emission using the data assimilation results of their regional model. RC4. Their simulated results (without assimilation) also exhibited an underestimated dust ...
SOLA, 2011, Vol. 7A, 036−039, doi:10.2151/sola.7A-010
36
The Effects of Snow Cover and Soil Moisture on Asian Dust: I. A Numerical Sensitivity Study Taichu Y. Tanaka, Tsuyoshi T. Sekiyama, Takashi Maki, and Masao Mikami Meteorological Research Institute, Tsukuba, Japan
Abstract Semi-arid regions of East Asian dust source areas are frequently covered by snow. We conducted a sensitivity study of East Asian dust storms to investigate the effects of snow cover and soil moisture with a global aerosol model (MASINGAR). The simulated dust concentration greatly underestimated the dust event in late March 2007 relative to the observed PM10 concentration. When the effect of soil moisture is not taken into consideration, the simulated total dust emission amount is almost doubled, and the simulated dust concentration is comparable or overestimated with regard to the PM10 observations. The result of the sensitivity study suggests that the underestimation of the dust event is due to excessive soil moisture, which suppresses the dust emission. In contrast, the simulated dust concentration of the control experiment is in agreement with the observed PM10 concentration in May, suggesting that the effects of snow cover and soil moisture on the dust event in May were very small. To improve our ability to forecast Asian dust events in March, the treatment of the hydrological cycles of snow in the land surface model and the soil moisture dependence of dust emission flux should be regarded as the key factors.
1. Introduction Mineral dust aerosol has multiple roles in the climate system and the atmospheric environment. In East Asia, dust aerosol is mainly generated by dust storms when extratropical cyclones pass over arid and semi-arid regions. The dust storms may cause severe air quality hazards because the dust source areas are located adjacent to densely populated areas, such as the coastal regions of China, Korea and Japan. Because of the potential risks of dust and sand storms, Japan, China and South Korea have agreed to conduct joint research on dust and sandstorms (DSS) based on the agreement of the 8th Tripartite Environment Ministers Meeting (TEMM) (Ministry of the Environment of Japan 2008). Recently, the meteorological and environmental agencies of these countries have been providing information on Asian dust events based on numerical simulations. However, the numerical dust models still have major uncertainties, especially regarding the dust emission process (Uno et al. 2006). The outbreak of the dust particles is controlled by the meteorological conditions as well as by a number of land surface conditions, such as soil moisture, land use, vegetation, snow cover, and soil particle size distribution. The land surface conditions of the dust source regions in East Asia are complex, because of the heterogeneity of land use and seasonally varying obstacles such as snow cover and vegetation. A characteristic feature of the dust source areas in East Asia is that semi-arid regions of them are frequently covered by snow because they are located at relatively high latitudes. Snow cover decreases the emission of dust through two processes. First, the snow directly suppresses dust emission when the snow covers the potential dust source areas. Second, when the snow melts, it increases the soil moisture and, consequently, supCorresponding author: Taichu Y. Tanaka, Meteorological Research Institute, 1-1 Nagamine, Tsukuba, Ibaraki 305-0052, Japan. E-mail: yatanaka@ mri-jma.go.jp. ©2011, the Meteorological Society of Japan.
presses dust outbreaks. From their statistical study in East Asia, Kurosaki and Mikami (2004) demonstrated that the snow cover affects the threshold wind velocity of dust outbreaks. Mukai et al. (2003) suggested that the early spring snow is an influential factor in dust emission, and soil moisture impacts on the seasonal variation in dust emission from their long-term simulation. However, the impacts of snow cover and soil moisture on numerical dust prediction have not yet been discussed in detail. We conducted a sensitivity study of dust storms in East Asia to investigate the effects of snow cover and soil moisture using our global aerosol model. We discuss the effects of snow cover and soil moisture on the strength of a large-scale East Asian dust event that occurred in March 2007.
2. Method We used a global aerosol transport model referred to as the Model of Aerosol Species IN the Global Atmosphere (MASINGAR), which was coupled with an atmospheric general circulation model (AGCM) MRI/JMA 98. MASINGAR treats dust aerosols in ten size classes, ranging from 0.2 to 20 μm in diameter, that are transported independently and assumed to be non-interacting. The emission flux of dust is calculated from the friction velocity, vegetation cover, snow cover, land use type and soil type. Tanaka and Chiba (2005) described and validated the model in detail using available observations. The model resolutions were set to a T106 Gaussian horizontal grid (about 1.125° × 1.125°) and 30 vertical layers from the surface to a height of 0.4 hPa. An earlier version of this model has been used as the operational dust forecasting model by the Japan Meteorological Agency (JMA) since January 2004. The surface hydrological variable such as snow cover and soil moisture were prognostically calculated in the 3-layer Simple Biosphere model (SiB), which is the land-surface model used in the AGCM. Supplemental Fig. S1 displays the simulated snow cover distributions in March 2007, which is reasonably consistent with satellite observations. Snow on bare ground directly suppresses dust emission. The dust-emission flux is assumed to be linearly reduced with the areal fraction of snow cover As in the grid. Thus, dust flux F is calculated by: F = (1 − As ) Fno snow ,
(1)
where Fno snow is the dust-emission flux if the snow cover fraction is zero. Soil moisture increases the cohesive forces among soil particles and thereby increases the threshold friction velocity (u*t) of the soil particles. The parameterization of Fécan et al. (1999) was applied to calculate the effect of soil moisture on the threshold friction velocity as u*t = fw u*t dry , where fw is the soil moisture factor, and u*t dry is the threshold friction velocity under dry conditions. The soil moisture factor fw is parameterized as: 1 w £ wr ïì f w = ïí , ïïî 1 + a[100( w - wr )]b w > wr
(2)
where w and wr are the gravimetric soil water content and the threshold gravimetric soil water content is kg kg−1, respectively. The parameters a and b are empirical constants and were set to values of 1.21 and 0.68, respectively. The threshold gravimetric soil water content wr is calculated as wr = 0.17Mclay + 0.14Mclay2,
37
Tanaka et al., Impact of Snow Cover and Soil Moisture on Asian Dust: I. Sensitivity Study
where Mclay is the mass fraction of clay in the soil. We performed a series of simulations from 1 March to 31 May 2007, with a five-month spin-up run. During the period from March to May 2007, several large dust events occurred (e.g., Yumimoto et al. 2008; Sekiyama et al. 2010; Sugimoto et al. 2010). In late March 2007, an intense dust storm occurred around the Gobi desert, and the dust aerosol caused severe dust phenomena in the coastal regions of China, Korea, and Japan. Our study focused on this large-scale dust event. To elucidate the effects of the snow cover and the soil moisture, we conducted simulations using three sets of conditions. Hereafter, we refer to the default setup of the model simulation as the “control” run, the experiment with the soil moisture factor fw = 1 in Eq. (2) as the “fw1” run, and the experiment with snow cover As = 0 in Eq. (1) and the soil moisture factor fw = 1 in Eq. (2) as the “As0fw1” run. The difference between the fw1 and the control run indicates the effect of soil moisture, and the difference between the As0fw1 and the fw1 run represents the direct effect of snow cover on dust emission. To obtain realistic atmospheric fields, the horizontal wind fields were assimilated with the six-hourly data of the JMA global optimal analysis, while the sea-surface temperature and the ice data were taken from the JRA-25/JCDAS reanalysis (Onogi et al. 2007). Data assimilation of the aerosol concentration (Sekiyama et al. 2010) was not applied in these experiments. The radiative effect of the aerosol is not fed back into the dynamics of this simulation as different dust distributions produce different heating rates, resulting in different atmospheric fields, even if the assimilation is enforced. We compared our simulated dust concentrations with the surface PM10 observational data shared by the TEMM joint research project. For this paper, the data were taken from the Acid Deposition Research Center (ADORC), the Korea Meteorological Agency (KMA), and the Korea National Institute of Environmental Research (NIER).
3. Results and discussion Figure 1 shows the time series of simulated dust concentration for the sensitivity study and the observed PM10 concentrations at selected Korean and Japanese stations for dust events that occurred from late March to early April (Figs. 1a to 1d) and in late May for 2007 (Figs. 1e and 1f). In the late March 2007 dust event, the simulated dust concentrations with the control run were greatly underestimated relative to the PM10 observations. Other modeling studies (Yumimoto et al. 2008; Sugimoto et al. 2010; Maki et al. 2011) have also reported an underestimation of the dust
concentration for this dust event. However, the As0fw1 and fw1 runs exhibited comparable or overestimated dust concentrations. On the other hand, the As0fw1 and fw1 runs overestimated the dust concentration on 25 March (Figs. 1a to 1c). This suggests that the dust concentration in late March is very sensitive to the soil moisture in dust emission region. The differences between Asfw1 and fw1 runs are small, which indicate the effect of snow cover on dust concentration is small. In contrast, in the late May 2007 dust event case (Figs. 1e and 1f), all of the experiments indicate agreement or slight overestimation relative to the observations. Figures 1e and 1f show that the effect of soil moisture is small, and the effect of snow cover is negligible on the dust event in May case. This suggests that the snow cover and the soil moisture were not the dominant controlling factors of Asian dust in the May 2007 event. Figure 2a shows the horizontal distribution of the dust emission flux from 26 to 31 March 2007 for the control run, and Fig. 2b shows the difference between the control run and the fw1 run. The difference between these runs indicates that soil moisture reduces the simulated dust emissions over Mongolia and Inner Mongolia. The direct effect of snow cover to the dust emission, which is displayed as the difference between As0fw1 run and fw1 run (Fig. 2c), is small during this dust event. The dominant dust emission region, the Gobi and the Taklimakan deserts, are not affected by the snow cover or soil moisture during this event. Figure 2d shows the distribution of the surface synoptic observation (SYNOP) stations that reported the dust emission phenomena (weather report with 7−9 and 30−35) confirming that dust storm events were observed over the vast areas of Mongolia and Inner Mongolia. The total simulated dust emission amounts from the eastern China (90°E−130°E) during this dust event (28 to 31 March 2007) were 30 Tg, 49 Tg and 51 Tg with the control run, fw1 run, and As0fw1 run, respectively. Because these regions are in close proximity to Korea and Japan, the accuracy of the dust emission flux over these regions may have a greater impact on the dust concentrations at the stations presented in Fig. 1. Figures 3a and 3b show the simulated distributions of the snow cover and soil moisture on 29 March 2007 when the dust storm event occurred. In the dust storm areas of Mongolia and Inner Mongolia depicted in Fig. 2d, the snow cover had almost disappeared even though the soil moisture content was sufficiently high to suppress the dust emission in the model. Figure 3c shows the snow depth, soil moisture and precipitation at a represented grid in Inner Mongolia in March. The snow cover remained until mid-March, and as the snow melts, the soil moisture content increases. The soil moisture decreased with time, and keeps at a low level after 18 Mar. Then the soil moisture
Fig. 1. A comparison of the observed PM10 (circles with dash) and simulated surface dust concentrations at selected sites in spring 2007 (unit: μg m−3). (a) Seoul, (b) Busan, (c) Banryu, and (d) Hedo: dust event in late March 2007. (e) Busan and (f) Ganghwa: dust event in late May 2007. The green, red, and blue lines indicate the control run, fw1, and As0fw1 run, respectively, and the light blue line with circles indicates the PM10 observations.
SOLA, 2011, Vol. 7A, 036−039, doi:10.2151/sola.7A-010
38
Fig. 2. Simulated dust emission distribution over East Asia from 28 to 31 March 2007. (a) Control run (b) Difference between fw1 and the control run. (c) Difference between the As0fw1 run and the fw1 run. (d) Synoptic observation of the dust emission events (weather report with 7−9 and 30−35) from 28 to 31 March 2007 (Yellow, blue, green and red symbols correspond to the stations that reported dust events on 28, 29, 30, and 31 March 2007, respectively).
increased as the result of the weak precipitation on 28 March 2007 due to the cyclone that caused the dust storm. However, the observations for the dust storms (Fig. 2d) imply that the soil in Mongolia and Inner Mongolia is preferable for dust emission, suggesting that the soil moisture content is overestimated for late March. The distribution of snow cover on 31 March (Fig. S1d in Supplement) shows overestimation of snow in Mongolia and Inner Mongolia, suggesting a possibility of the overestimation of the precipitation with the dust storm. This result implies that the underestimation of the dust storm event by the control run is primarily attributed to the increase of soil moisture due to the precipitation. Figure 3d compares the simulated dust emission in Inner Mongolia in each sensitivity experiment. The difference between As0fw1 and fw1 is large in the early March but it is less
in late March, which suggests the snow cover suppresses dust emission during the early March but the effect is small in late March. In late March, the dust emission difference between the control run and both fw1 is very large, which indicates that the dust emission in Inner Mongolia is very sensitive to the treatment of soil moisture. Generally, land surface models coupled with large-scale atmospheric models have difficulties in treating the snow distribution because of their coarse resolutions that cannot treat small-scale topographical characteristics and vertical structures of the soil. In addition, treatments of the physical processes of snow, especially snow albedo, are empirical parameters with large uncertainties (e.g., Aoki et al. 2003). The results suggest that the dust prediction results in the snow-melting season are highly sensitive to inac-
Fig. 3. Simulated snow cover and soil moisture over China and Mongolia. (a) Snow cover on 29 March 2007. (b) Soil water content on 29 March 2007. (c) Temporal change of snow depth, precipitation, and soil moisture in Inner Mongolia. (d) Simulated dust emission from Inner Mongolia in March 2007.
39
Tanaka et al., Impact of Snow Cover and Soil Moisture on Asian Dust: I. Sensitivity Study
curacies in the soil moisture predicted by the land surface models. Another possibility for the uncertainty of dust is the soil moisture dependence on the threshold friction velocity (Eq. 2). The relationship between the soil moisture and the dust emission should be reexamined for the conditions prevailing in Mongolia and Inner Mongolia. Sugimoto et al. (2010) obtained a similar simulated dust emission bias, and discussed the effect of vegetation growth on dust emission using the data assimilation results of their regional model RC4. Their simulated results (without assimilation) also exhibited an underestimated dust emission outcome for late March and an overestimated dust emission outcome for late May, and they attributed these disagreements to the vegetation growth in comparison with the vegetation growth map of Mongolia. However, they ignored the effect of soil moisture because their simulated results did not depend on the soil moisture. The discrepancy between our results and those of Sugimoto et al. (2010) may arise from the difference in the land surface schemes. Our result suggests that the underestimation of the dust in March may be explained by the inaccurate soil moisture. Our results also exhibit a slight overestimation of the dust concentration in May. As Sugimoto et al. (2010) discussed, this overestimation of dust in May was possibly caused by the uncertainty in vegetation growth.
4. Conclusions We conducted a sensitivity study of the effects of the snow cover and soil moisture on the Asian dust event in spring 2007 using a global aerosol model, MASINGAR. From the comparisons with the PM10 observation, the control run underestimated the dust concentrations at the Korean and the Japanese stations in March but overestimated them in May. The results of the sensitivity study suggest that the numerical prediction of the Asian dust event in March strongly depends on the soil moisture content, especially in Mongolia and Inner Mongolia. On the late March dust storm event, the effect of snow cover on the dust emission was only marginal. Further understanding of the role of snow cover and its melt water, we should investigate dust storm events in early March, and statistical analysis will be required. To improve the accuracy of the Asian dust simulation in March, the treatment of the hydrological cycles of snow in the land surface models and the soil moisture dependence of dust emission flux were identified as the key factors that should be further investigated. Based on this sensitivity study, some strategies are suggested for the improvement of the prediction of the Asian dust event in March. First, the effects of soil moisture on sand and dust flux (such as the parameterization of Fécan et al. 1999) should be reexamined and improved. Second, to express the subgrid heterogeneity of snow cover and soil moisture, the introduction of a more sophisticated subgrid scale treatment in the land surface model, such as the mosaic approach, would improve the snow cover and soil moisture and, thus, improve the predicted dust emission flux in March. Third, the incorporation of near-real-time observational data of snow cover and soil moisture will reduce the uncertainties in the prediction of dust emission. To achieve this incorporation, the four-dimensional data assimilation of the dust emission together with the snow cover and the soil moisture would improve the dust simulation for the month of March (Sekiyama et al. submitted to SOLA).
Acknowledgments This research was partly supported by the Environmental Research and Technology Development Fund (B-0901) of the
Ministry of the Environment of Japan. This research was also partially supported by the Grant-in-Aid for Scientific Research (20244078) of the Ministry of Education, Culture, Sports, Science and Technology, Japan. This research paper contributes to the Joint Research of Dust and Sand Storms under TEMM-WG1.
Supplement Fig. S1. Comparison of the simulated snow cover with satellite observational product of Interactive Multisensor Snow and Ice Mapping System (IMS) (http://www.natice.noaa. gov/ims/). Fig. S2. Monthly mean distribution of the gravimetric soil moisture.
References Aoki, Te., A. Hachikubo, and M. Hori, 2003: Effects of snow physical parameters on shortwave broadband albedos. J. Geophys. Res., 108, 4616, doi:10.1029/2003JD003506. Fécan, F., B. Marticorena, and G. Bergametti, 1999: Parameterization of the increase of the aeolian erosion threshold wind friction velocity due to soil moisture for arid and semi-arid areas. Annales Geophysicae, 17, 149−157. Kurosaki, Y., and M. Mikami, 2004: Effect of snow cover on threshold wind velocity of dust outbreak. Geophys. Res. Lett., 31, L03106, doi:10.1029/2003GL018632. Maki, T., T. Y. Tanaka, T. T. Sekiyama, and M. Mikami, 2010: The impact of ground-based observations on the inverse technique of Aeolian dust aerosol. SOLA, 7A, 21−24. Ministry of the Environment of Japan, 2008: Dust and Sandstorms, 2nd version, http://www.env.go.jp/en/earth/dss/pamph/pdf/ full.pdf (last accessed 15 Nov. 2010). Mukai, M., T. Nakajima, and T. Takemura, 2004: A study of longterm trends in mineral dust aerosol distributions in Asia using a general circulation model. J. Geophys. Res., 109, D19204, doi:10.1029/2003JD004270. Onogi, K., and co-authors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369−432. Sekiyama, T. T., T. Y. Tanaka, A. Shimizu, and T. Miyoshi, 2010: Data assimilation of CALIPSO aerosol observations. Atmos. Chem. Phys., 10, 39−49. Sekiyama, T. T., T. Y. Tanaka, T. Maki, and M. Mikami, 2010: The effects of snow cover and soil moisture on Asian dust: II. emission correction in lidar data assimilation. SOLA, (submitted). Sugimoto, N., Y. Hara, K. Yumimoto, I. Uno, M. Nishikawa, and J. Dulam, 2010: Dust emission estimated with an assimilated dust transport model using lidar network data and vegetation growth in the Gobi desert in Mongolia. SOLA, 6, 125−128. Tanaka, T. Y., and M. Chiba, 2005: Global simulation of dust aerosol with a chemical transport model, MASINGAR. J. Meteor. Soc. Japan, 83A, 255−278. Uno, I., and co-authors, 2006: Dust model intercomparison (DMIP) study over Asia: Overview. J. Geophys. Res., 111, D12213, doi:10.1029/2005JD006575. Yumimoto, K., I. Uno, N. Sugimoto, A. Shimizu, Z. Liu, and D. M. Winker, 2008: Adjoint inversion modeling of Asian dust emission using lidar observations. Atmos. Chem. Phys., 8, 2869−2884. Manuscript received 15 January 2011, accepted 20 March 2011 SOLA: http://www. jstage. jst. go. jp/browse/sola