基于CLDAS强迫CLM3.5模式的新疆区域土壤温度陆面过程模拟及验证

发布时间：2018-06-14 15:42

本文选题：土壤温度 + CLM.　；参考：《生态学报》2017年03期

【摘要】：利用中国气象局国家气象信息中心研发的中国气象局陆面数据同化系统(China Meteorological Administration Land Data Assimilation System,CLDAS)大气近地面强迫资料,驱动美国国家大气研究中心公用陆面模式(Community Land Model,CLM3.5),对中国新疆地区土壤温度时空分布进行逐小时Off-line模拟(模拟时段为2009—2012年);利用国家土壤温度自动站(新疆区域105站点)数据验证CLDAS驱动场强迫下的CLM3.5模式在中国新疆地区3个土壤层(5cm、20cm和80cm)的土壤温度模拟能力。研究发现:在月变化方面,第1层(5cm)土壤温度模拟与实测值差异最大,在每年7月最大差异达5k左右;第2层(20cm)在每年7月达最大差异(3k左右),而第3层(80cm)在每年7月均模拟的很好。造成这种现象的原因可能因为新疆地区7月前后浅层土壤温度变化剧烈,温度白天最高可达300K以上,昼夜温差大,导致模式不能很好抓住浅层土壤温度的变化趋势。研究还发现,在80cm土壤深度,模式在1月、12月的模拟结果均较前两层差。在日变化方面,研究发现:较浅的两层(5cm和20cm)土壤温度模拟值在夏季和秋季均较差。与月变化模拟结果类似的是,80cm土壤层日变化在1、12月模拟较差,然而在其他时段却模拟的很好。在小时变化方面,分析发现:第1层土壤(5cm)模拟结果在每年的1—4月及9—11月的全天(即24 h),模式也会有不同的偏差:其中,在03UTC—21UTC之间主要表现为模式结果比观测结果偏高,而在日内21UTC—00UTC主要表现为模拟结果偏小。在每年的5—8月,全天模拟值都偏小,其中在09UTC达当日最大值。而距离第2层(20cm)处的土壤温度模拟值在大部分月份都偏差较小(-1K至1k之间),并在日内12UTC偏差达到当日最大值。研究发现,在土壤20cm处,模式模拟的最大值较观测值提前,而第3层(80cm)的土壤温度基本不受日内变化影响,表现较为平稳。造成这种影响的原因可能是因为新疆地区5—8月、9—11月为昼夜温差大,深层土壤温度较浅层土壤温度温差变化小,这也造成了模式对于浅层土壤模拟较深层差的主要原因。总体研究表明:CLDAS驱动场强迫下的CLM3.5模式可较为精确的模拟中国新疆地区多年平均土壤温度时空分布,并较为准确的反映中国新疆地区土壤温度的小时、日、月及年际的变化规律。模式浅温度模拟不好的原因可能与模式参数化方案及地表参数有关,后期将继续修正该问题。
[Abstract]:Using the atmospheric near-surface forcing data of China Meteorological Administration Land data Assimilation system CLDAS-developed by the National Meteorological Information Center of China Meteorological Administration, the land surface data assimilation system of China Meteorological Administration is used. Driving the National Atmospheric Research Center Common Land Model CLM 3.5, the hourly Off-line simulation of soil temperature distribution in Xinjiang, China is carried out (the simulation period is 2009-2012), and the national soil temperature automatic station (Xinjiang region) is used to simulate the soil temperature distribution. The soil temperature simulation ability of CLM3.5 model forced by CLDAS driving field in three soil layers (20cm and 80cm) in Xinjiang region of China was verified by the data from the 105 site. The results show that the difference between the simulated and measured values of soil temperature in the first layer (5 cm) is the largest, and the maximum difference is about 5 k in July every year. The second layer (20 cm) has the largest difference of about 3 k in July, while the third layer (80 cm) simulates well in July of each year. The reason for this phenomenon may be that the temperature of shallow soil in Xinjiang region changes sharply before and after July, the highest temperature can be more than 300K during the day, and the diurnal temperature difference is large, which leads to the model can not grasp the change trend of shallow soil temperature very well. It was also found that at the soil depth of 80cm, the simulation results of the model in January and December were worse than those of the first two layers. In terms of diurnal variation, it was found that the simulated values of soil temperature in the two shallower layers of 5 cm and 20 cm were worse in summer and autumn. The results are similar to those of monthly variation. The diurnal variation of 80 cm soil layer is in January, but the simulation in December is not good. However, in other periods, the simulation is very good. In terms of hourly variation, it was found that the simulated results of the first layer of soil were different from those of the whole day (i.e. 24 h) in January-April and September-November of each year: among them, Between 03UTC-21UTC, the model results are higher than the observed ones, while the in-day 21UTC-00UTC results show that the simulation results are small. In May-August of each year, the full-day simulation value is small, and the maximum value is reached at 09UTC. The simulated values of soil temperature at the distance of 20 cm from the second layer were smaller in most months and reached the maximum of 12UTC in the day. It was found that the maximum value of the model was earlier than the observed value at the soil 20cm, while the soil temperature of the third layer (80cm) was not affected by the variation in the day, and the soil temperature was stable. The reason for this effect may be that the temperature difference between day and night is larger in May / August and September / November in Xinjiang, and the temperature difference of deep soil is smaller than that of shallow soil, which also causes the main reason that the model is worse for simulating shallow soil. The overall study shows that the CLM3.5 model forced by the drive field of the: CLDAS can accurately simulate the temporal and spatial distribution of the average soil temperature in Xinjiang, China, and more accurately reflect the hours and days of soil temperature in Xinjiang, China. The law of change between months and years. The reason for the bad simulation of model shallow temperature may be related to the model parameterization scheme and surface parameters, and the problem will continue to be corrected at a later stage.
【作者单位】：中国水利水电科学研究院;新疆大学干旱生态环境研究所;中国气象局国家气象信息中心;中国科学院寒区旱区环境与工程研究所冰冻圈科学国家重点实验室;中国科学院新疆生态与地理研究所;中国气象局华云信息技术工程有限公司;
【基金】：水利部公益性行业科研专项经费(201301103) 国家自然科学基金重点项目(41130531)
【分类号】：S152.8

【参考文献】