DNICast Report on Combination of nowcasted DNI

Abstract

Within the EU FP7 DNICast project (Direct Normal Irradiance Nowcasting methods for the optimized operation of concentrating solar technologies) a set of innovative and improved forecast methods are proposed. One additional task in the project is the proposal and analysis of different options to combine the data provided by forecast sources. This report is focused on this last task of combination, describing the proposed methodologies for the combination. Inputs for the combination come from nowcasting data sets based on different methods already developed in the project. After this task, the data sets are prepared for their validation and their application in CST (Concentrating Solar Technology) models. After a review of forecasts combination methodologies, the report is focused on showing three different methodologies for combining Direct Normal Irradiance (DNI) nowcast outputs. The first combination methodology weighs the different nowcast data sets according to their uncertainty estimations (UWA). The second combination method uses a time-dependent multi-regressive model (MRA) and the third is based on a weight calculation based on the Euclidean distance between forecasts as functional data (DWA). The three combination approaches are described in this report. Detailed analysis of the result provided by each methodology is presented in the deliverable 3.13.