Estimation of Precipitation and No Precipitation Areas Using Spatial Classification

• Activity Number: 84 - Contributed Poster Presentations: Section on Statistics and the Environment
• Type: Contributed
• Date/Time: Monday, August 3, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Melissa Innerst* and Kyuhee Shin and Bo-Young Ye and GyuWon Lee and Joon Jin Song
• Companies: Juniata College and Kyungpook National University and Kyungpook National University and Kyungpook National University and Baylor University
• Keywords: machine learning; spatial classification; rainfall

Abstract

A common question of interest is whether or not there will be precipitation at a given location. There are many statistical and machine learning methods available to answer this question. In this talk, statistical and machine learning methods for estimating precipitation area are considered and compared. The data we use in this study were collected by a network of VIS rain gauges, a network of automated weather system (AWS) tipping-bucket rain gauges, and a S-band dual-polarimetric weather radar during ten different rainfall events of varying lengths.Additionally, two models are compared, one with the latitude and longitude of an observation included in the model covariates, and one without spatial information included. The mean squared prediction error (MSPE) resulting from leave-one-out cross validation (LOOCV) is used to measure the performance of the methods. Of the methods, the decision tree and random forest methods result in the lowest MSPE. Of the models, the model containing spatial information results in the lowest MSPE.