Probability of a dust emission event during the period that the soil active layer remains wet after a precipitation event

Soil moisture in the active aeolian layer (the top ~2 mm of the soil) impacts dust emission by increasing the threshold for emission, and thus precipitation has the potential to suppress dust emission. The purpose of this study was to use reanalysis and satellite data similar to those used in global and regional dust emission models to calculate the probability that a high wind event happens during the period that antecedent precipitation would have left the active layer wet. The results indicate that the answer to this question is more strongly related to regional climate than soil texture. For more than half of the globe with mean annual precipitation < 500 mm/year, the probability of precipitation influencing dust emission is greater than 30 – 40%. Thus, rain-derived soil moisture in the active layer should not be ignored in models throughout much of the world’s dust producing regions.

Methods
A global map of sand, silt, and clay fractions was derived from the global WISE v. 3.1 30 x 30 arcsec database which is based on the Harmonized World Soil Database (Batjes, 2016). For each cell of this dataset, there are up to nine different soil components. For each cell, percent sand, silt and clay were derived by calculating the average of all available components, weighted by their fractional area in the cell. These average fractions were resampled to 0.25° x 0.25°, the same resolution as the precipitation data, and drying time was calculated using Equation 1. 

Ravi et al. (2006) conducted experiments on six soils that provide estimates of the time it takes the active layer to dry (i.e., for a wet surface to return come into equilibrium with ambient relative humidity). A parameterization for these drying times (DTs) for soils based on soil texture data created by regressing DT against fractions of sand, silt, and clay:

                DT (minutes) = 15.95 % Sand + 28.05 % Silt + 20.28 % Clay – 1494.          (Equation 1)       

This relationship fits the published values from Ravi et al. (2006) with R2 = 0.99.

A histogram was constructed for each cell (0.25° x 0.25°) of the length of time between each precipitation event (3B42RT > 0, GES DISC, 2016) and the next period during which the 10-m MERRA2 (Molod et al., 2015) wind exceeded 7 m/s (e.g. Marsham et al., 2011).  If the (3-hour) wind exceeded threshold during the same time step as the precipitation event, a 0 was recorded. If wind exceeded the threshold during the time step immediately following the rain event, a 1 was recorded. If wind exceeded the threshold during two timesteps following the rain event, a 2 was recorded, and so on. To account for the uncertainty of when the above-threshold wind occurred during the 3-hour TRMM period, Drying Time (DT) was also rounded down to the nearest 3-hour period (e.g., 4 hours rounded down to 3 hours) and up to the nearest 3-hour period (e.g., 4 hours rounded up to 6 hours). For the season with the highest DUP, the probability that a wind event occurs after a rain event but before the soil is dry, and can therefore suppress emission, is given by: Psuppress = P (Tinterval ≤ DT).

Batjes, N. H. (2016). Harmonized soil property values for broad-scale modelling (WISE30sec) with estimates of global soil carbon stocks. Geoderma, 269, 61-68. https://www.sciencedirect.com/science/article/pii/S0016706116300349

GES DISC. (2016). RMM (TMPA-RT) Near Real-Time Precipitation L3 1 day 0.25 degree x 0.25 degree V7. http://disc.gsfc.nasa.gov/datacollection/TRMM_3B42RT_Daily_7.html. Retrieved from: http://disc.gsfc.nasa.gov/datacollection/TRMM_3B42RT_Daily_7.html 

Marsham, J. H., Knippertz, P., Dixon, N. S., Parker, D. J., & Lister, G. M. S. (2011). The importance of the representation of deep convection for modeled dust-generating winds over West Africa during summer. Geophysical Research Letters, 38(16), L16803. http://dx.doi.org/10.1029/2011GL048368

Molod, A., Takacs, L., Suarez, M., & Bacmeister, J. (2015). Development of the GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2. Geosci. Model Dev., 8(5), 1339-1356.

Ravi, S., Zobeck, T. M., Over, T. M., Okin, G. S., & D'Odorico, P. (2006). On the effect of wet bonding forces in air-dry soils on threshold friction velocity of wind erosion. Sedimentology, 10.1111/j.1365-3091.2006.00775.x

活跃风成层（active aeolian layer，即土壤表层约2毫米的土层）中的土壤水分会通过提升沙尘排放阈值来影响沙尘排放过程，因此降水具备抑制沙尘排放的潜力。本研究旨在采用与全球及区域沙尘排放模型所用数据类似的再分析与卫星数据，计算前期降水使得活跃风成层保持湿润的时段内，强风事件发生的概率。研究结果表明，该问题的答案与区域气候的相关性要强于土壤质地。在全球超过一半的年平均降水量低于500毫米/年的区域，降水影响沙尘排放的概率高于30%~40%。因此，在全球多数沙尘源区的模型中，活跃风成层中由降水产生的土壤水分不应被忽视。

## 研究方法
全球砂粒、粉粒、黏粒含量分布图源自基于世界协调土壤数据库（Harmonized World Soil Database，Batjes, 2016）构建的全球WISE v3.1版本30角秒×30角秒数据库。该数据集的每个网格单元最多包含9种不同的土壤组分。针对每个网格单元，通过对所有可用土壤组分按其在单元内的面积占比加权平均，计算得到砂粒、粉粒、黏粒的百分含量。将得到的平均含量重采样至0.25°×0.25°的空间分辨率（与降水数据的分辨率一致），并通过公式1计算干燥时间。

Ravi等人（2006年）针对6种土壤开展了实验，估算了活跃风成层干燥所需的时间（即湿润表层恢复至与环境相对湿度平衡的时长）。基于土壤质地数据，通过对干燥时间（DT）与砂粒、粉粒、黏粒含量进行回归分析，构建了土壤干燥时间的参数化方案：
干燥时间（分钟）= 15.95×砂粒百分含量 + 28.05×粉粒百分含量 + 20.28×黏粒百分含量 – 1494。 （公式1）
该回归关系与Ravi等人（2006年）的公开实验数据拟合度极佳，决定系数（R²）达0.99。

针对每个0.25°×0.25°的网格单元，构建了降水事件（3B42RT＞0，GES DISC，2016）与下一次10米高度MERRA2再分析风速（Molod等人，2015年）超过7米/秒的时段之间的时间长度直方图（例如Marsham等人，2011年）。若降水事件发生的同时，3小时时间步长内的风速超过阈值，则记为0；若降水事件发生后的第一个时间步长内风速超过阈值，则记为1；若降水事件发生后的两个时间步长内风速超过阈值，则记为2，以此类推。为了考虑3小时TRMM时段内超标风速发生时刻的不确定性，干燥时间（DT）也分别向下取整至最近的3小时时段（例如4小时向下取整为3小时）和向上取整至最近的3小时时段（例如4小时向上取整为6小时）。在沙尘排放最活跃的季节（DUP最高的季节），降水事件后、土壤干燥前发生强风事件（从而可抑制沙尘排放）的概率由下式给出：
Psuppress = P (Tinterval ≤ DT)

## 参考文献
1. Batjes, N. H. (2016). 用于大尺度建模的协调土壤属性数值（WISE30sec）及全球土壤碳储量估算. *Geoderma*, 269, 61-68. https://www.sciencedirect.com/science/article/pii/S0016706116300349
2. GES数据信息中心（GES DISC）. (2016). RMM (TMPA-RT) 近实时降水L3产品，1日分辨率，0.25°×0.25°，V7版本. http://disc.gsfc.nasa.gov/datacollection/TRMM_3B42RT_Daily_7.html. 获取自：http://disc.gsfc.nasa.gov/datacollection/TRMM_3B42RT_Daily_7.html
3. Marsham, J. H., Knippertz, P., Dixon, N. S., Parker, D. J., & Lister, G. M. S. (2011). 夏季西非沙尘起风模拟中深对流表征的重要性. *Geophysical Research Letters*, 38(16), L16803. http://dx.doi.org/10.1029/2011GL048368
4. Molod, A., Takacs, L., Suarez, M., & Bacmeister, J. (2015). GEOS-5大气环流模式的研发：从MERRA到MERRA2的演进. *Geoscientific Model Development*, 8(5), 1339-1356.
5. Ravi, S., Zobeck, T. M., Over, T. M., Okin, G. S., & D'Odorico, P. (2006). 风干土壤中的湿粘结力对风蚀临界摩擦速度的影响. *Sedimentology*, 10.1111/j.1365-3091.2006.00775.x