积雪特性参数分析及雪深模型建立

通过2015~2016年冬季积雪观测试验,测量自然条件下雪深、雪层温度、雪层密度、雪层液态含水率等积雪特性参数,分析积雪特性参数影响因子和发育变化规律,构建径向基神经网络雪深模型。结果表明,空气、地表温度及水汽压是影响雪参数发育三大主要气象因子。在时长为124 d覆雪期中,雪深最大平均融雪速率达3.20 mm·d-1,出现在融雪后期。雪层密度随雪深增加波浪式上升,其中稳定期全层密度0.1~0.4 g·cm-3;雪层液态含水率0.5%~3.5%,分层液态含水率中层较大,积雪表层和底部波动较小;全层积雪孔隙率平均值为72.3%,积底层孔隙率廓线下降最快,极差最大。经两次优化后,雪深模型最大绝对误差为0.614 1 cm,最大相对误差为5.85%,模拟精度控制在6 mm以内。

Analysis on snow characteristic parameters and construction of snow depth modeling

By snow observation experiment in 2015-2016,snow parameters of depth,temperature,density and liquid water content were collected under the natural conditions,the influence factors and the development law of snow parameters were analyzed,and the snow depth model was constructed by radial basis neural network.The results showed that air temperature,surface temperature and water vapor pressure were the three main meteorological factors affected snow parameters.In the 124 d during the whole snow cover period,the maximum snow melting rate of the snow depth reached 3.20 mm· d-1 and appeared in the late snowmelt period.With the increase of deep snow density wave risen,the snow density varied between 0.1 to 0.4 g· cm3 during the stable period,and the average liquid water content of snow increased from 0.5％ to 3.5％.The bottom snow layer exhibited larger parameter variations compared to those in the surface and middle layers.The average porosity of snow was 72.3％ in the entire snowpack,ranged from 4％ to 19％ during the accumulation period and from 7％ to 25％ during the snowmelt period,in which the porosity of the bottom layer displayed the fastest decline.The maximum absolute error of the snow depth model after two times optimization was 0.6141 cm,the maximum relative error was 5.85％,and the simulation precision could be controlled within 6 mm.