初值对中国东部初夏土壤湿度可预报性影响

The Impact of Initial Conditions on Soil Moisture Predictability in Early Summer in Eastern China

土壤湿度是影响天气和气候非常重要的因子之一,但目前针对土壤湿度可预报性的研究报道相对较少。该文在对BCC_CSM模式进行了适合的陆面初始化的条件下,设计了两组在中国东部地区采用不同土壤湿度初值的回报试验研究该地区土壤湿度的可预报性及初值对其可预报性影响问题。试验结果表明:BCC_CSM模式在真实的外场强迫下可以模拟出相对合理的土壤湿度;土壤湿度的可预报性在表层约为3候,随着深度的增加,土壤湿度的可预报性持续时间增加,在中层预报性甚至能达到月尺度以上;初值对于土壤湿度的预报存在影响,在表层影响时间约为2～3候,影响时间随着深度增加;浅层土壤湿度受降水的影响较大,浅层土壤湿度变化滞后降水变化约1～2d,中层土壤湿度变化与降水变化存在5d左右的滞后关系。

Soil moisture has an important impact on weather and climate. Some previous soil moisture initialization experiments indicate that appropriate initial soil moisture can improve the forecasting skill of the model. The development of land surface initialization method is the urgent demand for improving the forecasting ability of the land model. However, studies about the predictability of the soil moisture in model are still relatively insufficient. To investigate the impact of soil moisture initial value on the prediction of soil moisture in BCCCSM2MR model, two experiments are carried out using a respectively true initial value of soil moisture（OH experiment） and an ideal initial value of soil moisture climatology（CH experiment）.The lack of soil moisture observations and coordination between the observation and model is a difficult problem in the development of land initialization method. The GSWP（Global Soil Wetness Project） introduces a new method to get the appropriate land initial data. The method indicates land initial data can be obtained using the accurate observed field（including atmospheric field, precipitation data, and radiation field） to force the land model. Therefore, an OF experiment is designed, which forces BCCCSM model from 1994 to 2013 by using the NCEP atmospheric reanalysis data and China National Meteorological Information Center（NMIC） precipitation data to produce a good model initial value. The soil moisture from OF experiment can reflect the observed distribution and variation of soil moisture over eastern China as the corresponding correlation is high. Thus, it will be used as observations. Based upon soil moisture data from OF experiment, OH and CH experiments are respectively initialized on 1 May during 1994 — 2013,and the lead time is six months. Comparison of two hindcast experiments shows that the predictability of soil moisture in BCCCSM model is about 3 pentads in surface layer and even exceeds 1 month in deep layer in some regions of China. At the same time, soil moisture initialization has an impact on forecast skill of soil moisture. This impact can reach 2 — 3 pentads in surface layer and longer in deeper layer. Variations of soil moisture in shallow layers strongly depend on variations of rainfall, and there is a 1 — 2 days lag between variations of soil moisture and rainfall, but about 5 days for that in the middle layers.