A novel convective-scale regional reanalysis COSMO-REA2: Improving the representation of precipitation

Pub­lished in Mete­o­rol­o­gis­che Zeitschrift, 26, 345–361, 2017:

Atmos­pher­ic reanaly­ses are a state-of-the-art tool to gen­er­ate con­sis­tent and real­is­tic state esti­mates of the atmos­pher­ic sys­tem. They pro­vide a syn­the­sis of var­i­ous het­ero­ge­neous obser­va­tion­al sys­tems and mod­el sim­u­la­tions using a phys­i­cal mod­el togeth­er with a data assim­i­la­tion scheme. Cur­rent reanaly­ses are main­ly glob­al, while region­al reanaly­ses are emerg­ing for North Amer­i­ca, the polar region, and most recent­ly for Europe. How­ev­er, deep con­vec­tion is still para­me­ter­ized even in the region­al reanaly­ses. A nov­el con­vec­tive-scale region­al reanaly­sis sys­tem for Cen­tral Europe (COSMO-REA2) has been devel­oped by the Hans-Ertel Cen­ter for Weath­er Research – Cli­mate Mon­i­tor­ing Branch. The sys­tem is based on the COSMO mod­el and uses obser­va­tion­al nudg­ing for region­al data assim­i­la­tion. In addi­tion to con­ven­tion­al obser­va­tions, radar-derived rain rates are assim­i­lat­ed using latent heat nudg­ing. With a hor­i­zon­tal grid-spac­ing of 2 km, the mod­el runs with­out para­me­ter­i­za­tion of deep moist con­vec­tion. COSMO-REA2 pro­duces hor­i­zon­tal wind fields that rep­re­sent a real­is­tic ener­gy spec­trum for hor­i­zon­tal scales above 14 km. COSMO-REA2 is cur­rent­ly avail­able for sev­en years from 2007 to 2013.This study illus­trates the improved rep­re­sen­ta­tion of local pre­cip­i­ta­tion over Ger­many by the con­vec­tive-scale reanaly­sis COSMO-REA2 com­pared to coars­er grid­ded Euro­pean and glob­al reanaly­ses. A sys­tem­at­ic ver­i­fi­ca­tion using rain gauge data reveals the added val­ue of high-res­o­lu­tion region­al atmos­pher­ic reanaly­ses on dif­fer­ent time scales. On month­ly to annu­al time scales, region­al reanaly­ses yield bet­ter esti­mates of the spa­tial vari­abil­i­ty of pre­cip­i­ta­tion pat­terns which can not be pro­vid­ed by coars­er grid­ded glob­al mod­els. On hourly to dai­ly time scales, the con­vec­tive-scale reanaly­sis sub­stan­tial­ly improves the rep­re­sen­ta­tion of local pre­cip­i­ta­tion in two ways. On the one hand, COSMO-REA2 shows an enhanced rep­re­sen­ta­tion of observed fre­quen­cies of local pre­cip­i­ta­tion, espe­cial­ly for high pre­cip­i­ta­tion events, which is main­ly due to the high spa­tial res­o­lu­tion. On the oth­er hand, the assim­i­la­tion of radar data large­ly improves the spa­tial and tem­po­ral coher­ence between mod­el out­put and local obser­va­tions. Both lead to rean­a­lyzed pre­cip­i­ta­tion val­ues which are com­pa­ra­ble to point obser­va­tions.

Authors: Wahl, S., Bollmey­er, C., Crewell, S., Figu­ra, C., Friederichs, P., Hense, A., Keller, J.D., Ohlwein, C. 

Link to arti­cle