1. Introduction
Warm-season rainfall in the central United States can be broadly categorized into two types with distinct characteristics: rainfall associated with mesoscale convective systems (MCSs) and rainfall resulting from other rain generation mechanisms (non-MCS). MCSs are organized clusters consisting of convective cores and large stratiform rain areas, spanning more than 100 km in length. MCSs can persist for many hours, which is well beyond the lifetime of isolated convective storms. MCS rainfall contributes 30%–70% of warm-season rainfall in the central United States (Fritsch et al. 1986; Haberlie and Ashley 2019) while non-MCS rainfall produced by isolated convective cells and nonconvective stratiform clouds accounts for the remaining amount. During spring, a synoptically favorable environment supports the development of MCSs but in summer, MCSs can develop even in the presence of weaker dynamical forcing, so MCSs are less predictable in summer compared to spring (Song et al. 2019; Feng et al. 2019).
Rainfall produced by MCS and non-MCS storms has drastically different characteristics. Because of the larger storm size and longer lifetime, MCS rainfall is more concentrated in time and space compared to non-MCS rainfall, which is more scattered spatiotemporally. Observations show that MCS rainfall is ~7 times more intense than non-MCS rainfall, but it occurs 3–5 times less frequently in space and ~2 times less frequently in time (Hu et al. 2020). MCS and non-MCS rainfall also displays different diurnal cycles, with MCS rainfall peaking in nocturnal hours and non-MCS rainfall peaking in the late afternoon (Maddox 1980; Houze et al. 1990; Feng et al. 2019). The diurnal cycle of MCSs is influenced by their tendency to propagate eastward from the Rocky Mountain foothills to the Appalachians (Laing and Fritsch 1997; Houze 2004; Ashley et al. 2003; Feng et al. 2019).
Given the contrasting characteristics in intensity, frequency, and diurnal cycle of MCS and non-MCS rainfall and the large spatial coverage of MCS rainfall east of the Rocky Mountains, it is important to understand the hydrologic footprint of the two types of rainfall. The partitioning of rainfall into surface and subsurface runoff is largely determined by the intensity of rainfall and soil moisture condition. Surface runoff is primarily produced by infiltration excess overland flow (Horton 1933) for rain rate exceeding the infiltration rate; after the soil reaches saturation, surface runoff is mainly produced by saturation excess overland flow (Cappus 1960; Dunne and Black 1970). Through its impacts on surface runoff, rain rate plays an important role in determining water infiltration into the soil and subsequently, evapotranspiration (ET) and subsurface processes. This suggests that MCS and non-MCS rainfall may have different impacts on land–atmosphere interactions. As the central United States has been identified as a hotspot for land–atmosphere interactions (Koster et al. 2004), quantifying the relative contributions of MCS and non-MCS rainfall to ET may improve our understanding of soil moisture–precipitation feedback and the variability and predictability of rainfall in the region.
Besides land–atmosphere interactions, MCS and non-MCS rainfall may also influence the ecologic systems of the central United States differently because storm characteristics play an important role in ecohydrology (Brooks et al. 2009; Taylor 2012; Evaristo et al. 2015). Light-to-moderate rains replenish soil moisture and benefit plants (Weltzin et al. 2003), while intense rainfall may cause flooding (Trenberth 2011; Good and Caylor 2011). Water uptake by plants through ET is a critical component of the surface water balance especially in summer due to the large energy demand. The ecohydrologic impacts of MCS and non-MCS rainfall have important implications for vegetation productivity and flood risks in the significant grain growing region of the central United States (Knapp et al. 2008; Conant et al. 2018; Kloesel et al. 2018).
Observational analyses have revealed an increasing trend of MCS rainfall at the expense of non-MCS rainfall in the United States during the recent decades (Feng et al. 2019; Hu et al. 2020). Future projections suggest a threefold increase of intense summertime MCS frequency in northeast North America (Prein 2020). Through their impacts on land–atmosphere interactions and ecohydrology, understanding how MCS and non-MCS rainfall governs each component of the surface water balance is integral to understanding and predicting the effects of climate variability and change on water security and ecosystem services (Oki and Kanae 2006; Evaristo et al. 2015). This may be accomplished by tracking the surface and subsurface transit of water sources from the two types of rainfall events. Since MCS and non-MCS rainfall covers a broad area in the central United States, land surface modeling with numerical tracers may be the only viable approach for quantifying how MCS and non-MCS rainfall influence the surface water budget components. This is because stable water isotope tracers are sparsely observed in space and time, so they are incapable of providing event-scale water transit characteristics over a large area to be climatologically relevant (McGuire and McDonnell 2007).
In this study, the water tracer enabled version of the Noah land surface model with multiparameterization options (WT-Noah-MP; Niu et al. 2011; Hu et al. 2018) is used with the goal to quantify the different impacts of MCS and non-MCS rainfall on the surface water balance in the central United States. The tracer module allows “tagging” of the water from MCS or non-MCS rainfall and tracking its subsequent life cycle through surface–subsurface mixing and transit. Analysis is focused on fluxes important to land–atmosphere interactions and ecohydrology and processes dominating the differences between the impacts of MCS and non-MCS rainfall during 1997–2018. In section 2, we describe the method, model, and dataset used in the study. Model simulations are evaluated using observations and compared with simulations from other models in section 3. Section 4 describes the results of the water-tagging experiments and analysis of the processes differentiating the hydrologic response to MCS and non-MCS rainfall. Conclusions and discussions are provided in section 5.
2. Method, model, and dataset
a. MCS and non-MCS rainfall
MCS rainfall has been distinguished from non-MCS rainfall at 1/8° resolution based on hourly data from the North American Land Data Assimilation System (NLDAS; Mitchell 2004; Xia et al. 2017) by Feng et al. (2016). Rainfall associated with MCSs in the warm season (April–August) is identified if the major axis length of a precipitation feature (PF) exceeds 200 km and persists longer than 4 h (see more details in Feng et al. 2016). A PF is defined as a contiguous rainy area with pixel-level rain rate exceeding 1 mm h−1. Rainfall that is not associated with MCSs is categorized as non-MCS rainfall. The MCS and non-MCS rainfall characteristics based on PF tracking applied to the NLDAS data have been verified against those derived from MCS tracking based on satellite brightness temperature data (Feng et al. 2016) and MCS tracking with an updated algorithm applied to both satellite brightness temperature and 3D radar reflectivity data (Feng et al. 2018; Hu et al. 2020). Although NLDAS spans a longer time period, we only use the most recent 22 years from 1997 to 2018 in this study because starting from 1997, the U.S. Next Generation Weather Radar network (NEXRAD) data are incorporated to provide more reliable rainfall characteristics.
b. WT-Noah-MP
Rainfall from MCS and non-MCS events are tagged using the water tracer tool embedded in Noah-MP (WT-Noah-MP) to trace their transit processes in the terrestrial system. Similar to other land surface models, Noah-MP (Niu et al. 2011) represents the timing and extent of soil water budget, precipitation interception, storage, and eventual loss to either runoff or ET (Henderson-Sellers et al. 2006). However, Noah-MP provides flexibility for customizing simulations because it includes multiple options to parameterize different processes (e.g., canopy stomata resistance, runoff partitioning).
With a water tracer module added in Noah-MP, WT-Noah-MP can numerically tag water of a particular event or series of storms and distinguish the tagged water from the rest of water in the model as it moves through the surface–subsurface continuum until it ultimately leaves the system as ET or runoff (Hu et al. 2018). This water tracer tool is used to tag water molecules from MCS and non-MCS rainfall, respectively, when each type of rainfall occurs in the warm season from 1997 to 2018. The water tracers, representing water from each rainfall type, mix with each storage (e.g., canopy intercepted water storage, soil moisture in each vertical soil layer) along with its transit and contribute to the corresponding outward fluxes proportionately based on the relative composition of the tracers in each storage (see more details in Hu et al. 2018). Different from isotope-enabled land surface models (e.g., iso-MATSIRO, iCHASM, ECHAM5-JSBACH-wiso) that favor a direct comparison with isotope observations, WT-Noah-MP directly tracks the event water rather than the water isotopes. This allows a more straightforward interpretation of water transit without the complexity of isotope physics and the model is easier to run because no isotope data are needed as input to the model (Hu et al. 2018). Due to these advantages, similar water tracer tools have been implemented in climate models (Bosilovich and Schubert 2002; Insua-Costa and Miguez-Macho 2018). Implementation of tracers in land–atmosphere coupled system (Arnault et al. 2019) enables the impact of different rainfall types on subsequent rainfall events to be assessed.
c. Model configurations and forcing data
We configure our simulations using WT-Noah-MP with a similar spatial coverage as the NLDAS dataset (1/8°), but with a higher resolution 4-km grid spacing. Soil, vegetation and topographic parameters are derived from the 30-arc-s-resolution geographic dataset (https://www2.mmm.ucar.edu/wrf/users/download/get_sources_wps_geog.html) to better represent spatial heterogeneity of the land surface. Soil layers are configured using the default four layers with thicknesses of 0.1, 0.3, 0.6, and 1.0 m from top to bottom. The options we use for the different physics are shown in Table 1.
Option set for physical processes used in Noah-MP.
For each year between 1997 and 2018, we perform simulations covering 1 March–31 August, driven by the NLDAS forcing including precipitation, radiation, near-surface air temperature, wind and humidity, and surface pressure. Soil conditions are initialized at 0000 UTC 1 March of each year based on the soil states of the Noah simulations from the NLDAS archive for the same year, and reaches equilibrium states within the first month. Rainfall due to MCS storms from April to August of each year is cumulatively tagged by water tracers in a continuous simulation, and rainfall from non-MCS events are cumulatively tagged in another simulation. This way, we obtain two sets of model outputs with the same estimates of bulk storages and fluxes, but different tracer contributions to each storage and flux components to represent the contribution of MCS or non-MCS rainfall. Similarly, we can also tag rainfall during daytime and nighttime, respectively, in order to estimate the effect of diurnal cycle differences in MCS and non-MCS rainfall on each term in the surface water budget. Because WT-Noah-MP can only handle a single tracer, we run separate simulations to tag rainfall from different sources, although the simulations have identical bulk storages and fluxes.
3. Evaluation of bulk quantities simulated by Noah-MP
To evaluate our simulations, we compare bulk quantities of total ET, surface runoff and soil moisture in our simulations with the corresponding quantities from NLDAS estimated by different models available from the NLDAS-2 archive: Noah, Mosaic, and VIC with 1/8° grid spacing. For ET, we use two additional observation-based estimates: the global ET estimates from the MODIS satellite (1 km and monthly resolution; Mu et al. 2011) and the gridded ET product from FLUXNET towers (0.5° and monthly resolution; Jung et al. 2009). Note that it is not feasible to evaluate the tracer dynamics simulated by the model using observations because matching isotope measurements are lacking.
Compared with the April–August monthly averaged ET from 1997 to 2018, our simulation using Noah-MP reproduces the large east–west gradients in ET from Noah in the central United States (the black box in Fig. 1a), with magnitudes within the range represented by Noah, MOSAIC, and VIC (Figs. 1d–f). However, our simulation tends to overestimate the ET flux especially over the northern Great Plains compared with the MODIS and FLUXNET data, which also differ noticeably between each other. It is important to note that the observation-based ET estimates may also be subject to biases (Zhang et al. 2020).
Comparison of monthly averaged ET (April–August) (a) from simulation with (b) MODIS, (c) gridded FLUXNET product, and (d)–(f) the three models from NLDAS. The MODIS data in (b) are averaged over 2000–14, the gridded FLUXNET data in (c) are averaged over 1997–2011, and the rest are averaged over 1997–2018.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
For surface runoff, our simulation shows a north–south-oriented region of surface runoff exceeding 12 mm month−1 in the dry-to-wet transition zone (~95°W). Wider and narrower regions of higher surface runoff are also present in Noah and MOSAIC, respectively (Figs. 2b,c), but VIC shows an evident maximum over the upper Mississippi river basin (Fig. 2d) not found in the other models. The warm-season soil moisture in the top 1 m shows smaller zonal gradient than other quantities, and our simulation is very similar to that of other models. The soil moisture fluctuations within each warm season are also very similar among the models (not shown). Thus, we conclude that our simulations using Noah-MP can well reproduce the bulk quantities in the surface–subsurface systems comparable to the NLDAS product, but comparison with observations show some biases in each bulk quantity. Cai et al. (2014) compared the hydrological performance of four land surface models in the NLDAS test bed, which includes Noah-MP, with observations. They noted that Noah-MP has the best performance in simulating soil moisture and terrestrial water storage while some other models perform better in simulating ET and streamflow. The Noah-MP simulations reported here show spatial features of the bulk water fluxes comparable to those discussed in Cai et al. (2014).
Comparison of April–August averaged estimates of (left) RS and (right) top 1-m SM from simulations using (a),(e) WT-Noah-MP with the three models from NLDAS: (b),(f) Noah; (c),(g) MOSAIC; and (d),(h) VIC. The black boxes indicate the area of the central United States we used for areal averaging.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
4. Water-tagging simulation results
a. Differences in water budget terms
Our main focus is the partitioning of MCS and non-MCS contributed water into different water budget terms revealed by the water tracers. Figure 3 shows the water tagged from MCS and non-MCS rainfall in each warm season (April–August) over the central United States (averaged over the black boxes in Fig. 1) and the partitioning into surface runoff (RS) and ET accumulated during the warm season and the change in soil moisture (SM) from 1 April to 31 August.
Warm-season rainfall (April–August) of each year from (a) MCS and (b) non-MCS, and their partitioning into the three key components: RS, ET, and SM at the end of August. (c),(d) The 22-yr averaged partitioning of MCS and non-MCS rainfall into the key components. (e) Comparison of the partitioning of MCS and non-MCS rainfall to the key components for each year.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
For both types of rainfall, a majority of the water input from rainfall leaves the terrestrial system as ET, indicating ET as the most important water flux in summer due to the large evaporative demand driven by the large net input energy. A smaller fraction of MCS/non-MCS rainfall remains in the soil by the end of August, followed by an even smaller fraction that goes into surface runoff. All other water budget components, including snow water, subsurface runoff and canopy intercepted water, are negligible (Figs. 3a–d). Compared with non-MCS rainfall, MCS rainfall is characterized by larger fractions contributing to surface runoff generation (10.3% for MCS and 6.9% for non-MCS rainfall). In contrast, non-MCS rainfall has a larger fractional contribution to ET than MCS rainfall (56.1% for MCS and 64.8% for non-MCS). As the difference in water fluxes associated with MCS and non-MCS rainfall is larger for ET than surface runoff, non-MCS rainfall contributes less to soil water storage by the end of August. Although there are large interannual variations of the MCS and non-MCS rainfall as well as the fraction of MCS-to-total rainfall (between 28.5% and 54.6%), differences in the partitioning into surface runoff, ET, and soil moisture are quite robust and consistent each year, as evident by the alignment of each annual flux ratio (each dot) for ET, surface runoff, and soil moisture change above or below the 1:1 line in Fig. 3e.
Within each warm season, Fig. 4a shows an earlier peak of MCS rainfall in May and June while non-MCS rainfall affects the central United States more uniformly throughout the warm season. The early-season MCS rainfall peak has a corresponding but amplified peak in RS (Fig. 4b), resulting in a larger difference between MCS and non-MCS contributed surface runoff between April and June. The amplified difference is due not only to the higher intensity of MCS rainfall but also the wetter soil conditions in spring than in summer, which increases the seasonality of flooding (Berghuijs et al. 2016; Ye et al. 2017). Hence MCS rain is more likely to produce flooding, with a more concentrated timing in spring, than non-MCS rain. Evidently, non-MCS rainfall generates significantly less surface runoff, especially during spring. Also evident, however, is the larger contribution of non-MCS rainfall to ET throughout the warm season (Fig. 4c). Unlike the larger difference in MCS versus non-MCS surface runoff in spring, the difference in ET is larger during summer when more energy is available (as a combined effect of solar insolation and plant phenology) to support ET.
(left) 22-yr averaged seasonal cycles (15-day running mean) of MCS and non-MCS contributed (a) rainfall, (b) RS, and (c) ET. (right) 22-yr averaged diurnal cycles of MCS and non-MCS contributed (d) rainfall, (e) RS, and (f) ET. The blue and green shading indicate the 25th and 75th percentile ranges.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
In response to the different diurnal cycles of MCS and non-MCS rainfall, MCS rainfall results in peak surface runoff during night hours that well exceeds the surface runoff from non-MCS rainfall (Fig. 4e). For ET, both MCS and non-MCS rainfall show a daytime peak in response to the peak energy demand but ET from non-MCS rainfall is clearly larger (Fig. 4f).
The spatial pattern of ET sourced from MCS and non-MCS rainfall is largely driven by their spatial distribution of two rainfall types. ET from MCS rainfall maximizes in the central United States (black boxes in Fig. 1), collocated with the maximum MCS rainfall (Figs. 5a,b), while ET from non-MCS rainfall is more evenly distributed over the central United States. RS from MCS rainfall is much greater than that from non-MCS rainfall (Figs. 5c,g), and its spatial distribution resembles that of the total surface runoff (Fig. 2a), featuring a band with maximum values along ~95°W. For soil moisture (SM) from MCS rainfall, it shows a similar pattern comparable to MCS rainfall with a slightly eastward shifted maximum. SM from non-MCS is mostly below 60 mm at the end of the warm season within our domain.
Spatial patterns of accumulated warm-season rainfall (April–August) associated with (a),(e) MCS and non-MCS events; (b),(f) ET; (c),(g) RS; and (d),(h) SM in the top 2 m on 31 Aug from the 22-yr averages (1997–2018).
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
Unlike the magnitude of each surface water budget term that highly depends on the distribution of its source rainfall, the partitioning of each component is more spatially homogeneous (Fig. 6), especially for the ET and SM fractions. However, regional differences are still apparent, reflecting the impacts of topography and soil properties on surface runoff, and hence ET and SM. Overall, non-MCS rainfall is largely balanced by ET (Fig. 6b) while all three fluxes contribute in the surface water balance of MCS rainfall (Figs. 6a–c). These results also confirm that MCS rainfall is the major contributor to surface runoff while non-MCS rainfall plays a more important role in supplying moisture for local recycling through ET.
Spatial patterns of the partitioning (in fraction) of warm-season (April–August) MCS and non-MCS rainfall into each component: April–August accumulated (a),(d) ET and (b),(e) RS, and (c),(f) SM in the top 2 m on 31 Aug averaged over 1997–2018.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
b. Dominant contribution of MCS rainfall to surface runoff
To understand the dominant controls of the greater contribution of MCS rainfall to surface runoff, it is important to examine the model partitioning of surface runoff and infiltration. The surface runoff and infiltration rate are parameterized in Noah-MP (Schaake et al. 1996) using Eqs. (1)–(3):
where Qs is surface runoff (m); Px is the water input to the soil surface, which is mostly from rainfall (m); Ic is the infiltration capacity that is calculated by Eq. (3) (m); I is the infiltration (m); Δt is the time interval of each model step (s); θ(i), θwilt(i), and θmax(i), are the simulated, wilting point, and saturation soil moisture at soil layer i (m3 m−3), respectively; Δz(i) is the soil thickness of layer i (m); and k is a parameter representing the nonlinear relationship between infiltration and soil moisture deficit (s−1). The portion of Px exceeding the infiltration capacity becomes surface runoff.
From Eq. (1), rainfall rate and soil moisture deficit can both affect infiltration rate. Figure 7a illustrates the relationship between infiltration rate and rainfall intensity, showing MCS rainfall rate well exceeding the infiltration rate, while non-MCS rainfall rate generally occurring below the infiltration rate. As the amount of rainfall exceeding the infiltration rate becomes surface runoff, Fig. 7a indicates that the dominant mechanism for generating surface runoff is infiltration excess flow, given the relatively dry soil conditions in the warm season. The dominant control of the MCS rainfall intensity on the greater surface runoff is supported by the insignificant differences between the soil moisture deficits when each type of rainfall occurs (Fig. 7b).
(a) The relationship between rainfall intensity and infiltration rate and (b) the distributions of infiltration capacity Ic (related to the antecedent soil moisture deficit) when MCS or non-MCS rainfall occurs using data from the 2011 season. (c) Scatterplot of RS generation ratios (RS/PP) and MCS-event rainfall amounts, with the boxplot showing the statistics of the RS generation ratio within each bin. (d) Scatterplot of RS generation ratio (RS/PP) and MCS-event rain rate for all MCS events (blue circles) and April MCSs (orange circles).
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
For MCS rainfall that usually exceeds the infiltration capacity, its contribution to RS is further investigated in relation with the precipitation strength of each MCS event. Figure 7c shows that the RS generation ratio (RS/PP) increases nonlinearly with the accumulated rainfall amount associated with each MCS event (PP) calculated from Eq. (4). The median value of the RS generation ratio exceeds 0.1 for MCS events with rainfall amount greater than 109 m3, resulting in significant surface runoff that can induce flooding. As larger MCSs also tend to last longer, relating the RS generation ratio with MCS-event rain rate [Eq. (5)] instead of rainfall amount results in a more linear relationship (Fig. 7d). Note that most of the outliers from the linear relationship are due to MCS events occurring in April when soil moisture is more saturated so a higher fraction of MCS rainfall is turned into RS regardless of the rain rate. This indicates a secondary role played by soil moisture in determining the RS generation ratio by MCS rainfall in the early part of the warm season, which is generally wetter due to precipitation in the previous winter:
where Pi,t is hourly MCS rainfall amount at pixel i and hour t, Ai,t is the corresponding pixel area, Nt is the total number of pixels that MCS rainfall occurs at hour t, and T is the lifetime of each MCS event.
c. Dominant contribution of non-MCS rainfall to evapotranspiration
To explain the differences of MCS and non-MCS rainfall in contributing to ET, we compare the three components of ET: evaporation from the soil (EDIR), evaporation from the canopy-intercepted water (ECAN), and transpiration (ETRAN). Note that both EDIR and ETRAN extract water from the soil so their behavior is closely linked with the soil moisture feedback to the atmosphere. While ETRAN from both sources of rainfall is quite similar (Fig. 8c), we found that the differences in ET are explained by ECAN and EDIR, both showing larger fluxes from non-MCS rainfall than MCS rainfall (Figs. 8a,b). The larger ECAN from non-MCS rainfall is closely related to the broader area impacted by this type of rainfall (Figs. 5a,e) and thus a greater portion intercepted by canopy leaves and evaporating through ECAN (Figs. 5b,f).
(a)–(d) Diurnal cycles of MCS and non-MCS contributed canopy evaporation, soil evaporation, plant transpiration, and ET. (e) Seasonal cycles of the soil moisture in the top soil layer and in the four layers sourced from (f) MCS and (g) non-MCS rainfall. (h)–(j) Soil moisture profiles sourced from MCS and non-MCS rainfall on 1 May, 1 Jul, and 1 Sep. The orange shaded layers in (h)–(j) indicate layers with significant differences between MCS and non-MCS contributed soil moisture (>95% confidence level), and all results are averages from 1997 to 2018.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
For EDIR, this process draws water from the top soil layer in Noah-MP. A larger contribution from non-MCS rainfall to soil moisture in the top layer is evident (Fig. 8e), and such difference is more evident toward the end of the warm season as MCS rainfall decreases after June (Fig. 4a). However, such soil moisture difference is inconspicuous in the deeper soil layers (Figs. 5f,g). The soil moisture profiles at three different times during the warm season clearly reveal such differences (Figs. 5h–j), showing that MCS rainfall penetrates deeper into the soil column while non-MCS rainfall is confined closer to the top of the soil column. Although a larger fraction of non-MCS rainfall infiltrates into the soil due to its smaller partitioning into RS by infiltration–excess runoff discussed earlier, the lighter non-MCS rainfall mostly stays in the top soil layer, thus contributing more to ET through EDIR. In contrast, despite a smaller fraction of MCS rainfall infiltrates into the soil column due to the greater partitioning into RS, the water tends to penetrate deeper, as more intense rainfall can produce greater water–header pressure differences to drive downward transit. Analysis of the matric potential difference between the top and second soil layers (not shown) confirms that the higher intensity MCS rainfall drives stronger downward percolation than non-MCS rainfall. Furthermore, with a longer lifetime than non-MCS storms, MCSs produce longer, continuous hours of rainfall that increase soil wetness, which may lead to increased hydraulic conductivity. Differences in subsurface water transit between MCS and non-MCS rainfall caused by differences in pressure head and hydraulic conductivity resulting from differences in rainfall intensity and lifetime are analogous to the heterogeneity of water transit due to differences in hydrologic connectivity: water can bypass the soil matrix through macropores in wet conditions and reach specific depths faster than water moving through the soil micropores when the soil is drier.
Besides rainfall intensity and lifetime that influence water infiltration into the soil and hence soil moisture profiles and soil evaporation, the difference in diurnal cycles of MCS and non-MCS rainfall may also contribute to the difference in ET. More specifically, the nocturnal timing of MCS rainfall allows more time for downward transit of MCS rainwater in the subsurface before evaporation increases during daytime and hence reduces the contribution of MCS rainfall to ET. In contrast, a better alignment of the late-afternoon peak of non-MCS rainfall with the energy demand increases the partitioning of non-MCS rainfall to ET. We hypothesize that nocturnal rainfall would contribute to a smaller fraction to ET than daytime rainfall. To determine the contribution of diurnal cycle to the ET difference, we perform additional simulations by tagging only the daytime (0800–2000 local time) and nighttime (2000–0800 local time) rainfall, respectively. Five out of the 22 years in 1997–2018 with the highest fractions of MCS-to-total rainfall (2008, 2011, 2015, 2016, and 2017) are selected for this experiment for more evident impacts of the diurnal cycle of rainfall.
From the daytime and nighttime water-tagging experiments, we found slightly higher fractions of the daytime rainfall becoming ET than that sourced from nighttime rainfall, and this applies to both MCS and non-MCS rainfall (Fig. 9a). Such higher fractions of daytime rainfall contributing to ET is compensated by the slightly lower partitioning into RS for MCS rainfall versus the slightly lower partitioning into SM for non-MCS rainfall. Decomposing ET into its three components, higher fractions of daytime rainfall become ECAN than nighttime rainfall (Fig. 9b), indicating the role of daytime energy demand in depleting the canopy-interception storage. The process of depletion and refilling of the canopy-interception storage by daytime rainfall accelerates the turnover of water intercepted by canopy leaves and increases ECAN during daytime compared to nighttime. Notably, ECAN is a more important component of ET for non-MCS rainfall than MCS rainfall due to its broader spatial coverage (Fig. 5f), as discussed earlier. However, such interception is limited in amount and a large fraction of water that drips through plants contributes to ET through EDIR. Balancing the day–night differences of ECAN, EDIR has a higher fractional contribution from nighttime rainfall than daytime rainfall for both MCS and non-MCS events (Fig. 9b). For ETRAN, no clear differences are found between daytime and nighttime rainfall especially for MCS rainfall. This ET component is more difficult to interpret because it is a mixture of water sources from different soil layers accessed by root systems that vary with plants. Also, soil moisture at deeper layers carries a longer memory of the previous rainfall events well beyond days, making it more difficult to differentiate the impacts of daytime and nighttime rainfall. Overall, the tagging experiments suggest that the differences in diurnal cycles of MCS and non-MCS rainfall play a negligible role on their ET differences.
Comparison of (a) partitioning of daytime and nighttime rainfall into ET, RS, and SM and (b) partitioning of ET from daytime and nighttime rainfall into ETRAN, EDIR, and ECAN. Each dot/star represents the partitioning of each year in 2008, 2011, 2015, 2016, and 2017 by 31 Aug.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
d. Contribution to daily surface runoff and evapotranspiration
The tagging capability not only enables us to investigate the contributions of MCS and non-MCS rainfall to different surface water components, it also allows us to understand how the RS and ET sourced from MCS and non-MCS rainfall combines into daily RS and ET. An especially interesting aspect is to quantify the fraction of extreme RS and ET derived from MCS and non-MCS rainfall to understand the importance of MCS and non-MCS rainfall to flooding and moisture recycling (moisture supply for precipitation).
Figure 10a clearly reveals that the fraction of daily RS sourced from MCS rainfall increases with increasing percentile of daily RS, while the opposite is true for non-MCS rainfall. For daily RS in the top 10th percentile bin (see Table 2 for the corresponding percentile values), 74.6% comes from MCS rainfall, indicating a critical role of MCS rainfall in local flooding. Such a large fraction of top 10th percentile RS coming from MCS rainfall is strongly linked to the MCS precipitation strength and its impact on RS: MCS events with larger rainfall amount also have a larger fraction of the rainfall contributing to RS (Fig. 7c), which combine to explain why a majority of the extreme RS events are generated by MCS events. The top 10th percentile RS events occur more frequently in April, May, and June (Fig. 10a), consistent with the predominant RS from MCS rainfall in these months (Fig. 4b) due to higher MCS rainfall amount (Fig. 4a) and wetter soil conditions that allow large fractions of MCS rainfall to contribute to RS even for MCS events with low relatively intensity rain rate (Fig. 7d). Interestingly, days with RS in the 0th–10th percentile also occur more frequently in April and May, suggesting substantial variabilities in daily RS during these months. RS in July and August are more likely to be moderate (between 20th and 70th percentiles). Note that the small fraction of RS in the 0th–10th percentile bin not explained by MCS or non-MCS rainfall comes from delayed melt of the snowpack initialized on 1 April, as most RS comes from infiltration excess runoff that is not affected by antecedent soil moisture.
(a) The contributions from MCS and non-MCS rainfall to RS in each 10th percentile range (left bars), and the contribution from each month for days within each 10th percentile range (right bars). (b) As in (a), but the contributions from MCS and non-MCS rainfall to ET and the contribution from each month at each 10th percentile range. The corresponding magnitudes of daily RS and ET in each percentile range are listed in Table 2.
Citation: Journal of Hydrometeorology 21, 10; 10.1175/JHM-D-20-0081.1
The values of daily RS and ET at every 10th percentile.
For ET, the fraction contributed by non-MCS rainfall is always greater than that from MCS rainfall, regardless of the percentile of the daily ET. However, notable differences in the MCS and non-MCS fractions are evident across the season. In April when ET is weaker (mostly in the 0th–30th percentiles), MCS and non-MCS rainfall is mainly stored in the soil while a majority of ET (~70%) comes from antecedent soil moisture that is present before tagging (Fig. 10b). The antecedent soil moisture contributes to daily ET at all percentiles, but the contributions are less for the higher percentile daily ET. The daily ET fraction contributed by MCS and non-MCS rainfall reaches a maximum for daily ET between the 70th and 100th percentiles. At these percentiles, MCS and non-MCS rainfall contributions increase to ~28% and ~36%, respectively, with larger contributions coming from June and July when solar radiation is strongest. August daily ET occurs more frequently in the 40th–60th percentile range as available energy decreases, but the fraction of ET from MCS and non-MCS rainfall reaches a maximum after accumulation in the soil throughout the warm season.
5. Conclusions and discussions
We examine the contributions of MCS and non-MCS rainfall to the surface water balance using a water tracer tool embedded in Noah-MP. In the central United States, MCS and non-MCS rainfall differs in their intensity, diurnal timing, and seasonal and spatial distributions, as elaborated in Hu et al. (2020). Using numerical water tagging in land surface simulations, we show that MCS rainfall is more important to surface runoff while non-MCS storms are more important to ET. The large difference in rainfall intensity (~7 times) is mostly responsible for the differences in the flux partitioning. The much more intense MCS rainfall contributes more importantly to surface runoff through infiltration excess flow. Over 70% of the top 10th percentile runoff is contributed by MCS rainfall, demonstrating the important role of MCS rainfall to local flooding. More intense rainfall from MCS also produces higher pressure heads and higher hydrologic connectivity to drive faster downward transit of rainwater into deeper soils. With lighter rain rate, non-MCS rainfall mostly increases soil moisture in the top soil layer. Differences in the soil moisture profiles of MCS and non-MCS rainfall events dictate more soil evaporation into the atmosphere for non-MCS rainfall. Larger rain area coverage also favors larger canopy-intercepted evaporation from non-MCS rainfall relative to MCS rainfall. As a result, RS from MCS rainfall is of particular importance to extreme RS while non-MCS rainfall plays a more important role in ET throughout the warm season. Among the differences in rainfall characteristics, rain intensity, and to a lesser extent rainfall area, play important roles in controlling the distinct surface water balance associated with MCS and non-MCS rainfall, while the difference in diurnal timing has negligible effect.
The differences in soil moisture profiles associated with MCS and non-MCS rainfall are particularly interesting, because these differences can affect soil moisture–precipitation feedback. Water from non-MCS rainfall that is confined more in the shallow soil layers and is more homogeneous in space may play a more important role in supplying moisture for subsequent rainfall through ET. On the other hand, MCS rainfall tends to penetrate deeper into the soil but with greater heterogeneity in space, it may be more important in controlling where surface-induced perturbation occurs. Thus, both rainfall types are critical components of surface fluxes that can affect the strength and location of subsequent rainfall events (see an example in Arnault et al. 2019). In addition, the different soil moisture profiles from MCS and non-MCS rainfall can also be linked with the responses of the ecosystems. Vegetation with shallow roots might a have higher dependence on non-MCS rainfall for water supply while plants with deep roots can access water from MCS rainfall through a deeper soil layer. As increasing trend of heavy rainfall through MCSs has been observed in recent decades at the expense of more moderate non-MCS rainfall, agricultural plants with shallow roots might have to rely more on irrigation in order to maintain agricultural productivity.
It is important to caution that our water-tagging results are highly dependent on the model representations of different terrestrial processes by Noah-MP. Because the tagging components are partitioned from the total flux in Noah-MP, the bias of simulating each component by Noah-MP can result in bias in the water tracer components. For example, the overestimated ET by Noah-MP is likely to overestimate ET amount from MCS and non-MCS rainfall, but how it affects the fraction from each component remains unclear unless observational estimates, such as stable water isotopes, are available. In addition, we assume the tracers are completely mixed within each storage and it may cause slower tracer transit in the soil especially when the soil is wet, and water can transit through preferential flow pathways. This is particularly important for MCS rainfall with high intensity so the depth to which MCS water can percolate could be underestimated. Plant hydraulic processes have important effects on plant transpiration and hence ET. For example, hydraulic redistribution of soil water facilitated by plant roots can modulate the vertical distribution of soil moisture and root access to soil moisture for ET (Caldwell and Richards 1989). Although mechanistic plant hydraulic schemes have been developed for use in land surface models (e.g., Kennedy et al. 2019), plant hydraulic processes are not currently represented in Noah-MP. Last, canopy-intercepted evaporation contributes to the ET difference between MCS and non-MCS rainfall as well as daytime versus nighttime rainfall. This process is represented in Noah-MP using a formulation similar to Shuttleworth (1988) that accounts for subgrid rainfall variability. Wang et al. (2007) noted that canopy-interception can be better represented by distinguishing the rain type, suggesting that parameterizations of ECAN can also contribute to uncertainty in quantifying the flux partitioning for MCS and non-MCS rainfall. Despite using the same atmospheric forcing, the diverse water fluxes simulated by land surface models (e.g., Figs. 1 and 2; Cai et al. 2014) motivate the need to evaluate the uncertainty associated with the flux partitioning of MCS and non-MCS rainfall to better understand their impacts on land–atmosphere interactions and ecohydrology.
Another important process that is not represented by Noah-MP is lateral flow in the land surface and subsurface. Such ridge-to-valley redistribution of water through shallow and deep pathways in the land surface is not only important for runoff generation, but can also substantially modulate the ET through its impacts on near-surface soil moisture and groundwater table dynamics (Fan et al. 2017; Chang et al. 2018; Fan et al. 2019). This regulation on ET by lateral flow can affect MCS rainfall percolation in the soil and accumulation in downstream valleys, hence affecting the partitioning of MCS rainfall to ET and runoff at seasonal scale. Quantifying the impacts of horizontal redistribution of the tagged water may be achieved using models that better incorporates hillslope hydrology [e.g., WRF-Hydro framework used in Arnault et al. (2019)] and could be pursued in future studies.
Besides uncertainty in land surface modeling and water tagging, our results may also be sensitive to how MCS rainfall is defined. Here, we define MCS-associated rainfall as a continuous rainy area with pixel-level rain rate exceeding 1 mm h−1, while the fringe areas of the MCSs with rain rate below this cutoff threshold are treated as non-MCS rainfall. Whether to include the fringe areas of the MCSs as MCS or non-MCS rainfall and the choice of the cutoff rain rate might induce differences in rainfall intensity, and thus can result in variations in our tagging results. Despite the various sources of uncertainty, the relative importance of MCS and non-MCS rainfall to ET and RS is likely robust as the dominant processes controlling the different responses can be mechanistically attributed to the distinct characteristics of MCS and non-MCS rainfall. To further investigate the role of MCS and non-MCS rainfall in land–atmosphere interactions, future studies will use coupled land–atmosphere models with water tracers to quantify their contributions to subsequent rainfall events.
Acknowledgments
This research is supported by the U.S. Department of Energy Office of Science Biological and Environmental Research as part of the Regional and Global Modeling and Analysis program area. PNNL is operated for the Department of Energy by Battelle Memorial Institute under Contract DE-AC05-76 RL01830. The NLDAS dataset is obtained from NASA (http://ldas.gsfc.nasa.gov/nldas/NLDAS2forcing.php#AppendixC). The monthly MODIS evapotranspiration data are obtained from University of Montana (http://files.ntsg.umt.edu/data/NTSG_Products/MOD16). The monthly gridded FLUXNET data is obtained from Max-Planck Institute for Biogeochemistry, Germany (https://www.bgc-jena.mpg.de/geodb/projects/Home.php). The WT-Noah-MP code can be accessed through the github link: https://github.com/huancui/WT-Noah-MP.
REFERENCES
Arnault, J., J. Wei, T. Rummler, B. Fersch, Z. Zhang, G. Jung, S. Wagner, and H. Kunstmann, 2019: A joint soil-vegetation-atmospheric water tagging procedure with WRF-hydro: Implementation and application to the case of precipitation partitioning in the upper Danube River Basin. Water Resour. Res., 55, 6217–6243, https://doi.org/10.1029/2019WR024780.
Ashley, W. S., T. L. Mote, P. G. Dixon, S. L. Trotter, E. J. Powell, J. D. Durkee, and A. J. Grundstein, 2003: Distribution of mesoscale convective complex rainfall in the United States. Mon. Wea. Rev., 131, 3003–3017, https://doi.org/10.1175/1520-0493(2003)131<3003:DOMCCR>2.0.CO;2.
Berghuijs, W. R., R. A. Woods, C. J. Hutton, and M. Sivapalan, 2016: Dominant flood generating mechanisms across the United States. Geophys. Res. Lett., 43, 4382–4390, https://doi.org/10.1002/2016GL068070.
Bosilovich, M. G., and S. D. Schubert, 2002: Water vapor tracers as diagnostics of the regional hydrologic cycle. J. Hydrometeor., 3, 149–165, https://doi.org/10.1175/1525-7541(2002)003<0149:WVTADO>2.0.CO;2.
Brooks, J. R., H. R. Barnard, R. Coulombe, and J. J. McDonnell, 2009: Ecohydrologic separation of water between trees and streams in a Mediterranean climate. Nat. Geosci., 3, 100–104, https://doi.org/10.1038/ngeo722.
Cai, X., Z.-L. Yang, Y. Xia, M. Huang, H. Wei, L. R. Leung, and M. B. Ek, 2014: Assessment of simulated water balance from Noah, Noah-MP, CLM, and VIC over CONUS using the NLDAS test bed. J. Geophys. Res. Atmos., 119, 13 751–13 770, https://doi.org/10.1002/2014JD022113.
Caldwell, M. M., and J. H. Richards, 1989: Hydraulic lift: Water efflux from upper roots improves effectiveness of water uptake by deep roots. Oecologia, 79 (1), 1–5, https://doi.org/10.1007/BF00378231.
Cappus, P., 1960: Etude des lois de l'écoulement-Application au calcul et à la prévision des débits. Houille Blanche, 4A, 493–520, https://doi.org/10.1051/lhb/1960007.
Chang, L.-L., and Coauthors, 2018: Why do large-scale land surface models produce a low ratio of transpiration to evapotranspiration? J. Geophys. Res. Atmos., 123, 9109–9130, https://doi.org/10.1029/2018JD029159.
Conant, R. T., and Coauthors, 2018: Northern Great Plains. Impacts, Risks, and Adaptation in the United States: The Fourth National Climate Assessment, Vol. II, U.S. Global Change Research Program, 136–140.
Dunne, T., and R. D. Black, 1970: An experimental investigation of runoff production in permeable soils. Water Resour. Res., 6, 478–490, https://doi.org/10.1029/WR006i002p00478.
Evaristo, J., S. Jasechko, and J. J. McDonnell, 2015: Global separation of plant transpiration from groundwater and streamflow. Nature, 525, 91–94, https://doi.org/10.1038/nature14983.
Fan, Y., G. Miguez-Macho, E. G. Jobbágy, R. B. Jackson, and C. Otero-Casal, 2017: Hydrologic regulation of plant rooting depth. Proc. Natl. Acad. Sci. USA, 114, 10 572–10 577, https://doi.org/10.1073/pnas.1712381114.
Fan, Y., and Coauthors, 2019: Hillslope hydrology in global change research and Earth system modeling. Water Resour. Res., 55, 1737–1772, https://doi.org/10.1029/2018WR023903.
Feng, Z., L. R. Leung, S. Hagos, R. A. Houze, C. D. Burleyson, and K. Balaguru, 2016: More frequent intense and long-lived storms dominate the springtime trend in central US rainfall. Nat. Commun., 7, 13429, https://doi.org/10.1038/ncomms13429.
Feng, Z., L. R. Leung, R. A. Houze Jr., S. Hagos, J. Hardin, Q. Yang, B. Han, and J. Fan, 2018: Structure and evolution of mesoscale convective systems: Sensitivity to cloud microphysics in convection-permitting simulations over the United States. J. Adv. Model. Earth Syst., 10, 1470–1494, https://doi.org/10.1029/2018MS001305.
Feng, Z., R. A. Houze Jr., L. R. Leung, F. Song, J. C. Hardin, J. Wang, W. I. Gustafson Jr., and C. R. Homeyer, 2019: Spatiotemporal characteristics and large-scale environments of mesoscale convective systems east of the Rocky Mountains. J. Climate, 32, 7303–7328, https://doi.org/10.1175/JCLI-D-19-0137.1.
Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contribution of mesoscale convective weather systems to the warm-season precipitation in the United States. J. Climate Appl. Meteor., 25, 1333–1345, https://doi.org/10.1175/1520-0450(1986)025<1333:TCOMCW>2.0.CO;2.
Good, S. P., and K. K. Caylor, 2011: Climatological determinants of woody cover in Africa. Proc. Natl. Acad. Sci. USA, 108, 4902–4907, https://doi.org/10.1073/pnas.1013100108.
Haberlie, A. M., and W. S. Ashley, 2019: A radar-based climatology of mesoscale convective systems in the United States. J. Climate, 32, 1591–1606, https://doi.org/10.1175/JCLI-D-18-0559.1.
Henderson-Sellers, A., and Coauthors, 2006: Stable water isotope simulation by current land-surface schemes: Results of iPILPS Phase 1. Global Planet. Change, 51, 34–58, https://doi.org/10.1016/j.gloplacha.2006.01.003.
Horton, R. E., 1933: The role of infiltration in the hydrologic cycle. Eos, Trans. Amer. Geophys. Union, 14, 446–460, https://doi.org/10.1029/TR014i001p00446.
Houze, R. A., Jr., 2004: Mesoscale convective systems. Rev. Geophys., 42, RG4003, https://doi.org/10.1029/2004RG000150.
Houze, R. A., Jr., B. F. Smull, and P. Dodge, 1990: Mesoscale organization of springtime rainstorms in Oklahoma. Mon. Wea. Rev., 118, 613–654, https://doi.org/10.1175/1520-0493(1990)118<0613:MOOSRI>2.0.CO;2.
Hu, H., F. Dominguez, P. Kumar, J. McDonnell, and D. Gochis, 2018: A numerical water tracer model for understanding event-scale hydrometeorological phenomena. J. Hydrometeor., 19, 947–967, https://doi.org/10.1175/JHM-D-17-0202.1.
Hu, H., L. R. Leung, and Z. Feng, 2020: Observed warm-season characteristics of MCS and non-MCS rainfall and their recent changes in the central United States. Geophys. Res. Lett., 47, e2019GL086783, https://doi.org/10.1029/2019GL086783.
Insua-Costa, D., and G. Miguez-Macho, 2018: A new moisture tagging capability in the weather research and forecasting model: Formulation, validation and application to the 2014 Great Lake-effect snowstorm. Earth Syst. Dyn., 9, 167–185, https://doi.org/10.5194/esd-9-167-2018.
Jung, M., M. Reichstein, and A. Bondeau, 2009: Towards global empirical upscaling of FLUXNET eddy covariance observations: Validation of a model tree ensemble approach using a biosphere model. Biogeosciences, 6, 2001–2013, https://doi.org/10.5194/bg-6-2001-2009.
Kennedy, D., S. Swenson, K. W. Oleson, D. M. Lawrence, R. Fisher, A. C. Lola da Costa, and P. Gentine, 2019: Implementing plant hydraulics in the Community Land Model, version 5. J. Adv. Model. Earth Syst., 11, 485–513, https://doi.org/10.1029/2018MS001500.
Kloesel, K., and Coauthors, 2018: Southern Great Plains. Impacts, Risks, and Adaptation in the United States: The Fourth National Climate Assessment, Vol. II, U.S. Global Change Research Program, 987–1035.
Knapp, A. K., C. Beier, D. D. Briske, A. T. Classen, and Y. L. Bioscience, 2008: Consequences of more extreme precipitation regimes for terrestrial ecosystems. BioScience, 58, 811–821, https://doi.org/10.1641/B580908.
Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 1138–1140, https://doi.org/10.1126/science.1100217.
Laing, A. G., and J. M. Fritsch, 1997: The global population of mesoscale convective complexes. Quart. J. Roy. Meteor. Soc., 123, 389–405, https://doi.org/10.1002/qj.49712353807.
Maddox, R. A., 1980: Mesoscale convective complexes. Bull. Amer. Meteor. Soc., 61, 1374–1387, https://doi.org/10.1175/1520-0477(1980)061<1374:MCC>2.0.CO;2.
McGuire, K., and J. McDonnell, 2007: Stable isotope tracers in watershed hydrology. Stable Isotopes in Ecology and Environmental Science, 2nd ed. M. Robert and L. Kate, Eds., Blackwell Publishing, 334–373.
Mitchell, K. E., 2004: The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system. J. Geophys. Res., 109, D07S90, https://doi.org/10.1029/2003JD003823.
Mu, Q., M. Zhao, and S. W. Running, 2011: Improvements to a MODIS global terrestrial evapotranspiration algorithm. Remote Sens. Environ., 115, 1781–1800, https://doi.org/10.1016/j.rse.2011.02.019.
Niu, G.-Y., and Coauthors, 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. J. Geophys. Res., 116, D12109, https://doi.org/10.1029/2010JD015139.
Oki, T., and S. Kanae, 2006: Global hydrological cycles and world water resources. Science, 313, 1068–1072, https://doi.org/10.1126/science.1128845.
Prein, A. F., 2020: Simulating North American mesoscale convective systems with a convection-permitting climate model. Climate Dyn., 55, 95–110, https://doi.org/10.1007/s00382-017-3993-2.
Sakaguchi, K., and X. Zeng, 2009: Effects of soil wetness, plant litter, and under-canopy atmospheric stability on ground evaporation in the Community Land Model (CLM3.5). J. Geophys. Res., 114, D01107, https://doi.org/10.1029/2008JD010834.
Schaake, J. C., V. I. Koren, Q. Y. Duan, K. Mitchell, and F. Chen, 1996: Simple water balance model for estimating runoff at different spatial and temporal scales. J. Geophys. Res., 101, 7461–7475, https://doi.org/10.1029/95JD02892.
Shuttleworth, W. J., 1988: Macrohydrology - The new challenge for process hydrology. J. Hydrol., 100, 31–56, https://doi.org/10.1016/0022-1694(88)90180-1.
Song, F., Z. Feng, L. R. Leung, R. A. Houze Jr., J. Wang, J. Hardin, and C. R. Homeyer, 2019: Contrasting spring and summer large-scale environments associated with mesoscale convective systems over the U.S. Great Plains. J. Climate, 32, 6749–6767, https://doi.org/10.1175/JCLI-D-18-0839.1.
Taylor, R. G., 2012: Evidence of the dependence of groundwater resources on extreme rainfall in East Africa. Nat. Climate Change, 3, 374–378, https://doi.org/10.1038/nclimate1731.
Trenberth, K. E., 2011: Changes in precipitation with climate change. Climate Res., 47, 123–138, https://doi.org/10.3354/cr00953.
Wang, D., G. Wang, and E. N. Anagnostou, 2007: Evaluation of canopy interception schemes in land surface models. J. Hydrol., 347, 308–318, https://doi.org/10.1016/j.jhydrol.2007.09.041.
Weltzin, J. F., M. E. Loik, S. Schwinning, and D. Williams, 2003: Assessing the response of terrestrial ecosystems to potential changes in precipitation. BioScience, 53, 941–952, https://doi.org/10.1641/0006-3568(2003)053[0941:ATROTE]2.0.CO;2.
Xia, Y., D. Mocko, M. Huang, B. Li, M. Rodell, K. E. Mitchell, X. Cai, and M. B. Ek, 2017: Comparison and assessment of three advanced land surface models in simulating terrestrial water storage components over the United States. J. Hydrometeor., 18, 625–649, https://doi.org/10.1175/JHM-D-16-0112.1.
Ye, S., H.-Y. Li, L. R. Leung, J. Guo, Q. Ran, Y. Demissie, and M. Sivapalan, 2017: Understanding flood seasonality and its temporal shifts within the contiguous United States. J. Hydrometeor., 18, 1997–2009, https://doi.org/10.1175/JHM-D-16-0207.1.
Zhang, B., Y. Xia, B. Long, M. Hobbins, X. Zhao, C. Hain, Y. Li, and M. C. Anderson, 2020: Evaluation and comparison of multiple evapotranspiration data models over the contiguous United States: Implications for the next phase of NLDAS (NLDAS-Testbed) development. Agric. For. Meteor., 280, 107 810–107 821, https://doi.org/10.1016/j.agrformet.2019.107810.