Seasonal and Regional Variability of Et over the Conterminous University of Maryland United States

According to Equations 15.1 and 15.2, variability of Et or Q, is extracted from the seasonal and regional variability of input terms: P and Qsw (1 - a). In this section, we discuss the roles and variability of net solar radiation, precipitation, land-cover type, and Et over the Conterminous United States (CONUS) through reviewing state-of-the-art observational datasets and numerical model simulation results from the North American Land Data Assimilation System (NLDAS).111 The NLDAS project constructs atmospheric forcing and LSM datasets from the best available observations and model outputs. Unlike in situ or satellite data, NLDAS can provide spatially and temporally complete datasets that satisfy energy and budget balances in Equations 15.1 and 15.2. The use of separate observational platforms for datasets of P, Et, and other terms can lead to inconsistencies and imbalances in the closure of the water and energy cycle equations. For this section, we have compiled a 30-year seasonal climatology (March/April/May (MAM); June/July/August (JJA); September/October/November (SON); December/January/ February (DJF)) from NLDAS Phase 2|21 datasets using precipitation observations and output from the Mosaic LSM from 1979 to 2009. It is important to note that NLDAS-2 precipitation is a state-of-the-art dataset over CONUS for such a long period, based primarily on a l/8,h-degree daily gridded analysis of rain gauge observations. This precipitation (and other NLDAS-2 surface meteorology, such as Qsw and QLk) is used as input to the LSM. The Mosaic LSM (like all LSMs) contains several parameterizations and assumptions in its model formulations to calculate Et, R, dW, Ql and QH. In the following sub-sections, we discuss the seasonal and regional variability of the key water and energy terms.

Precipitation and Solar Radiation

Surface precipitation, P, varies seasonally and regionally, characterized primarily by synoptic weather (i.e., large-scale circulation), mesoscale circulation (local disturbance), and thermodynamics (daytime heating). When precipitation reaches the surface, some water may be trapped on the canopy, stored on the surface of the soil (including as snow), be infiltrated into the soil, or run off to a river. Infiltrated moisture to the soil becomes the storage term, dW (Equation 15.1), that is the direct source of Et, through transpiration and/or soil evaporation.

Figure 15.1 shows a 30-year seasonal climatology of the precipitation from NLDAS-2. In the MAM, DJF, and SON periods, the coastal regions of the North West show a large amount of the precipitation generated by extratropical cyclones from the Pacific Ocean. The moisture flux from the north Pacific also generates snowfall or orographic rainfall over the mountains in the west. In the eastern domains, precipitation is peaked around the lower Mississippi Basin due to moisture flux from the Gulf of Mexico. Precipitation is higher along a narrow range of the northeast coastal region in the MAM and the SON periods. In the JJA period, the largest precipitation appears in the Deep South due to effects of sea breezes. Excepting winter, the South West tends to be drier for all seasons. The Great Plains become very dry during winter, and have light amounts of precipitation during other seasons.

Surface incoming solar radiation (Qsw) varies along with seasonal cycles. Its seasonal variability is, at the first order, governed by the mean solar zenith angle due to tilt of earth’s rotation axis relative to the sun. Its regional variability is affected by the presence of clouds and aerosols. Surface albedo (a: defined as a fraction of the incoming shortwave flux that is reflected by the surface) also affects the regional variability of the surface net solar radiation. Albedo differs widely depending on the land-cover type and vegetation senescence, ranging from 10% with dense forest to 50% over desert, but it could be up to 95% with the presence of fresh snow.

Figure 15.2 shows a 30-year seasonal climatology of the surface net solar radiation, Qsw (1 - a), from NLDAS-2 Mosaic. It is obvious that the largest net solar radiation is in the JJA period, whereas the lowest is in the DJF period, and is generally stratified in the latitudinal zones. This confirms that the largest factor for determining the net surface solar radiation is the mean solar zenith angle. Even in the same latitudinal zone, net solar radiation shows regional (longitudinal) variability at some extent, especially highlighted in the MAM period. For example, net solar radiation is ~40 W/m2 higher in California than in Nevada. In Colorado, the Rocky Mountain region has spotty regions with low net solar radiation. These types of regional variability are attributed to differences in albedo associated with the land-cover type and the presence of cloud/aerosols and surface snow.

(See color insert.) Thirty-year seasonal climatology (Mar 1979-Feb 2009) of surface precipitation in NLDAS Phase 2

FIGURE 15.1 (See color insert.) Thirty-year seasonal climatology (Mar 1979-Feb 2009) of surface precipitation in NLDAS Phase 2.

(See color insert.) Tliirty-year seasonal climatology (Mar 1979-Feb 2009) of surface net solar radiation from the Mosaic LSM in NLDAS Phase 2

FIGURE 15.2 (See color insert.) Tliirty-year seasonal climatology (Mar 1979-Feb 2009) of surface net solar radiation from the Mosaic LSM in NLDAS Phase 2.

Natural vegetation competes and evolves by taking the best advantage of solar radiation and rainfall patterns. Figure 15.3 shows the satellite-derived land-cover map used in NLDAS. Forest types extend over much of the eastern domain of the CONUS, coastal regions of the North West, and spotted over the Mountain region, where enough orographic rainfall sustains their life. Trees with a large canopy require more soil moisture to maintain their photosynthesis, transpiration, and mineral uptake. Some tree species adapt to the region where mean temperature significantly changes seasonally by becoming dormant during the cold seasons (e.g., deciduous forest). For areas with moderate climate throughout the seasons, tree species tend to be evergreen (e.g., evergreen forest). Grass and shrub species have smaller bodies, which require less number of resources to maintain. They can be dormant for long periods during drought and/or cold. Thus, they are well adapted for the dry regions, such as in the Great Plains, South West, and some of the Mountain regions. Cropland exists elsewhere in the CONUS, especially extended over the Midwest, lower Mississippi river basin, and the Great Plains. Agricultural practices have modified crop species, soil, and moisture sources for their climate and economy through a long-time history. Thus the distribution of cropland is not strictly regulated by the rainfall and net radiation patterns.

Transpiration and Physical Evaporation

For a given solar insolation and downwelling thermal radiation, surface total available energy is determined, balanced by energy release from the turbulent sensible heat flux (QH), the turbulent latent heat flux (QJ, and the thermal flux emitted from the surface (QLWemit). For further detail, these three terms in Equation 15.2 can be expressed as follows.

University of Maryland (UMD) dominant land-cover type map used in NLDAS Phase 2

FIGURE 15.3 University of Maryland (UMD) dominant land-cover type map used in NLDAS Phase 2.

For a plant-leaf level,

and for soil level,

where pMm is dry air density; X is latent heat of vaporization; air is water vapor mixing ratio of surrounding air; eaf) is saturated water vapor mixing ratio at leaf temperature (Tjeaf);£| is water vapor conductance (stomatal and leaf aerodynamic conductance for plant leaf, soil aerodynamic conductance for soil);gH is heat conductance (leaf or soil aerodynamic conductance); Cp is the specific heat of dry air, Talr is surrounding air temperature; e is emissivity; a is Stefan-Boltzmann constant; qxn] is water vapor mixing ratio of soil surface; Tsoil is soil surface temperature.

Et plays a primal role to determine the energy balance between QH, Ql> and QLWemit. For example, strong wind and high stomatal conductance provide the best environment for plants to transpire more water vapor from leaves to the surrounding air, and consequently less energy is available for sensible heat flux and thermal emission, which is strongly coupled with leaf temperature. Consequently, leaf temperature must be reduced until it balances the net radiation through Equation 15.3a. Large Et indicates larger net solar radiation and available soil moisture from frequent precipitation, thus linking to larger exchange of C02 for photosynthesis (plant production) and larger uptake of root-zone soil moisture associated with various minerals required for their growth. Alternatively, a plant may become highly stressed due to lack of available soil moisture or a lack of wind, and transpiration and associated latent heat flux will be restricted, which will quickly increase leaf temperature, enhancing sensible heat flux and thermal emission. If such condition is extended for a long term, plants start wilting and become dormant to survive in a severe environment.

Soil evaporation performs a similar way to transpiration; it controls soil skin temperature and sensible heat flux (Equation 15.3b). A significant difference between soil evaporation and plant transpiration is the depths of soil moisture. Soil evaporation is linked to top-soil moisture, whereas plant transpiration is linked to root-zone soil moisture (up to several meters in depth). During short-term drought situation, deep-root tree species can maintain high Et and QL with deep soil moisture, even if physical evaporation is limited from the soil surface. It should be also noted that Equation 15.3a will be scaled to the leaf area index. So Equation 15.3a dominates the Equation 15.3b for dense forests, and vice versa for a less vegetated area.

Figure 15.4 shows the seasonal Et from the 30-year NLDAS-2 Mosaic climatology. Overall patterns of seasonal and regional variability of Et can be extracted from those of precipitation (Figure 15.1) and net solar radiation (Figure 15.2). From MAM to DJF, the trend of CONUS-mean Et is well explained by that of net solar radiation; e.g., Et is largest in JJA, decreases from MAM to SON, and lowest in DJF. The same trend is shown in the surface net solar radiation. For a given net solar radiation, seasonal precipitation and landcover type explains the regional variability. In MAM, light Et (~l-2 mm/day) is spread over the Great Plains and West in general. Moderate Et (2-4 mm/day) appears in eastern domains of the CONUS and the North West Coast. The similar regional pattern continues in the JJA period, but Et over the eastern domains is ~1 mm/day larger than the MAM period. Also, from MAM to JJA, the peak of Et shifts from the South to Midwest cropland regions. In the SON period, minimal Et (<0.5 mm/day) is shown across the South West, and Et decreases approximately 2 mm/day over eastern domains. In the DJF period, the minimal Et region is spread widely over the West and Midwest, and small Et (0.5-2 mm/ day) remains in the Southeast and the West Coast.

Figure 15.5 shows a scatter plot between NLDAS-2 precipitation and Mosaic Et. All values are a 30-year seasonal climatology, each mark represents a single grid-point within the l/8lh-degree NLDAS grid

(See color insert.) Scatter plot between surface precipitation and Mosaic Et for the MAM, JJA, SON, and DJF periods in NLDAS Phase 2

FIGURE 15.5 (See color insert.) Scatter plot between surface precipitation and Mosaic Et for the MAM, JJA, SON, and DJF periods in NLDAS Phase 2. Each scattered point represents a grid point over CONUS, and all values are based on a 30-year seasonal climatology (Mar 1979-Feb 2009).

(52,476 marks for each season); it therefore correlates Figures 15.1-15.4. For all seasons, precipitation and Et are well correlated with each other; it suggests that precipitation variability characterizes the regional Et variability. Slopes between Et and precipitation (defined as Et/precipitation) become steeper in the order from DJF (~0.25), SON (~0.6), MAM (~1.0), to JJA (~1.25), which is in the same order to the seasonal variability of surface net solar radiation over the CONUS (Figure 15.2). This means that JJA (DJF) has the largest (smallest) energy that is used to recycle surface precipitation back to the atmosphere by Et. Also of note is the number of points where the Et is larger than precipitation, especially during JJA, as shown in Figure 15.5 and comparing Figures 15.1 and 15.4. This effect is caused by changes in the dW storage term, showing evidence of soil moisture and snow cover that fell as precipitation in previous seasons being converted to Et during a season when more energy is available. Overall, Figure 15.5 suggests that, at first order, variability of mean surface net solar radiation determines the seasonal variability of Et, whereas precipitation variability determines the regional variability of Et. Variability of land-cover (vegetation) type (Figure 15.3) and regional variability of net surface radiation (Figure 15.2), as well as soil moisture and snow cover storage also affect the spread of the scatter plots.

 
Source
< Prev   CONTENTS   Source   Next >