Journal Club: 22-05-2018

Article title: Bias correction of a novel European reanalysis data set for solar energy applications

This paper examines the skill of a regional reanalysis COSMO-REA6 compared to BSRN observations. They found a systematic error in the reanalysis:

  • An underestimation of GHI in clear sky conditions – due to the aerosol climatology which causes too strong solar extinction.
  • An overestimation of GHI in cloudy sky conditions – due to the underestimation of optical depth of clouds.

Therefore, a post-processing method was developed to improve both regimes separately. A different scaling factor is developed for clear sky and cloudy sky situations.  The situations were separated based on transmissivities.

For cloudy conditions twelve scaling factors were estimated, one for each month.  While for clear sky conditions six scaling factors were estimated for different solar elevation regimes. There was a reduction in the systematic bias (Figure 1).

Figure 1: Monthly mean bias for COSMO-REA6 (dashed) and COSMO-REA6pp (solid lines) for (a) clear sky situations and (b) cloudy sky situations.

They also examined ramp events on scales of 3h, 1h and 30mins. The observed ramp rates are well represented on the longer time scales but on smaller time scales the performance decreases and ramp rates are underestimated.

Relevancy for our work:

  • Consider separating our post-processing methods depending on current weather situations.
  • Examine the solar energy ramp rates.
  • To remember when working with daily mean values, BSRN stations only take measurements when the solar angles are greater than 10°, whereas some reanalysis estimates include all values.