MM5/VIC Modeling Evaluation of the Influence of Antecedent Soil Moisture on Variability of the North American Monsoon System Chunmei Zhua, Yun Qianb, Ruby Leungb, David Gochisc, and Dennis P. Lettenmaiera aDepartment of Civil and Environmental Engineering Box 352700, University of Washington, Seattle, WA 98195 bPacific Northwest National Laboratory, Richland WA 99352 cNational Center for Atmospheric Research, PO Box 3000, Boulder, CO 80307
Winter Precipitation - Monsoon Rainfall Hypothesis to be tested by: (Zhu et al., J. Climate, 2005, 2007) Winter Precipitation - Monsoon Rainfall feedback hypothesis Higher (lower) winter precipitation & spring snowpack More (less) spring & early summer soil moisture Lower (higher) spring & early summer surface temperature Weak (strong) monsoon
Modification of coupled MM5/VIC modeling system by UW Links Docum Publicatios MM5-VIC coupled model system First coupled by Drs. Ruby Leung at PNNL and Xu Liang at University of California, Berkeley Precipitation Pressure Radiation Wind Humidity Air temperature Sensible heat flux Latent heat fluxes … Modification of coupled MM5/VIC modeling system by UW PNNL UW vegetation type: Single Multiple elevation band: Single Multiple Parameters: Soil, veg type dependent cell dependent initialization: Spin up 3 months Offline VIC
Domain MW 1 MS MM5/VIC model setup: Late Early Regions for which winter precipitation are related to summer monsoon in MW and MSa in Zhu et al 2005, 2007. 150*178 grid cells at 30km resolution in a Lambert-Conformal projection MW (eastern AZ and western NM) 1 NAMS Soil Moisture prescribing domain (1) MS (2) (northwestern Mexico) MM5/VIC model setup: The soil moisture prescribing domain is defined based on the winter precipitation related region from our previous papers on MW and MSa. Late Early ● Kain-Fritsch (KF) scheme ● Rapid Radiative Transfer Model (RRTM) long-wave scheme ● simple ice-explicit microphysics ● medium-range forecast (MRF) boundary layer scheme ● NCEP/NCAR Reanalysis LBC
Experimental Design ► 2 Initial soil moisture prescribed at ► The initial soil wetness condition on May 15 is a surrogate for previous winter precipitation condition. ► Control simulation s. moisture prescribed from offline VIC LDAS (3 mo spin-up, Mar-Apr-May). Initial soil moisture prescribed at Field capacity Wilting point Oct Sep Aug July June SM free running May 15 ► Simulations performed on wet and dry monsoon years to represent different atmospheric circulation conditions
Selection of wet/dry years: MW JJAS Precipitation 1984 1990 1995 1979 1973 1989 1984 1989 1995 1979 1993 Wet year: 1984 Dry year:1989 MSa JJAS Precipitation (dark) and Onset (gray)
Validation of coupled MM5/VIC modeling system 1984 wet year: Mean monthly daily precipitation Control Simulation Observation June July Aug Sep June July Aug Sep This and next slides compare the JJAS monthly precipitation spatial pattern over Mexico and most of the continental U.S. between model control run and observation for 1984 and 1989 respectively. The control simulation in 1984 captures the observed precipitation patterns and timing reasonably well in western Mexico and the U.S. Southwest (SW) especially in June and July, but overestimates SW precipitation in August and September. The precipitation in eastern Mexico and south-eastern Texas near the Gulf of Mexico are overestimated in July and this wet bias persists in eastern Mexico in August. The simulation missed the synoptic events in the eastern U.S. in August and in the Midwest in September, and produced too dry conditions there. The model simulated well the out-of-phase relationship between the precipitation over the SW and the Great Plains-northern tier and the in-phase relationship between precipitation over the SW and the eastern U.S. in July.
1989 dry year: Mean monthly daily precipitation Control Simulation Observation June July June July Aug Sep Aug Sep For the dry year 1989, the model initiates a much earlier monsoon than was observed, hence produces too much precipitation in northwestern Mexico. At the same time, it almost totally missed the heavy precipitation in the southeastern U.S. and the East Coast in June. July and August are the best simulated months excepting the SW in July (too much precipitation) and missed synoptic events in the Midwest in August. In September, the model overestimated precipitation in the SW and had a dry bias in the Great Plains and southern Mexico. In general, though, the core monsoon region of northwestern Mexico was well simulated from July to September, which provides some confidence for using 1989 as the dry year in our study of monsoon land surface feedbacks. MM5/VIC more aggressive In precipitating during ‘dry’ year
Positive Soil Moisture-Monsoon Rainfall Feedback ? mean monthly precipitation difference 1984-wet minus 1984-dry 1989-wet minus 1989-dry June July Aug Sep June July Aug Sep Several papers (Higgins et al. 1998, 2000, Hu and Feng 2002, 2004, Zhu et al 2005, 2007) have documented the existence of a statistically significant negative relationship between monsoon and previous winter precipitation in the SW and northwestern Mexico, even though this relationship is not robust all the time. Wet winters are expected to be followed by weak and late monsoons, and the reverse is true for dry winters. We expect to observe weaker monsoons for the “wet” experiments than for the control runs and “dry” simulations. Surprisingly though, the initial wet soil conditions seem to enhance the formation of monsoon precipitation in the model, and the dry conditions initial conditions do the reverse.
Winter Precipitation - Monsoon Rainfall The reverse of the proposed negative -- Winter Precipitation - Monsoon Rainfall feedback hypothesis 1 Higher (lower) winter precipitation & spring snowpack More (less) spring & early summer soil moisture 2 3 Lower (higher) spring & early summer surface temperature Weak (strong) monsoon Begin to examine 3 links……
Soil moisture differences between the wet and dry runs persist until mid-summer First Layer Third Layer June July Aug Sep June July Aug Sep 1984 The land surface memory process is the first link in our feedback hypothesis . The differences in modeled first layer soil moisture fields between “wet” and “dry” sensitivity experiments for 1984 and 1989, respectively. These two figures show that how the land surface memory in initial soil moisture persists. In particular, this anomaly in the extended NAMS region is retained until July and then disappears over most of NAM domain by August and September. 1989
Land surface memory – surface thermal conditions (1984) Difference maps between 1984-wet and 1984-dry runs June July First layer soil moisture Aug Sep June July + Aug Sep The strong soil moisture anomaly persistence shown above may exert an energy link to the atmosphere through its governing of the partitioning of solar radiation into latent and sensible fluxes, and hence modulation of the monsoon system, which is in fact the second link in the hypothesis. wet soil raises the latent heat flux throughout the season and reduces the sensible heat flux by nearly equal amounts for both 1984 and 1989, resulting in decreased surface skin temperature; the reverse is true for dry soil. In general, the surface temperature difference has a similar spatial pattern as the latent heat, but with an inverse relationship. The higher the latent heat, the cooler the surface, and vice versa. Latent heat June July -- Surface skin temperature Aug Sep
Difference maps between 1989-wet and 1989-dry runs June July First layer soil moisture June July Aug Sep + Aug Sep Similarly for 1989. Overall, the modeling results indicate that the land surface does have a memory of initial soil moisture, and this memory lasts over most of the monsoon domain until August. This land memory has an inverse relationship with the surface thermal condition over the NAMS domain and adjacent areas. June July Latent heat -- Surface skin temperature Aug Sep
Larger Thermal contrast– stronger monsoon ? Larger Thermal contrast– stronger monsoon Difference map between 1984-wet and 1984-dry runs: June July Aug Sep Monthly mean surface skin temperature June July Aug Sep Seasonal changes in the thermal contrast between land and adjacent oceanic regions are a cornerstone of the conceptual basis for understanding the onset of NAMS. According to this widely accepted land-sea thermal contrast theory, we expect that the “wet” simulation will produce a dry monsoon and/or late onset and the “dry” simulation will generate a wet monsoon with early onset because the “wet” runs always produce cooler surface temperature than “dry” runs over the NAMS region for both the wet and dry years. But surprisingly, the opposite is true, i.e. the strong (weak) monsoon is characterized by cooler (warmer) surface temperatures over SW and northwestern Mexico both before and during the monsoon season. Monthly mean precipitation
? Southwest surface heat low – monsoon strength Difference maps between 1984-wet and 1984-dry runs 925mb Geopotential Height Surface Skin Temperature June July Aug Sep June July Aug Sep The local surface pressure rises due to the increasing geopotential height at lower levels and the decreasing geopotential height at higher levels. This weakens the Southwest surface heat low – a very important component for initiating the monsoon. These results confirm that the strength of the heat low and intensity of the thermal contrast don't necessarily have positive relationships with monsoon strength (Kanamitsu and Mo 2003). June July In MM5-VIC increased local surface pressure weakens the Southwest surface heat low, but is related to the stronger monsoon? 500mb Geopotential Height Aug Sep
Monthly mean 925 mb meridional moisture flux (QV) averaged over longitude (107-113 oW) at 32 oN June July August September 1984-Wet 0.0047 0.0133 0.0150 0.0090 1984-Dry 0.0020 0.0134 0.0101 0.0088 1989-Wet 0.0054 0.0053 0.0033 0.0100 1989-Dry 0.0030 0.0036 0.0087 The impact of the thermal low on the monsoon is through its influence on low level large-scale circulation, since it is related to the moisture flux into the heated land from the Gulf of California. In order to roughly describe the magnitude of the moisture flux from GOC, we calculated the cross section of the meridional moisture flux (qv) at 925 mb averaged over 107o – 113o W near the entry point of the moisture flux into AZNM at 32N) as defined in Kanamitsu and Mo (2003). We choose the 925 mb level for the calculation to represent the low-level circulation which provides the major moisture contribution to the monsoon. Table shows that the average moisture flux is stronger in response to the prescribed initial higher soil moisture in 1984 across the monsoon season than for the “dry” runs. This preliminary analysis suggests that the weakening of the thermal low in response to the colder surface temperature, induced by the wetter soil, changes the low level circulation, and results in greater moisture flux into the interior of the NAMS region. Even though the SW thermal low is essential for the NAMS, its relationship with monsoon intensity is not as straightforward as is suggested by the thermal contrast monsoon driving concept, and there seem to be some other factors that determine the moisture flux into the region and in turn the monsoon strength. This is understandable because the strong low-level northerly surge of moist tropical air along the GOC is produced when a midlatitude trough is in the proper phase relationship with a tropical easterly wave (Stensrud et al. 1997). Weakening of the thermal low in MM5/Vic sims results in greater moisture flux into the interior of the NAMS region, likely from increased moisture availability due to increased regional evaporation instead of increased low level winds
Shallower Boundary Layer Boundary layer height difference between 1984-wet and 1984-dry runs Wet soil moisture conditions reduce the depth of the boundary layer, therefore increase the boundary layer moist static energy and the frequency and magnitude of rainfall from local convective storms. June July Aug Aug Monthly mean planetary boundary layer height (PBL) in the NAMS domain Besides the influence of the large-scale circulation, local land-atmosphere interactions may also play an important role in the observed precipitation enhancement. Eltahir (1998) and Zheng and Eltahir (1998) argue that wet soil moisture conditions increases the total heat flux from the surface (also confirmed by Small and Kurc 2003) and reduce the depth of the boundary layer, therefore increasing the boundary layer moist static energy. At local scale, this larger moist static energy increases the frequency and magnitude of rainfall from local convective storms. In large scale, the increase of the vertical gradient of moist static energy results in strengthening of the large-scale circulation, resulting in more rainfall. The warmer surface temperature produces a deeper boundary layer; the reverse is true for colder surface. Table also shows that the boundary layer height decreases substantially by about 500, 400, 300 and 100 m in June, July, August and September in response to the wetter soil, colder land surface, and smaller sensible heat flux over NAMS domain. Further investigation of the details of these mechanisms is needed. June July August September 1984-Wet 1354.2 1165.5 1060.9 1386.8 1984-Dry 1812.9 1552.1 1377.3 1436.5 1989-Wet 1704.1 1252.1 1332.0 1143.3 1989-Dry 2148.8 1647.3 1594.8 1264.0 local land-atmosphere interaction
Large-scale circulation or local land-atmosphere interaction ? Changes moisture convergence and precipitation. Changes the surface pressure and the flow field Meehl G. A., 1994: J. Climate shallower boundary layer Increased convective instability and potential for precipitation Meehl (1994) noted that for the south Asian monsoon the soil moisture – rainfall feedback could be either positive or negative. Two competing effects exist with respect to the soil moisture feedback – local enhancement of precipitation which leads to increased soil moisture, and a regional decrease of monsoon precipitation which leads to decreased land – sea temperature contrast and weaker monsoon flow according to the thermal contrast monsoon driving concept. So the impact of soil moisture on the precipitation could be local and direct or nonlocal and indirect . The wet soil could increase latent heat and reduce the sensible heat at the surface, leading to colder surface temperatures. This tends to build up a comparatively shallow boundary layer and provides a source of convective instability, increasing the potential for convective precipitation (Schar et al. 1999). Also, elevated soil moisture induces greater evaporation which in turn provides an increased moisture source for enhanced precipitation, leading to a strong monsoon, further increases in soil moisture, and so on (Meehl 1994). On the other hand, the lower surface temperature induced by increased evaporation changes the surface pressure and consequently the flow field, which changes moisture convergence and accordingly precipitation. Schar C et al. 1999: J.Climate Mo K. C. and H. H. Juang, 2003: J. Geophy. Res
Summary and Conclusions ● The MM5-VIC control sims reproduce reasonable monsoon precipitation for 1984 and 1989 over northwestern Mexico (1989 somewhat wet vs. obs) ● The model land surface has memory of the initial soil wetness that lasts for several months (until August). This land memory has a negative relationship with surface thermal conditions over the NAMS domain and its larger adjacent area. ● In contrast to the original hypothesis, the wet year 1984 and dry year 1989 experiments exhibit similar positive soil moisture – rainfall feedbacks over the NAMS domain. In essence, it appears that local-regional recycling of moisture dominates in sustaining increased precipitation in the model. However magnitude of imposed anomaly likely imparts excessive influence. ● In nature, both the large-scale circulation changes and local land-atmospheric interactions in response to soil moisture conditions likely play important roles in the soil moisture – monsoon precipitation feedback. The symbiosis of these features needs to be studied in more detail.
Limitations of the experiments Extreme wet and dry soil conditions in the sensitivity experiments extreme surface temperature anomalies exaggerated surface low (not the optimal strength and location to start monsoon) very intense local evaporation Contribute to apparent positive soil moisture – rainfall feedback Future Work The extreme wet and dry soil conditions in the sensitivity experiments produce extreme surface temperature anomalies, and in turn an exaggerated surface low, which may not be the optimal strength and location to start the monsoon. Also, under these extreme conditions, local evaporation become very intense, which normally enhances monsoon precipitation as noted in the next paragraph, which may be a major cause of the apparent positive soil moisture – rainfall feedback. Further research is needed to explore the relationship between antecedent soil moisture and monsoon rainfall under less extreme soil conditions over SW, and to identify the relative importance of large-scale circulation and local evaporation. Explore the relationship between antecedent soil moisture and monsoon rainfall under less extreme soil conditions, and to identify the relative importance of large-scale circulation and local evaporation.
Monthly means of energy components in the NAMS region (2) (1) SM1 Tgrd LH SH June 1984-Wet 0.262 302.8 83.8 61.8 1984-Dry 0.229 306.2 57.4 88.4 1989-Wet 0.240 304.2 78.4 65.4 1989-Dry 0.182 307.7 47.6 94.0 July 0.281 301.6 85.2 51.4 0.258 305.4 66.1 72.0 0.273 303.8 85.4 58.2 0.257 307.5 66.5 74.9 August 0.282 300.2 85.9 44.3 0.266 302.6 78.8 56.1 0.265 77.9 55.5 305.0 67.1 65.2 September 0.259 299.2 86.7 35.7 0.260 299.5 83.6 37.4 0.280 298.8 68.6 37.5 0.267 299.6 40.3 Wet soil raises the latent heat and reduces the sensible heat by nearly equal amounts, resulting in decreased surface skin temperature Wet soil raises the latent heat flux throughout the season and reduces the sensible heat flux by nearly equal amounts for both 1984 and 1989, resulting in decreased surface skin temperature; the reverse is true for dry soil.