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

Pub­lished in Solar Ener­gy, 164, 12–24, 2018: 

One of the major chal­lenges dur­ing the tran­si­tion phase of the ener­gy sys­tem is to main­tain the bal­ance between ener­gy sup­ply and demand. Ris­ing ques­tions are often relat­ed to site map­ping, vari­abil­i­ty, extremes and com­pen­sa­tion effects for exam­ple. A fun­da­men­tal source of infor­ma­tion to answer these ques­tions are high qual­i­ty data sets of renew­able ener­gy relat­ed vari­ables. As reanaly­ses pro­vide all rel­e­vant data to assess wind and solar pow­er gen­er­a­tion over a long peri­od of time (decades) in a grid­ded con­sis­tent way, they exhib­it great poten­tial in the field of renew­able ener­gy. A new region­al reanaly­sis is COSMO-REA6, which cov­ers the Euro­pean domain over the years 1995–2014 with a hor­i­zon­tal res­o­lu­tion of about 6 km and a tem­po­ral res­o­lu­tion of 15 min. In this paper, we first assess the qual­i­ty of the Glob­al Hor­i­zon­tal Irra­di­ance (GHI) pro­vid­ed by COSMO-REA6. High qual­i­ty GHI mea­sure­ments obtained through the Base­line Sur­face Radi­a­tion Net­work (BSRN) are used as ref­er­ence and reveal sys­tem­at­ic short com­ings in the reanaly­sis: (1) an under­es­ti­ma­tion of GHI in clear sky sit­u­a­tions and (2) an over­es­ti­ma­tion of GHI in cloudy sky sit­u­a­tions. In order to reduce these sys­tem­at­ic regime depen­dent bias­es, a post-pro­cess­ing is devel­oped. The applied post-pro­cess­ing method is a scal­ing based on orthog­o­nal dis­tance regres­sions for two dif­fer­ent regimes, i.e., “clear sky” and “cloudy sky”. The two regimes are dis­tin­guished by the use of a trans­mis­siv­i­ty thresh­old. The post-processed GHI shows a sig­nif­i­cant reduc­tion of the sys­tem­at­ic bias­es and an improve­ment in rep­re­sent­ing the mar­gin­al dis­tri­b­u­tions. A spa­tial cross-val­i­da­tion shows the applic­a­bil­i­ty to the whole mod­el domain of COSMO-REA6. More­over, COSMO-REA6 as well as the post-processed GHI data reveal an added-val­ue when com­pared to glob­al reanaly­sis ERA-Inter­im and MERRA‑2. The high­er res­o­lu­tion reanaly­sis exhibits a sig­nif­i­cant­ly bet­ter per­for­mance of rep­re­sent­ing GHI vari­abil­i­ty, as well as bias­es, RMSE and oth­er con­ven­tion­al scores. The post-processed GHI data are freely avail­able for down­load.

Authors: Frank, C. W., Wahl, S., Keller, J. D., Pospichal, B., Hense, A., Crewell, S.

Link to arti­cle