Forecasting seeing for the Maunakea observatories with machine learning
T. Cherubini, Ryan Lyman, Steven Businger
Abstract
ABSTRACT The staff at the Maunakea Weather Center (MKWC) has provided daily forecasts of optical turbulence for the summit of Maunakea for more than 20 yr. Observational measures of optical turbulence at Maunakea with which to validate official MKWC forecasts have been available since mid-2009. This paper presents a machine-learning approach to translate the MKWC experience into a forecast of the nightly average optical turbulent state of the atmosphere. Maunakea observational and forecast data were collected to build a predictive model of the total and free atmospheric seeing for the following five nights. The motivation for this work is two-fold: to provide a tool/guidance to the MKWC forecaster and allow for a dynamic calibration of the optical turbulence algorithm implemented in the MKWC Weather Research and Forecasting (WRF) model.