Abstract
In order for numerical weather prediction (NWP) models to correctly predict solar irradiance reaching the earth’s surface for more accurate solar power forecasting, it is important to initialize the NWP model with accurate cloud information. Knowing where the clouds are located is the first step. Using data from geostationary satellites is an attractive possibility given the low latencies and high spatio-temporal resolution provided nowadays. Here, we explore the potential of utilizing the random forest machine learning method to generate the cloud mask from GOES-16 radiances. We first perform a predictor selection process to determine the optimal predictor set for the random forest predictions of the horizontal cloud fraction and then determine the appropriate threshold to generate the cloud mask prediction. The results show that the random forest method performs as well as the GOES-16 level 2 clear sky mask product with the ability to customize the threshold for under or over predicting cloud cover. Further developments to enhance the cloud mask estimations for improved short-term solar irradiance and power forecasting with the MAD-WRF NWP model are discussed.
Highlights
Numerical weather prediction (NWP) models require an accurate cloud initialization to provide reliable solar irradiance forecasting
In this study we have focused on improving cloud detection prediction using a machine learning method to capture the non-linear relationships among Geostationary Operational Environmental Satellites (GOES)-R
An intermediate step in the cloud detection retrieval is the retrieval of the cloud fraction on our target grid, which can be used to complement other GOES-R
Summary
Numerical weather prediction (NWP) models require an accurate cloud initialization to provide reliable solar irradiance forecasting. This is especially the case for short term forecasts or nowcasts where the forecast window is only a few hours long [1,2]. Data from the current generation of sensors on geostationary satellites are well suited for nowcasting applications [3] These satellites provide high-spatio temporal resolution sampling in the visible and infrared with similar spectral resolution as polar orbiting satellites. The data is available shortly after the samples are processed, which provides around 1 min latencies to determine cloud properties for NWP initialization
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have