Evaluation of ERA5 and MERRA2 reanalyses to estimate solar irradiance using ground observations over Indonesia region
Yesi Sianturi, Marjuki, Kwarti Sartika
Abstract
This study examines the feasibility of using reanalyses datasets to estimate solar irradiance over the Indonesia region. Solar irradiance from two reanalyses, the ERA5 from ECMWF and MERRA2 from NASA's GMAO were evaluated using the quality-controlled global horizontal irradiance obtained from the ASRS network, for a one-year period. The evaluation is done on an hourly and monthly value basis. Statistical evaluation and data visualizations were used to evaluate both datasets. The biases under different levels of cloudiness were also assessed. The hourly solar irradiance estimated using the ERA5 reanalysis shows positive biases against the observed hourly solar irradiance, except in Bengkulu station (mean bias error ranges from -7 to 34 W/m2). On the other hand, lower estimates are shown by the MERRA2 reanalysis (mean bias error ranges from -40 to 1 W/m2). The evaluation over the monthly solar radiation also presents the same result; ERA5 overestimates the monthly solar radiation in most stations, whereas MERRA2 tends to underestimate the monthly solar radiation. The clearness index over Indonesia region varies from 0.1 - 0.7, which indicates the common cloudy condition over this area. On clear sky condition, or lower clearness index, the observed solar irradiance tends to be underestimated, whilst on cloudy days the estimates from both reanalyses tend to be higher than the observed ones. In general, the estimates from ERA5 show better agreement with the ground observations. The information about biases can be used to correct-bias the estimates. Cloud cover affects the biases of those reanalysis datasets and this finding should be kept in mind in further usage of the reanalysis dataset.