationship depends on the spectrum. Use compatible quantum-sensor measurements or a calibrated spectral integral, retaining the position, orientation and operating conditions.
For hypothetical PPFD readings of 100 representing 1 m² and 300 representing 3 m², the same area-weighted expression gives 250.00 µmol·m⁻²·s⁻¹. A user reference of 200 separates 25% of the declared area below and 75% above. That is a description of these supplied readings, not a plant target or a prediction of crop response. Representative PPFD and time can then support a daily light integral plan.
Daylight coverage weights occupied time separately
Spatial samples taken under common conditions answer a different question from a sequence of daylight readings. The daylight-time mode counts the fraction of a declared occupied horizon at or above a user threshold. Each row represents a non-overlapping occupied period at one zone; different zones must have non-overlapping representative areas. Readings must describe daylight alone, with artificial light excluded. The tool does not establish a suitable time sampling interval or threshold.
For each zone, account for the entire occupied horizon. Use an explicit missing reading for an unmeasured period; a measured zero remains zero. Zone IDs keep the same area across their periods. With more than one zone, the weight is represented m² × occupied h. The known coverage is the sum of weights whose original entered daylight readings are at least the threshold, divided by the whole declared area-time.
The occupied horizon is 3 h and known coverage is 1/3 = 33.33%. Recorded exposure is 100 × 2 + 300 × 1 = 500 lx·h, while the observed time mean is 500/3 = 166.67 lx. Exposure, mean and threshold coverage describe different properties; one cannot replace another. Adding 1 h at exactly 200 lx produces 2/4 = 50.00% coverage because equality is included.
If 1 h of that 3 h horizon is missing, it is neither a confirmed failure nor a confirmed success. The output shows missing area-time, coverage within observations, and lower/upper whole-horizon bounds: missing data could be below or at/above the threshold. The recorded exposure excludes it and must not be read as a complete exposure total. These descriptive records do not become annual spatial daylight autonomy (sDA), useful daylight illuminance (UDI) or Australian compliance results.