Abstract
The study presents a comparative analysis of solar radiation data from ground-based measurements conducted at the Botanical Garden in Riga. The measurements, recorded at one-minute intervals, were compared with corresponding datasets from the Copernicus Atmosphere Monitoring Service (CAMS) Radiation Service, the DWD-ICON numerical weather prediction model, and the ERA5 reanalysis. The high temporal resolution of the ground-based data enables the detection of short-term fluctuations caused by transient cloud events, which are often smoothed out in model or reanalysis datasets with coarser (15-minute to hourly) resolution. Results show that CAMS generally reproduces the seasonal and diurnal patterns of global horizontal irradiance (GHI) with good accuracy but tends to overestimate peak radiation values, while ERA5 underestimates GHI under clear-sky summer conditions. To improve comparability, a LIDAR-based shading correction was applied using local obstruction data (vegetation and buildings), which increased the overall accuracy index up to 84 %. The findings highlight the importance of integrating high-resolution ground-based measurements with reanalysis and model outputs for solar energy applications in urban environments, where local shading and fine-scale atmospheric dynamics play a critical role.