Statistics Glossary → Monte Carlo (simulation) envelope
Monte Carlo (simulation) envelopeconcept
A band built from the values of a summary function (e.g. Ripley's K or L) in many simulations under the null hypothesis, used to check visually whether the observed curve is consistent with chance.
In point-pattern analysis the null hypothesis is often complete spatial randomness (CSR): points are scattered independently and uniformly over the study area. Random patterns with the same number of points (or the same intensity) in the same area are generated nsim times under CSR, and the summary function is computed for each. At every distance, the smallest and largest simulated values (or the values of rank nrank) form the lower and upper limits of the envelope. If the observed curve rises above the envelope there is clustering at that distance; if it falls below, the pattern is regular (inhibitory).
A pointwise envelope must be interpreted with care: for a test at a single pre-chosen distance, the significance level is α = 2·nrank / (nsim + 1) (e.g. 0.02 for 99 simulations and nrank = 1). When the curve is examined at many distances at once, however, the probability that it leaves the envelope somewhere is much larger than this; this is a multiple-testing problem. Global envelopes control the familywise error rate over all distances by building a single test based on the maximum deviation. The more simulations, the more stable the envelope.
An everyday example
You map the locations of 200 trees in a forest and wonder whether they grow in groups. The computer places 200 trees completely at random in the same area 99 times and, each time, counts how many neighbours trees have within a given distance. If, at short distances such as 10 metres, the real forest has more neighbours than all 99 random forests, you conclude that the trees are clustered at that scale.
In MerQur
It is controlled in Point Pattern Analysis (Ripley K/L, G/F/J, CSR) with the 'Number of CSR simulations', 'Envelope rank (nrank)' and 'Fixed n' options.
Analyses where this concept appears
Point Pattern Analysis (Ripley K/L, G/F/J, CSR)