Abstract
To find an optimal location for a mm-wave astronomical observatory, all factors that directly or indirectly affect the water vapor column density should be considered. After estimating a weighted classification of these factors to obtain a range of acceptable values, places satisfying as many of these suitability conditions can be proposed as candidates. Here we analyze data from NASA and IDEAM to find places satisfying the best conditions to build a mm-wave astronomical observatory in Colombia, according to seven variables grouped into four classes. From NASA, we analyze the satellite data of: (i) relative humidity and (ii) cloud coverage/direct normal radiation, averaged monthly from 1983 to 2005. From IDEAM, we analyze data of relative humidity, sunshine, and (iii) precipitation/number of days with rain, averaged yearly over each month from 1981 to 2010. The data has been obtained in-situ by 2046 weather monitoring stations across Colombia, for which their (iv) altitude is known. For each quantity, we do a Principal Component Analysis, reducing the dimensionality of the yearly-averaged data to 2 components covering >90% of the variance. After this, we make a classification of the reduced-dimension data using a 4 cluster Gaussian Mixture Model (GMM), identifying similar geographic and climatological patterns. After selecting clusters of stations sharing optimal conditions (i.e. high altitude, low rain, etc.), we group and look for geographical clusters by applying a GMM on a Monte Carlo sampling of latitude, longitude, and altitude data in order to correct for biases. This method allows us to find regions of interest where further in-situ measurements of atmospheric absorption of mm-wave should be carried out in the future.
