Snow parametrisations in COSMO-RU analysis of winter-spring forecasts

Snow parametrisations in COSMO-RU: analysis of winter-spring forecasts 2009 -2010 COLOBOC Workshop. Moscow, 6 September 2010

Basic characteristics of New snow scheme ¡ Multilayerness ¡ Radiation is described explicitly ¡ Description of snow compaction by metamorphism and gravity ¡ Phase transition of liquid water in snow cover ¡ Description of water percolation with its next freezing and release of heat COLOBOC Workshop. Moscow, 6 September 2010

Methods and data ¡ Comparison 2 versions model COSMORu: “new” (with New snow scheme in TERRA) and “old” (previous snow scheme in TERRA) ¡ Integration period – 78 h (from 00 UTC) ¡ Data: - station measurements - decade measurements of snow survey on Roshydromet’s stations COLOBOC Workshop. Moscow, 6 September 2010

Research region: European part of Russia north centre south

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) north 83 mm 91 mm 95 mm Medvegegorsk : SWE – measurements and 72 h forecasts COLOBOC Workshop. Moscow, 6 September 2010

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) north 18 mm 16 mm Medvegegorsk : SWE – measurements and 72 h forecasts without mistakes COLOBOC Workshop. Moscow, 6 September 2010

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) centre 70 mm 81 mm 108 mm Inza : SWE – measurements and 72 h forecasts COLOBOC Workshop. Moscow, 6 September 2010

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) centre 37 mm 0 0 Inza : SWE – measurements and 72 h forecasts without mistakes COLOBOC Workshop. Moscow, 6 September 2010

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) south Harabaly : SWE – measurements and 72 h forecasts COLOBOC Workshop. Moscow, 6 September 2010

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) south Harabaly : SWE – measurements and 72 h forecasts without mistakes COLOBOC Workshop. Moscow, 6 September 2010

Snow characteristics: SWE and snow depth (1. 12. 09 -28. 02. 10) Correlation coefficients for snow depth according station data and 72 h forecasts north station new centre old Verhnyaya Toyma 0, 89 0, 94 Krasnoborsk 0, 88 0, 95 Lalsk 0, 90 0, 92 Medvegegorsk 0, 92 0, 96 Pinega 0, 92 0, 94 Blagodarniy new old station new old 0, 84 0, 85 0, 84 Kolomna 0, 92 0, 94 Bologoe 0, 93 0, 94 Michurinsk 0, 90 0, 96 Buzuluk 0, 93 Mogga 0, 90 0, 95 Buy 0, 87 0, 92 Morshansk 0, 94 0, 96 Vetluga 0, 85 0, 94 Poniri 0, 85 0, 90 0, 81 0, 86 Radishevo 0, 96 Dmitrov old 0, 81 0, 94 0, 95 Rybinsk 0, 92 0, 97 Inza 0, 95 Rilsk 0, 83 0, 86 Kamishin new Anna Gotnya south station 0, 86 0, 94 Spas-Demensk 0, 92 0, 95 Verhniy Baskunchak 0, 55 0, 42 Karabulak 0, 91 0, 90 Urupinsk 0, 75 0, 77 Modok 0, 87 0, 72 Karachev 0, 91 0, 97 Frolovo 0, 78 0, 69 Nalchik 0, 77 0, 65 Prohladnaya 0, 86 0, 78 Harabaly 0, 89 0, 84

Snow depth north 00 С centre 00 С south 00 С

T 2 m (11. 03. -29. 03. 09) night Kursk Moscow day Kursk Moscow COLOBOC Workshop. Moscow, 6 September 2010

T 2 m (29. 03. 10)

Snow fractional cover - ice covered part of grid element - SWE - parameter

Differences between forecasts with “standard” cf=0. 015 m and cf=0. 05 m Snow depth T 2 m Day 1 Day 2

T 2 m (29. 03. 10). Experiments if then

T 2 m (29. 03. 10). Experiments Kaluga Smolensk Kursk

Snow fractional cover. Experiments If hsnow>0. 3 m then cf_snow=0. 01 m

Snow characteristics: snow density (1. 12. 09 -28. 02. 10) Snow density for field, kg/m 3 date december 10 region 20 north 170 centre 31 137 south january 5 198 10 15 20 25 31 5 10 15 197 199 201 209 210 217 230 196 136 184 195 229 308 25 240 20 197 105 february 28 196 244 251 Snow density forest date december region 10 20 january 31 5 10 15 february 20 25 31 north 193 178 193 centre 85 158 195 5 10 197 251 188 15 20 203 214 216 25 28 195 249 248

Conclusions • - SWE Forecasts are considerably overestimated by two versions of model COSMO-Ru. The cause is an inaccuracy of SWE initial data, though snow depth initial data is quite correct. Needed: correction of the forming algorithm of SWE initial data using improved values of snow density • - Snow density varies during the period of snow cover existing in time scale and in each region (from north to south). It also have some differences between field and forest. So SWE calculation may take into consideration regional features of snow density. • - During snow accumulation period New snow scheme tends to overestimate snow depth after snowfalls. So, in New snow scheme Recommendation: to do the correction of the computing algorithm of fresh snow density in New scheme • - During snow melting period New snow scheme reproduce more realistically: - time dependence of SWE; - T 2 m at night (due to using New snow scheme there is the improvement of T 2 m forecast by 1, 5 -2 C). It is connected with New snow scheme’s description of freezing daily melted water and following release of heat. • • - During snow melting period the biggest mistakes of T 2 m forecasts were in vast regions for two versions (till 10 C), connected with equating to 0 C surface temperature of cell with snow. - Modifications with the snow fractional cover algorithm allowed to reduce the region area with mistaken temperature. Recommendation: to improve the algorithm of taking into account surfaces without snow during snow melting periods.

Thank you for your attention! COLOBOC Workshop. Moscow, 6 September 2010