Modelling and Validation of Satellite-Derived Hourly Global Horizontal Irradiance in Northeast Iraq Based On a New Quality Control Procedure

The thesis presents a procedure for obtaining high quality hourly Global Horizontal Irradiance (GHI) in areas with limited ground measurements of GHI such as northeast Iraq. This was done by comparing two broadly Quality Control (QC) approaches, developing a local test and investigating a new one utilising Air Temperature (AT) and Sunshine Duration (SD) variables for stations when only GHI is available. This also was achieved by using QC GHI data to validate two Satellite-Derived Datasets (SDDs) in a new area and correcting SDDs under cloudy-sky conditions with a simple method and by combing SDD with climate variables as new input combinations to model hourly GHI data in Artificial Neural Network (ANN) models. The results showed vast differences between the two QC procedures, while the two QC are applied in literature separately. The results of the local test showed better quality check than the other two. The new QC test using SD and AT has detected additional questionable data, which were not flagged by the other QC approaches. The SD tests can be used as a consistency check in GHI data, unlike AT tests for the same target. The hourly GHI ground data were well presented by the two SDDs, while the overall performance of HelioClim-3 version-5 (Hc3v5) is better than Copernicus Atmosphere Monitoring Service (CAMS) radiation service version-5 (CRSv3). The results of the bias correction of SDDs under cloudy-sky condition revealed high accuracy by decreasing bias by 10–80%. The results of modelling GHI hourly with ANN models were improved slightly by using the new input combinations generation on bias (0.0%) and root mean square error (9.5%). Overall, the results suggest that the SD can be used partly as a consistency test in QC of hourly GHI data, and AT can be used for a general check for the same target. The Hc3v5 and CRSv3 are considered to be a reliable source of GHI data in northeast Iraq. The results of ANN models with the new input combinations support their continued and future use to improve the outcomes of the hourly GHI modelling.