The Effects of Soil Moisture on Precipitation

During the day, vegetation and soil moisture affect the absorption of the sun’s heat between soil and the atmosphere. When droughts occur, soil lacks enough water for evapotranspiration to occur at its full potential. With decreased evapotranspiration, the lower atmosphere becomes hotter and drier. Therefore, soil moisture is an important factor in the development of storms that return moisture to soil through feedback loops. However, there is uncertainty regarding how soil moisture affects storm formation globally, due to a lack of observed evidence and uncertainty in large-scale models. Taylor et al. conducted a worldwide observational analysis of the relationship between soil moisture and precipitation (2012). The authors were able to conclude that afternoon rain falls predominantly over soils that are relatively dry compared to the soil in the surrounding area over certain spatial scales and seasons. This finding is most obvious in semi-arid regions where fluxes in heat and humidity are most sensitive to soil moisture and there is frequent convection of heat and moisture between the soil and atmosphere. The authors found that their results indicate increased afternoon moisture caused by increased heat flux into the atmosphere over drier soils, as well as possible variability in soil moisture over a scale of 50-100 km. This study suggests no positive feedback loops of increased precipitation coupled with wetter soils over the spatial scale studied. In contrast, six state-of-the-art global weather and climate models predict positive feedback between soil moisture and precipitation. The authors largely attribute this discrepancy to excessive drought predictions in large-scale hydroclimatic models. —Hilary Haskell
Taylor, C., de Jeau, R., Guichard, F., Harris, P., Dorigo, W., 2012. Afternoon rain more likely over drier soils. Nature 489, 423-426.

Taylor et al. evaluated the response of daytime moisture convection between soil and the atmosphere to soil moisture anomalies. The authors used global observational data sets of both surface soil moisture and precipitation at a resolution of 0.25° x 0.25° on daily and 3-hourly time scales in order to analyze the location of afternoon rain events relative to soil moisture, before the rain events occurred. Soil moisture data was retrieved between 60°south and 60° north from the Advanced Microwave Scanning Radiometer for EOS and the MetOP Advanced Scatterometer.  The authors considered whether rain is more likely over soil that is either wetter or drier than other soil in the surrounding area. This methodology was also used in six global models of climate projection.
Soil moisture affects precipitation over a variety of time periods and geographical areas. When drought occurs, there is less evapotranspiration of water out of plant leaves into the atmosphere, causing overall atmospheric moisture content to decrease. Less atmospheric moisture leads to decreased precipitation. Anomalies in soil moisture disrupt atmospheric heating patterns, and thus synoptic-scale (large scale, usually 100 km or longer) atmospheric circulations and horizontal water vapor transport from the oceans. Small scale disruptions, such as regional precipitation and the formation of convective clouds due to atmospheric instability in heat and moisture content, can be affected by fluctuations in these factors throughout the day. Both convective clouds and rainfall occur due to the sun’s uneven heating of the earth’s surface and atmosphere, thus leading soil moisture to evaporate and turn to water vapor. When the water vapor rises and cools, it condenses back into liquid form in clouds, before eventually becoming heavy and falling in the form of precipitation. In an undisturbed atmosphere, surface feedback between soil and the atmosphere is determined by atmospheric temperature and humidity. Mesoscale (10-100vkm in length) variability in soil moisture can create feedback loops through daytime circulation of soil and atmospheric moisture.
            A variety of studies have indicated different correlations between soil moisture and precipitation. In Illinois and West Africa, studies found positive correlations, demonstrating a positive feedback between soil moisture and precipitation. Another study found increasing frequency of heavy convective rainfall coupled with high rates of evapotranspiration in North America. Sensible heat fluxes are the fluctuations of heat between the Earth’s surface and the atmosphere through conduction and convection at the boundary between the atmosphere and the soil. Satellite cloud data indicate an increase in afternoon precipitation frequency over areas with increased sensible heat fluxes from mesoscale circulations, caused by soil moisture or vegetation cover.
            Regionally, climate models consistently predict feedback between soil moisture and rainfall. Soil moisture limits evapotranspiration when convection occurs, due to the lack of available water in the soil for both processes. However, the strength of these feedback loops is less consistent, indicating that there is uncertainty in surface flux sensitivity to soil moisture and the impact of surface fluxes on the convection in the boundary layer between the atmosphere and soil. The authors found that whether the feedback relationship between soil moisture and precipitation is positive or negative depends on the model’s spatial resolution. This resolution is impacted by convective parameterization, which provides a quantitative measurement of atmospheric instability, a triggering mechanism for convection, the net effect of latent heating on the local atmosphere, and the amount of water vapor available in a convective process. This study suggests that smaller spatial scales produce different predictions of soil moisture and precipitation feedback patterns due to erroneous climate models and domination of large-scale atmospheric conditions.
            The authors focused on afternoon precipitation development, when sensitivity of convection of moisture and heat to land conditions is maximized. A precipitation event was defined as a 0.25 x 0.25 pixel location within a box measuring 5 x 5 pixels, and precipitation that exceeded 3 mm. To ensure that soil moisture measurement preceded rainfall, the authors excluded pixels with more than 1 mm of rain in the hours preceding the precipitation event. The authors also excluded locations with topographic heights above 300 m and regions containing water bodies or strong soil moisture gradients. Therefore, mountainous and coastal areas were not included due to their effects on mesoscale precipitation, or tropical forests due to the inability to retrieve soil moisture data beneath dense vegetation. Each maximum afternoon rainfall (L max) was paired with one or more pixels in the box where afternoon rainfall was at a minimum (L min). The authors then measured the difference in pre-rain-event soil moisture ∆Se by subtracting L min from L max, with an adjustment for the climatological mean soil moisture from both the control and experimental locations. The strength of the soil moisture’s effect on precipitation was quantified by a sample of precipitation events. The events were analyzed for predictability by comparing how unexpected the observed sample mean value of ∆Se was relative to a control sample, Sc, from the same location on non-rain-event days. The control sample was constructed from daily soil moisture differences between L max and L min, using data from the same month but non-rain-event areas. This was expressed as a percentile of typical values δe. Rain over drier soils is indicated by values less than ten, while rain over wetter soils is indicated by values greater than ten.
            The authors concluded that globally, 28.9% of the areas analyzed had percentile values less than ten (indicating rain over drier soils), compared to an expected frequency (which assumes no feedback between soil moisture and precipitation) of 10%. Lower percentiles are in semi-arid to arid regions, such as North Africa, Eastern Australia, Central Asia, and Southern Africa. This finding indicates that rain over drier soils is a common phenomenon. When computing the mean difference in soil moisture before rain events for both the experimental and control groups, this same outcome occurred with 99% statistical significance across all continents and their climate zones. However, upon repeating the analysis after lowering the spatial resolution from 0.25 ° to 1.0° degrees, this produced only about one-tenth of the number of events compared to the 0.25° data.  Still, for this repeated analysis, the authors found that afternoon rain over drier soils was more likely, and that this finding was statistically significant for parts of North Africa and Australia.
            According to Taylor et al., the satellite-driven data sets are subject to error in providing data for the rain events. However, analyzing data from many precipitation events should still yield accurate estimates of δe. Additionally, satellites are a valuable data source in that they are able to demonstrate the spatial structure of rainfall. This further substantiates the authors’ methodology by providing consistency in the spatial variability of soil moisture and rainfall studied in the independent data sets.
            After concluding that rain tends to fall over the drier soils within regions, Taylor et al. also studied whether drier soils are consistent with land surface feedback between soil and atmospheric moisture. For this to be the case, evapotranspiration must be limited by soil water deficit. The consistency in land surface feedback and drier soil is only present during some seasons and in regions where water stress coincides with the convective activity. Low percentiles of  δe occur in relatively dry areas, and are due largely to convective storms. The most negative values occurred in the driest mean conditions. This finding is consistent with soil moisture feedback, meaning that the sensitivity of heat fluxes (sensible and latent) to soil moisture increase as mean soil moisture decreases.
            For there to be a soil moisture feedback, there must be strong daily tendency for afternoon precipitation to fall over soils drier than other soils in the surrounding region. The authors repeated their analysis a third time and evaluated the onset of precipitation at varying times after soil moisture observations were taken at 1:30 P.M. local time. This reanalysis produced values of δe   that demonstrate a daily cycle, with the most negative values occurring during daytime (between 12:00 P.M. and 3:00P.M.) over dry soils and positive values during nighttime (between 9:00 P.M. and 3:00 A.M.) over wetter soils. The early afternoon minimum over drier soil is consistent with negative soil moisture feedback. This type of feedback occurs when convective instability, effects of surface properties on the planetary boundary layer between the soil and the atmosphere, and mesoscale flows are maximized. However, the mechanisms to explain the positive values at night time are less obvious. After dusk, day-time surface-induced flows and thermals, columns of rising 
air in the lower altitudes of the atmosphere created by the uneven heating of the Earth’s surface by the sun, do not persist. However, nocturnal humidity anomalies do persist and last longer depending on the spatial scale of surface features and wind conditions. At night, pre-existing, fast-moving convective systems are the most influential factors in producing positive night values
            The authors repeated their analysis a final time using 3-hourly diagnostics from six global models, ranging in resolution from 0.5° to 2.0°. In contrast to the previous analyses, this analysis found a strong preference for rain over wet soils for most parts of the world. Only one of the models produced more than the expected 10% results of with P < 10 (with more negative
δe valuesindicating more rainfall over drier soil). The P < 10 areas were also not distributed as globally as they were in the authors’ previous analyses, and were concentrated in the mid-latitudinal region. This analysis across the six global models produced the opposite result compared to the author’s previous analyses. Increased precipitation over wetter soils, especially in the tropics, indicates the failing of convective parameterizations to represent land feedbacks on daytime precipitation. This failure is largely attributable to the lag in the daily cycle of precipitation, with rainfall starting several hours too early. Furthermore, this failure is amplified by a lack of clouds within the boundary layer between soil and atmosphere in some models. Convective precipitation occurs rapidly and intensively over a limited area, due to unstable moisture and heat in the atmospheres, and is highly responsive to the daytime increase of moisture instability. This over sensitivity causes precipitation to occur faster over wetter soils, indicating a positive feedback cycle. When rainfall occurs prematurely in daily cycles, there are other daytime factors that contribute to convection of moisture into the atmosphere are limited in their effect.
            Still, Taylor et al.’s finding that afternoon rain over locally dry soils on scales of 50 -100 km is consistent with studies using remotely sensed data. The authors’ failure to find instances of positive feedback is an indication that surface-induced flows over 50-100 km are important in triggering convection. In addition, the authors cite mixing processes in the formation of convective clouds as an important contributing factor. However, neither of these factors is currently included in the one-dimensional analyses utilized in this study. The authors also consider whether models that use convective parameterizations represent the aforementioned processes adequately. The models analyzed in this study (HadGEM2, CNRM-CM5 and INMCM4) do not resolve soil moisture structures across 50-100 km or their impacts on triggering the convective heat and moisture flux between the atmosphere and soil. Still, all the models support the finding that rain tends to fall over wetter soils, despite the lack of observational data. The authors do not aim to conclude that soil moisture feedback is negative for temporal or spatial scales not analyzed in the study, such as those on large spatial scales. Taylor et al. postulate that the accumulation of moisture in the lower atmosphere from transpiration over land surfaces may provide more favorable dawn conditions for daytime convection of moisture into the atmosphere than the equivalent accumulation in a region affected by drought. Or, the large-scale response to soil moisture may dominate regional responses in some areas. The authors conclude that the sensitivity of convection patterns used in the six climate models used in this study is inaccurate, and contributes to the tendency of large-scale models to erroneously extend measures of drought periods and exaggerate soil moisture feedbacks in climate systems.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s