A study of species distribution prediction based on kernel density estimation

Abstract

The occurrence of Vespa mandarine in the State of Washington is terrible news for the local Western bees, as Vespa mandarine is a very effective predictor. The Washington State government is taking the Vespa mandarine invasion seriously and hopes that people will report it when pests are found. This article processes the sighting reports provided, establishes a model, predicts the potential distribution of coriander, and can effectively provide the possibility of a new report being positive so as to quickly determine the most likely positive report and help the government allocate resources rationally. First, a species distribution prediction model based on kernel density estimation (KDE) was established. We took the Gaussian function as the kernel function, performed KDE on the given data, and controlled the search radius of the kernel function to be 30km from the center of the nest to obtain a probability density heat map reflecting the possibility of coriander spreading to a certain area, which shows that the scope of the spread and the possibility of each location are predictable. It is also essential to consider the circumstances under which the hives are cleared. Therefore, when a hive is removed, we will update the KDE model. If the distance is less than 30km, the point closest to the coordinates of the cleared hive will be deleted.