Climate Time Series Prediction by Convex Polygons and the Kalman Filter

The number of monitoring stations essential for a region is key for installation and maintenance cost purposes, but also for data correction in the event that stations fail and miss data. Estimates that incorporate the distance to secondary stations are generally used, but climate components are spatial and should take into account only one dimension. In this paper, we propose an algorithm for performing approximations of climate data, such as temperature and relative humidity based on neighboring stations, using the area of the largest convex polygon containing the point of interest to calculate the estimates, instead of the distance; In order to test this method, real data observations are used and compared using different indices, with the simulated values for a given station.