Confidential manuscript submitted to Geophysical Research Letters

Does soil moisture affect warm season precipitation over the Southern Great Plains? 1

J. Welty1, X. Zeng1 2

1Department of Hydrology and Atmospheric Sciences, University of Arizona, Tucson, Arizona, 3 USA 4

Corresponding author: J. Welty (welty@email.arizona.edu) 5

Key Points: 6

• Relationship between morning soil moisture and subsequent convective precipitation is 7 modulated by advection 8

• Precipitation is amplified over drier (wetter) soils when advection is limited (enhanced) 9

• Different pathways are provided for precipitation amplification for weak and strong 10 advection regimes 11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

Confidential manuscript submitted to Geophysical Research Letters

Abstract 28

Numerous observational and modeling studies have addressed the impact of soil moisture on 29 subsequent precipitation (primarily its initiation), yet consensus remains elusive. Here we 30 quantify the relationship between soil moisture and precipitation amplification over the U.S. 31 Southern Great Plains. Warm season (June-September) days for the 2002-2011 period (with 32 ~1220 total days) are partitioned into low, medium, and high synoptic advection regimes, among 33 which certain days are identified as convective event days based on simple criteria. We find that 34 antecedent soil moisture conditions are negatively correlated with subsequent afternoon 35 precipitation magnitude for low advection regimes, but this correlation becomes positive for high 36 advection regimes. In contrast, this correlation is markedly reduced in magnitude and becomes 37 insignificant when all regime days are considered. These results are also confirmed by analyzing 38 the precipitation histogram and diurnal cycle. Furthermore, different pathways are provided for 39 precipitation amplification for low and high advection regimes. 40

41

1 Introduction 42

Land-atmosphere coupling as manifested in soil moisture (SM)–precipitation (P) 43 interactions has been the subject of extensive studies for decades, but there are still large 44 discrepancies between model/reanalysis representation of land-atmosphere coupling (e.g. 45 precipitation recycling) and observations [Ruiz-Barradas and Nigam, 2005; Zeng et al., 2010]. 46 While it is straightforward to understand the impact of P on SM, it remains a challenge to 47 understand the feedback of SM to P. 48

From the Global Land-Atmosphere Coupling Experiment (GLACE), boreal summer 49 model ensemble results suggested the existence of a number of “hot spots” of regionally-50 enhanced land-atmosphere coupling [Koster et al., 2004], one being the U.S. Southern Great 51 Plains (SGP). Findell et al. [2011] utilized the North American Regional Reanalysis (NARR) 52 data to assert that afternoon convective precipitation is positively correlated with pre-noon 53 evaporative fraction most notably over the eastern U.S. and Mexico, and that the positive 54 feedback is primarily modulated by precipitation frequency (triggering) rather than intensity 55 (amplification). Additionally, they posited that the surface wields little influence in the nature of 56 afternoon convective precipitation west of the Mississippi (including the SGP). Froidevaux et al. 57 [2014] employed an idealized cloud-resolving model to demonstrate that background horizontal 58 advection results in preferential precipitation occurrence over wet patches and that suppression 59 of the background influence yields preferential triggering over dry patches. 60 In addition to these model and reanalysis studies, there is also an abundance of 61 observational studies that investigated the behavior of land-atmosphere interaction and SM-P 62 coupling. Findell and Eltahir [2003a,b] isolated what have been coined as the “dry-soil 63 advantage” and “wet-soil advantage” regimes in the context of convective precipitation 64 triggering by utilizing a framework that assimilates information about low-level humidity and 65 low- to mid-tropospheric lapse rates. D’Odorico and Porporato [2004] found that there is a 66 positive SM-P relationship, primarily via triggering (not amplification) over Illinois. Alfieri et al. 67 [2008] found that, though there are regions of positive and negative temporal SM-P coupling, 68 there is no appreciable evidence of soil moisture influence on subsequent convective or 69 stratiform rainfall intensity over the upper Midwest. Taylor et al. [2012] performed a global 70 analysis of observational data and found that, overall, afternoon precipitation occurs 71

Confidential manuscript submitted to Geophysical Research Letters

preferentially over drier soils, with some regions exhibiting more robust coupling than others. 72 Tuttle and Salvucci [2016] found that the correlation between precipitation triggering and soil 73 moisture over the U.S. is generally positive in the more arid western regions and negative in the 74 more humid east (with no statistically significant feedback over the SGP on seasonal timescales). 75 A number of recent studies have also addressed the SM-P relationship over the SGP, some of 76 which indicated that coupling is either very weakly positive [Song et al. 2016] or negligible 77 [Phillips and Klein, 2014; Ford et al., 2015b]. 78

As is evident, the literature is laden with varying conclusions regarding the nature of 79 land-atmosphere coupling as encapsulated in SM-P relationships, many of which are 80 inconclusive and/or contradictory to others. Here we focus on the SGP (rather than other regions, 81 or extending the study globally) for two reasons: i) questions have emerged in recent years about 82 the SGP’s designation as a “hot spot” region of land-atmosphere coupling, and ii) the 83 observational network and data availability/diversity across the domain is unparalleled, allowing 84 for more thorough and extensive exploration of SM-P, and more broadly, land-atmosphere 85 coupling behavior. The goal of this study is to address if and how morning soil moisture affects 86 afternoon precipitation magnitude (“magnitude” used hereafter to encompass accumulation, 87 maximum, and intensity) over the SGP domain. Section 2 describes the data and methodology, 88 Section 3 presents the results, and Section 4 gives the conclusions. 89

90

2 Data and Methods 91

2.1 Data description 92

The study period comprises June-September (JJAS), when active, diurnal, local-scale 93 convection is the most prevalent, for the years 2002-2011 over the United States SGP (~1220 94 total days). A number of quantities are calculated using station data, radiosonde data, and 95 combined radar and gauge data to represent the processes occurring across the domain for each 96 given day. First and foremost, the ground-based U.S. Department of Energy (DOE) Atmospheric 97 Radiation Measurement (ARM) program data are indispensable in calculating surface and 98 atmospheric variables throughout the study period. Soil moisture measurements are from the Soil 99 Water and Temperature Profiling Systems (SWATS) across the SGP domain. Each SWATS 100 location consists of an east and west profile with sensors at eight different depths ranging from 5 101 cm to 175 cm below ground. Here we use soil moisture at 5 cm depth, corresponding to the 102 timescales of interest (sub-daily to daily). Volumetric soil moisture content values are averaged 103 across all available SWATS stations during the morning hours (0700 – 1100CST) for each day. 104 Then, these averaged values are used to calculate the seasonal standardized anomalies. Soil 105 temperatures at 5 cm depth are station- and time-averaged over the 1100-2300 CST period for 106 each day. 107

Turbulent and net radiation fluxes are retrieved from the SGP30BAEBBR value-added 108 product (VAP) which utilizes data from the Energy Balance Bowen Ration (EBBR) stations 109 across the domain. EBBR data are used instead of eddy covariance data (unavailable before 110 September of 2003) due to continuity throughout the time period. These fluxes are station- and 111 time-averaged from 1100-2300 CST for each day. PBL heights are retrieved from the 112 SGPPBLHTSONDE1MCFARL VAP with the PBL height estimates from Liu and Liang [2010]. 113 Other quantities such as the convective available potential energy (CAPE) and lifting 114 condensation level (LCL) are calculated using radiosonde data from the central facility (CF) in 115

Confidential manuscript submitted to Geophysical Research Letters

Lamont, OK. CAPE was calculated using a modified script following pseudo-adiabatic ascent 116 with ice processes included [George Bryan, UCAR]. LCL was calculated using the exact, 117 analytical formulation developed in Romps [2017]. 118

We utilize Stage-IV precipitation data which comprises merged gauge and radar data 119 over the continental U.S., regridded at 4x4 km resolution. Hourly values are used for this study 120 over a roughly 3°x3° SGP domain (35.25-38.25°N, 96-99°W) which corresponds to the 121 approximate domain used for regime partitioning via the Modern-Era Retrospective Analysis for 122 Research and Applications, version 2 (MERRA-2) integrated water vapor tendencies (discussed 123 next). 124

2.2 Synoptic advection and convective partitionings 125

Each of the warm season days are placed into low, medium, or high advection regimes by 126 utilizing the MERRA-2 data [Gelaro et al., 2017]. MERRA-2 is the latest, state-of-the-art 127 reanalysis product produced by NASA’s Global Modeling and Assimilation Office (GMAO) 128 with several enhancements over the previous version, including aerosol assimilation and 129 improvement of cryospheric representation. Included in the MERRA-2 output is a suite of 130 tendency terms for a myriad of atmospheric variables. These tendency terms represent the 131 various model constituent contributions to an overall given tendency. For instance, the integrated 132 water vapor tendency is affected by dynamics (i.e., water vapor advection), physics (including 133 moist processes and turbulence), chemistry, and analysis (data ingestion). To characterize the 134 influence of background synoptic conditions, this tendency due to dynamics is employed. 135

To begin, the hourly integrated water vapor tendency due to dynamics is retrieved over 136 all MERRA-2 grids within an approximate 3°x3° domain covering the SGP study area (Figure 137 1). The magnitudes of these values are then averaged over the domain over each local day. The 138 use of absolute values helps mitigate potential misclassifications. For instance, if an idealized 139 meridional line of convergence were to bisect the domain, negative and positive contributions to 140 the integrated water vapor tendency on opposite sides of the boundary could partially or largely 141 cancel out. The average of actual tendency values may lead to a low synoptic advection regime 142 day, though, in reality, dynamic factors may be relatively active. 143

Then, daily domain-averaged variable (absolute value of integrated water vapor tendency 144 due to dynamics) tritiles are used to partition each JJAS season into low, medium, and high 145 synoptic advection regime days. Thus, there are roughly 40-42 low, medium, and high regime 146 days, respectively, for each year 2002-2011. This yields a total of approximately 400-420 low, 147 medium, and high regime days, respectively, for the cumulative study period (~1220 total days). 148

Following the stratification of days into low, medium, and high advection regimes, each 149 day is classified further as a convective precipitation day (CD) or non-convective day (NoCD) by 150 utilizing the Stage-IV precipitation data. Figure S1 shows the convective precipitation 151 accumulations for JJAS 2006 over the domain as an example. If the domain receives 30 mm of 152 precipitation or more during the convective hours (defined as 1100 – 2300 CST) and less than 30 153 mm during the non-convective hours (defined as 0700 – 1100CST and 2300 – 2400 CST), then 154 this day is classified as CD. The second part of the requirement insures the exclusion of days that 155 are punctuated by precipitation episodes emerging later in the diurnal cycle and are more likely 156 to be associated with or reinforced by a burgeoning nocturnal low-level jet or other non-157 convective sources. Here we focus on CDs. 158

After pooling all of the days together for the 10-year period (~1220 total days), there are 159 a total of 161 CDs (~13% of total), of which 70, 63, and 28 are in low, medium, and high 160

Confidential manuscript submitted to Geophysical Research Letters

advection regimes, respectively. In the next section, various daily quantities are analyzed based 161 on CD and advection regimes. 162

163

3 Results 164

3.1 Observed soil moisture effect on precipitation 165

Before addressing specific relationships between soil moisture and precipitation on 166 convective days (CDs) across the three advection regimes, it is important to get a sense for the 167 overall synoptic configuration of these regimes. Figure 1 depicts the composite 500 mb 168 geopotential heights and 850 mb winds for the three advection regimes. The low regime is 169 characterized by expansive ridging over the southern continental U.S. (CONUS) with axis 170 oriented over the Great Plains and low-level anticyclone centered to the immediate east of the 171 SGP domain. The depth and proximity of synoptic high pressure would imply enhanced net 172 radiation (less synoptic cloud coverage). The progression from low regime to medium and high 173 regimes is characterized by an attenuation of the ridge in conjunction with an eastward shift of 174 the anticyclone. 175

Again, the foremost questions are: i) do SM and P exhibit significant time-delayed 176 correlation on CDs and ii) is any apparent relationship modified by advection regime? To 177 explore the impact of morning SM anomalies on subsequent convective precipitation 178 accumulation, simple linear regression is performed for all CDs pooled as a whole, then across 179 each of the dynamic regimes. The natural logarithm of precipitation is used following previous 180 studies [e.g., Koster et al., 2004] in order to reduce noise. 181

When examining all of the CDs as a whole (Fig. 2a), there is no significant relationship 182 between morning SM and subsequent afternoon precipitation accumulations. This confirms 183 results from previous studies. For instance, Ford et al. [2015b] indicated that unorganized 184 convection is preferentially triggered over drier soils, but that there is no significant correlation 185 between morning (0900 LST) SM and any of their quantitative metrics for unorganized 186 convective precipitation event characteristics (accumulation, duration, size). A number of other 187 studies also found no evidence of significant coupling between soil moisture and precipitation 188 over the SGP [Findell et al. 2011; Taylor et al., 2012; Phillips and Klein, 2014]. Song et al. 189 [2016] indicated a weak positive correlation between SM and peak afternoon precipitation for 190 both “dry-coupling” and “wet-coupling” regimes. 191 192 193 194 195 196 197 198

199 200

201

202

203

Confidential manuscript submitted to Geophysical Research Letters

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

Figure 1. Composite synoptic maps of 500 mb geopotential height (shading) and 850 mb winds 230 (vectors) for low, medium, and high synoptic advection regimes. Each regime represents the 231 average of all individual daily means (at 0000, 0600, 1200, and 1800 CST) for that regime. The 232 SGP domain used in this study is marked in each panel (black square). 233

234 235

Confidential manuscript submitted to Geophysical Research Letters

Figure 2. Relationship between the logarithm of precipitation accumulations (mm) from 1100–236 2300 CST and antecedent standardized soil moisture anomalies from 0700–1100 CST over 237 stations across the SGP domain for a) all CDs, b) low advection regime CDs, c) medium regime 238 CDs, and d) high regime CDs. Correlation coefficients (r) significant at p=0.05 are marked with 239 an asterisk (*), and n refers to the number of days. 240

241 However, by quantitatively accounting for the relative strength of the dynamic 242

(advection) regime governing SGP conditions on given CDs, we find statistically significant and 243 opposing correlations between morning SM and subsequent P accumulations for the low and 244 high regimes (Fig. 2b,d). Under the low regime, there is a clear negative correlation between 245 morning SM and convective P accumulations, whereas the correlation is positive under high 246 regime conditions. There is no significant relationship between the two quantities for the medium 247 regime (or all regimes together). These results suggest that, during the course of regional SM-P 248 coupling studies, it is essential that daily advection be accounted for in order to mitigate the 249 complications that may arise from synoptic influence yielding spurious or negligible 250 relationships. 251

To demonstrate the robustness of this new finding, Tables S1 and S2 (supplementary 252 material) present similar analyses, except for precipitation maximum and precipitation intensity, 253 respectively. The maximum is defined as the logarithm of the maximum hourly 4 km x 4 km 254 pixel value for precipitation accumulation over the SGP domain during the 1100-2300 CST 255 period, and precipitation intensity is the total accumulation during the 1100-2300 CST period 256 divided by the number of rain pixels (P>0). Both of these tables further support the opposite SM-257 P relationships for the low and high regimes. 258

Confidential manuscript submitted to Geophysical Research Letters

In order to address the issue of independence of CDs, a separate test is performed. In this 259 test, if two or more CDs occurred consecutively, then only the first of the series of CDs is 260 retained. This would reduce the number of CDs for the low, medium, and high advection regimes 261 from 70, 63, and 28 to 55, 55, and 27, respectively. Nevertheless, there would be no degradation 262 of the relationship exhibited between morning SM and subsequent P accumulation for the low 263 and high regimes. In fact, the magnitude of the correlation coefficients would increase slightly 264 for the low and high advection regimes, with values of -0.53 and 0.49, respectively (versus -0.51 265 and 0.48 in Figure 2b,d). 266

The finding in Figure 2 is further supported by the analysis of precipitation histograms 267 and diurnal cycle. Figure S2 shows that the spread of convective precipitation accumulations is 268 much larger and the majority of CDs (i.e., 48 of the 70 CDs, or 69%) occur over dry soils during 269 low advection regimes, and the opposite are true under high regimes. Figure S3 demonstrates 270 that, under the low advection regime, hourly precipitation rate is conspicuously elevated during 271 the afternoon convective hours, on average, over drier soils. Conversely, under the high 272 advection regime, afternoon precipitation rates are maximized over wetter soils and relatively 273 attenuated over drier soils. 274

3.2 Mechanisms associated with contrasting SM-P interactions 275

3.2.1 P amplification under low synoptic advection regime 276

In order to interpret the opposite SM-P correlations across the low and high advection 277 regimes, it is important to address the corresponding relationships between a number of variables 278 and P accumulations over the same regimes. The natural next step is to examine the relationship 279 between soil temperature and P accumulations. Enhanced soil temperatures during the diurnal 280 cycle indicate any combination of limited soil moisture, enhanced radiative input, and/or warmer 281 overlying atmosphere. Table 1 shows that soil temperature is most strongly correlated with P 282 accumulations, with positive and negative correlations for the low and high advection regimes, 283 respectively. 284

Net radiation is also significantly correlated with P accumulations for the low and high 285 regimes (Table 1). However, the correlation is much lower than that between soil temperature 286 and P accumulation for the low regime. To understand this difference, a linear regression with 287 SM anomalies and net radiation as the predictors and soil temperature as the predictand can be 288 done for low regime CDs. This yields a correlation coefficient of r=0.78 with both regression 289 coefficients highly significant at p = 0.05. Therefore, both SM deficits and net radiation 290 enhancements are significant drivers of increased soil temperatures under the low regime, and, 291 consequently, convective P accumulations when synoptic advection is limited. 292

293 294

295

296

297

298 299

Confidential manuscript submitted to Geophysical Research Letters

Table 1. 300 Relationship Between Variables and Accumulated Precipitation Across Regimes 301 P vs. All Low Medium High Morning SM Anomaly -0.06 -0.51* 0.12 0.48* Soil T 0.17* 0.48* 0.05 -0.45* Net Radiation -0.01 0.29* -0.16 -0.43* PBLHd/LCLd 0.15 0.42* 0.21 -0.44* 0-1 km T 0.13 0.41* 0.04 -0.20 CAPE 0.26* 0.41* 0.30* -0.08 SH 0.02 0.19 -0.10 -0.21 LH 0.00 0.12 -0.04 -0.20 Bowen Ratio 0.02 0.04 0.02 -0.04

Note. Correlation coefficients between the logarithm of precipitation accumulations (mm) from 302 1100–2300 CST and various quantities for CDs for all, low, medium, and high advection 303 regimes. SM anomalies are averaged from 0700-1100 CST. Soil temperature, net radiation, SH, 304 LH, and Bowen ratio are averaged from 1100-2300 CST. CAPE and 0-1 km T are computed 305 from the ~1200 CST sounding, and the PBLHd and LCLd are calculated as the respective 306 differences between ~0600 and ~1200 CST soundings (to capture the diurnal growth of each). 307 Correlation coefficients significant at p=0.05 are marked with an asterisk (*). 308

309 Net radiation can be decomposed into sensible (SH), latent (LH), and ground heat fluxes, 310

but neither these fluxes nor the Bowen ratio of SH/LH have statistically significant correlations 311 with P accumulations (Table 1). SH is the primary driver of near-surface temperature and 312 planetary boundary layer (PBL) growth, while LH affects near-surface humidity. The change of 313 surface air temperature and humidity also affects the computation of lifting condensation level 314 (LCL). To quantify the relative change of PBL and LCL, Table 1 includes a quantity, 315 PBLHd/LCLd, which is the ratio of the growth of diurnal PBL height to the growth of the LCL 316 height (1200 CST LCL minus 0600 CST LCL). Higher values signify that the early diurnal 317 growth of PBL height is relatively large compared to the early diurnal increase of the LCL. 318

Consistent with the relationship between soil temperature and P accumulation within the 319 low regime, Table 1 indicates that there is also a significant positive correlation between 320 PBLHd/LCLd and P accumulation. This relationship suggests that accelerated and enhanced 321 early diurnal PBL growth (relative to the LCL) is associated with larger convective P 322 accumulations over the SGP when advection (dynamic influence) is suppressed. Further 323 confirmation of this relationship is the 0-1 km temperature average (calculated from the 1200 324 CST sounding) which is also positively correlated with P accumulation under the low regime. 325

These results differ from the findings in Ford et al.[2015b]. They found that, for 326 unorganized convective events near Lamont, OK, larger PBL growth from 0600 to 1200 CST - 327 which they associated with drier soils – was negatively correlated with P accumulation, size, and 328 duration. They posited that deeper PBL growth indicates lower relative humidity eventually 329 yielding less afternoon P. However, their contradictory conclusions might change if different 330 advection regimes were separately considered in their study. Indeed our observation-based 331 results from the low regime (Table 1) suggest that the limiting factor on P magnitude is not PBL 332 relative humidity but rather the strength/frequency of thermals when advection is suppressed. 333

Confidential manuscript submitted to Geophysical Research Letters

The strength of thermals is directly tied to soil temperature, which is enhanced by increases in 334 net radiation and decreases in soil moisture (also see Table S3). 335

Yin et al. [2015] utilized a simplified zero-dimensional mixed-layer model to address the 336 triggering and intensity of moist convection by employing PBL height / LCL height dynamics in 337 conjunction with CAPE under the assertion that PBL/LCL dynamics are not alone sufficient to 338 explain the nature of subsequent convection. In this vein, the CAPE relationship with P 339 accumulation is also presented in Table 1. It is clear that CAPE is most strongly positively 340 correlated with P accumulation (and P maximum, in Table S1) under the low regime. Thus, 341 under low regime conditions, PBL growth propels the PBL top up to the LCL, after which point 342 enhanced CAPE allows for further amplification of P accumulation via more pronounced ascent 343 and latent heat release. 344

3.2.2 P amplification under high advection regime 345

While thermal forcing is the primary uplift mechanism under low advection regime, we 346 find that another mechanism is more influential under high advection regime. Corresponding to 347 the positive SM-P correlation under the high regime, the correlation of both soil temperature and 348 net radiation with P accumulation is significant and negative. Both the attenuation of net 349 radiation and increase of SM contribute to cooler soil temperatures and lower SH which leads to 350 slower diurnal PBL growth. Indeed the PBLHd/LCLd ratio is negatively correlated with P 351 accumulations under the high regime. In other words, under high advection regime, shallower 352 and slowly expanding PBLs are associated with enhanced P accumulations. These findings align 353 with those of Ford et al. [2015b] wherein shallower PBLs are associated with enhanced 354 convective event accumulation, duration, and size. The reasoning for this relationship is that 355 relative humidity is amplified within a shallower PBL. 356

In the absence of thermal forcing for the high advection regime, we posit that enhanced 357 moisture advection into the region and convergence would drive PBL air above the LCL, 358 allowing for condensation, latent heat release, and P formation. Thus, under high advection 359 regime, lift is not a limiting factor in P amplification, as PBL air is readily lifted by large-scale 360 forcing. 361

Song et al. [2016] found that wet-coupling days are preceded by an increase in horizontal 362 moisture convergence over the SGP. Wet-coupling days are characterized by an enhancement of 363 convective initiation likelihood by wetter soils and PBLs with higher relative humidity. 364 Therefore, while wet-coupling in that study and others [Findell and Eltahir, 2003a,b] typically 365 refers to convective initiation, our results in Table 1 suggest that similar concepts apply to the 366 daily convective amplification as well. In that vein, P amplification under the high regime does 367 not occur via thermal forcing, but instead by moisture transport and convergence within the SGP 368 domain. When advection and convergence of moisture is pronounced, air from shallower SGP 369 PBLs with correspondingly high relative humidity values more readily reach condensation upon 370 being dynamically lifted. 371

Thus, under the high regime, the limiting factor is PBL depth and relative humidity, 372 which determines the level at which condensation first occurs upon synoptic uplift. However, 373 under the low regime, the limiting factor for P amplification is the frequency and/or vigor with 374 which PBL parcels over the SGP can reach the LCL in the absence of dynamic forcing. Such a 375 mechanism necessitates thermal forcing, which is magnified by drier soils, enhanced net 376 radiation, increased soil (surface) temperatures, and more pronounced PBL growth. Overall, P 377

Confidential manuscript submitted to Geophysical Research Letters

appears to be modulated by the rate/vigor of the initial kick (lift) in the low regime, whereas P 378 modulation occurs after the initial kick in the high regime. 379

380

4 Conclusions 381

Observational evidence of contrasting morning SM and subsequent P magnitude 382 correlations modulated by synoptic influence over the SGP has been presented, including 383 potential pathways for this modulation. Though many prior studies focused on the relationship 384 between SM and convective P triggering, this study primarily focuses on SM and subsequent P 385 magnitude, as encapsulated by the metrics of accumulation, maximum, and intensity. 386

Overall, there is evidence of negative correlation between morning SM and afternoon 387 convective P accumulations over the SGP domain under relatively limited synoptic influence. 388 These findings suggest a “dry-coupling” pathway wherein thermal forcing, as manifested in 389 enhanced soil temperatures and deeper, drier PBLs, propels low-level air to the LCL. Under 390 more enhanced synoptic influence, SM and P accumulations become positively correlated, 391 indicating that large-scale forcing is the potentially necessary pathway by which uplift and ascent 392 occur over the SGP. 393

Part of the utility of the current study is its reproducibility and simplicity. Therefore, it 394 would be beneficial to examine the SM-P interaction over various regions across the globe to see 395 i) if/where there are any significant relationships and ii) if the partitioning by water vapor 396 advection changes the sign of any coupling relationships that may exist. Secondly, the nature of 397 this study would also allow it to be easily applied to global climate data, including reanalyses, 398 satellite data, and model output. While numerous studies have employed model simulations to 399 investigate SM-P coupling [e.g., Froidevaux et al., 2014], it would be useful to perform similar 400 studies with the lens of advection partitioning on regional and global scales. 401

402

Acknowledgments 403

This work was supported by the NASA SMAP (NNX16AN37G) and MAP (NNX14AM02G) 404 programs. MERRA-2 data were obtained from the NASA Goddard Earth Sciences Data and 405 Information Services Center Simple Subset Wizard (https://disc.gsfc.nasa.gov/SSW/). Data from 406 the U.S. DOE ARM program were obtained from (https://www.arm.gov/data). Stage-IV 407 precipitation data were obtained from the Earth Observing Laboratory data archive 408 (https://data.eol.ucar.edu/dataset/21.093), operated and maintained by NCAR/UCAR. 409

410

411

412

413

414

415

416

417

Confidential manuscript submitted to Geophysical Research Letters

References 418

419

Alfieri, L., Claps, P., D’Odorico, P., Laio, F., & Over, T. M. (2008). An Analysis of the Soil 420 Moisture Feedback on Convective and Stratiform Precipitation. Journal of 421 Hydrometeorology, 9(2), 280–291. https://doi.org/10.1175/2007JHM863.1 422

Berg, A., Findell, K., Lintner, B. R., Gentine, P., & Kerr, C. (2013). Precipitation Sensitivity to 423 Surface Heat Fluxes over North America in Reanalysis and Model Data. Journal of 424 Hydrometeorology, 14(3), 722–743. https://doi.org/10.1175/JHM-D-12-0111.1 425

Betts, A. K., & Ball, J. H. (1998). FIFE Surface Climate and Site-Average Dataset 1987–89. 426 Journal of the Atmospheric Sciences, 55(7), 1091–1108. https://doi.org/10.1175/1520-427 0469(1998)055<1091:FSCASA>2.0.CO;2 428

Bolton, D. (1980). The Computation of Equivalent Potential Temperature. Monthly Weather 429 Review, 108(7), 1046–1053. https://doi.org/10.1175/1520-430 0493(1980)108<1046:TCOEPT>2.0.CO;2 431

Chen, L., & Dirmeyer, P. A. (2017). Impacts of Land-Use/Land-Cover Change on Afternoon 432 Precipitation over North America. Journal of Climate, 30(6), 2121–2140. 433 https://doi.org/10.1175/JCLI-D-16-0589.1 434

D’Odorico, P., & Porporato, A. (2004). Preferential states in soil moisture and climate dynamics. 435 Proceedings of the National Academy of Sciences of the United States of America, 101(24), 436 8848–51. https://doi.org/10.1073/pnas.0401428101 437

Dirmeyer, P. A. (2001). An Evaluation of the Strength of Land–Atmosphere Coupling. Journal 438 of Hydrometeorology, 2(4), 329–344. https://doi.org/10.1175/1525-439 7541(2001)002<0329:AEOTSO>2.0.CO;2 440

Eltahir, E. A. B. (1998). A Soil Moisture-Rainfall Feedback Mechanism: 1. Theory and 441 observations. Water Resources Research, 34(4), 765–776. 442 https://doi.org/10.1029/97WR03499 443

Findell, K. L., & Eltahir, E. A. B. (2003). Atmospheric Controls on Soil Moisture–Boundary 444 Layer Interactions. Part I: Framework Development. Journal of Hydrometeorology, 4(3), 445 552–569. https://doi.org/10.1175/1525-7541(2003)004<0552:ACOSML>2.0.CO;2 446

Findell, K. L., & Eltahir, E. A. B. (2003). Atmospheric Controls on Soil Moisture–Boundary 447 Layer Interactions. Part II: Feedbacks within the Continental United States. Journal of 448 Hydrometeorology, 4(3), 570–583. https://doi.org/10.1175/1525-449 7541(2003)004<0570:ACOSML>2.0.CO;2 450

Findell, K. L., & Eltahir, E. A. B. (2003). Atmospheric controls on soil moisture-boundary layer 451 interactions: Three-dimensional wind effects. Journal of Geophysical Research, 108(D8), 452 8385. https://doi.org/10.1029/2001JD001515 453

Findell, K. L., Gentine, P., Lintner, B. R., & Kerr, C. (2011). Probability of afternoon 454 precipitation in eastern United States and Mexico enhanced by high evaporation. Nature 455 Geoscience, 4(7), 434–439. https://doi.org/10.1038/ngeo1174 456

Ford, T. W., Rapp, A. D., & Quiring, S. M. (2015). Does Afternoon Precipitation Occur 457 Preferentially over Dry or Wet Soils in Oklahoma? Journal of Hydrometeorology, 16(2), 458 874–888. https://doi.org/10.1175/JHM-D-14-0005.1 459

Ford, T. W., Rapp, A. D., Quiring, S. M., & Blake, J. (2015). Soil moisture–precipitation 460 coupling: observations from the Oklahoma Mesonet and underlying physical mechanisms. 461

Confidential manuscript submitted to Geophysical Research Letters

Hydrology and Earth System Sciences, 19(8), 3617–3631. https://doi.org/10.5194/hess-19-462 3617-2015 463

Froidevaux, P., Schlemmer, L., Schmidli, J., Langhans, W., & Schär, C. (2014). Influence of the 464 Background Wind on the Local Soil Moisture–Precipitation Feedback. Journal of the 465 Atmospheric Sciences, 71(2), 782–799. https://doi.org/10.1175/JAS-D-13-0180.1 466

Gelaro, R., McCarty, W., Suárez, M. J., Todling, R., Molod, A., Takacs, L., … Zhao, B. (2017). 467 The Modern-Era Retrospective Analysis for Research and Applications, Version 2 468 (MERRA-2). Journal of Climate, 30(14), 5419–5454. https://doi.org/10.1175/JCLI-D-16-469 0758.1 470

Gentine, P., Holtslag, A. A. M., D’Andrea, F., & Ek, M. (2013). Surface and Atmospheric 471 Controls on the Onset of Moist Convection over Land. Journal of Hydrometeorology, 14(5), 472 1443–1462. https://doi.org/10.1175/JHM-D-12-0137.1 473

Guillod, B. P., Orlowsky, B., Miralles, D. G., Teuling, A. J., & Seneviratne, S. I. (2015). 474 Reconciling spatial and temporal soil moisture effects on afternoon rainfall. Nature 475 Communications, 6, 6443. https://doi.org/10.1038/ncomms7443 476

Hsu, H., Lo, M.-H., Guillod, B. P., Miralles, D. G., & Kumar, S. (2017). Relation between 477 precipitation location and antecedent/subsequent soil moisture spatial patterns. Journal of 478 Geophysical Research: Atmospheres, 122(12), 6319–6328. 479 https://doi.org/10.1002/2016JD026042 480

Koster, R. D., Dirmeyer, P. A., Guo, Z., Bonan, G., Chan, E., Cox, P., … GLACE Team. (2004). 481 Regions of Strong Coupling Between Soil Moisture and Precipitation. Science, 305(5687), 482 1138–1140. https://doi.org/10.1126/science.1100217 483

Koster, R. D., Suarez, M. J., Higgins, R. W., & Van den Dool, H. M. (2003). Observational 484 evidence that soil moisture variations affect precipitation. Geophysical Research Letters, 485 30(5), n/a-n/a. https://doi.org/10.1029/2002GL016571 486

Liu, S., & Liang, X.-Z. (2010). Observed Diurnal Cycle Climatology of Planetary Boundary 487 Layer Height. Journal of Climate, 23(21), 5790–5809. 488 https://doi.org/10.1175/2010JCLI3552.1 489

Phillips, T. J., & Klein, S. A. (2014). Land-atmosphere coupling manifested in warm-season 490 observations on the U.S. southern great plains. Journal of Geophysical Research: 491 Atmospheres, 119(2), 509–528. https://doi.org/10.1002/2013JD020492 492

Romps, D. M. (2017). Exact expression for the lifting condensation level. Journal of the 493 Atmospheric Sciences, JAS-D-17-0102.1. https://doi.org/10.1175/JAS-D-17-0102.1 494

Ruiz-Barradas, A., Nigam, S., Ruiz-Barradas, A., & Nigam, S. (2005). Warm Season Rainfall 495 Variability over the U.S. Great Plains in Observations, NCEP and ERA-40 Reanalyses, and 496 NCAR and NASA Atmospheric Model Simulations. Journal of Climate, 18(11), 1808–497 1830. https://doi.org/10.1175/JCLI3343.1 498

Santanello, J. A., Roundy, J., & Dirmeyer, P. A. (2015). Quantifying the Land–Atmosphere 499 Coupling Behavior in Modern Reanalysis Products over the U.S. Southern Great Plains. 500 Journal of Climate, 28(14), 5813–5829. https://doi.org/10.1175/JCLI-D-14-00680.1 501

Seneviratne, S. I., Wilhelm, M., Stanelle, T., van den Hurk, B., Hagemann, S., Berg, A., … 502 Smith, B. (2013). Impact of soil moisture-climate feedbacks on CMIP5 projections: First 503 results from the GLACE-CMIP5 experiment. Geophysical Research Letters, 40(19), 5212–504 5217. https://doi.org/10.1002/grl.50956 505

Confidential manuscript submitted to Geophysical Research Letters

Sisterson, D. L., Peppler, R. A., Cress, T. S., Lamb, P. J., & Turner, D. D. (2016). The ARM 506 Southern Great Plains (SGP) Site. Meteorological Monographs, 57, 6.1-6.14. 507 https://doi.org/10.1175/AMSMONOGRAPHS-D-16-0004.1 508

Song, H.-J., Ferguson, C. R., & Roundy, J. K. (2016). Land–Atmosphere Coupling at the 509 Southern Great Plains Atmospheric Radiation Measurement (ARM) Field Site and Its Role 510 in Anomalous Afternoon Peak Precipitation. Journal of Hydrometeorology, 17(2), 541–556. 511 https://doi.org/10.1175/JHM-D-15-0045.1 512

Taylor, C. M., de Jeu, R. A. M., Guichard, F., Harris, P. P., & Dorigo, W. A. (2012). Afternoon 513 rain more likely over drier soils. Nature, 489(7416), 423–426. 514 https://doi.org/10.1038/nature11377 515

Tuttle, S., & Salvucci, G. (2016). Empirical evidence of contrasting soil moisture-precipitation 516 feedbacks across the United States. Science (New York, N.Y.), 352(6287), 825–8. 517 https://doi.org/10.1126/science.aaa7185 518

Wei, J., Dickinson, R. E., & Chen, H. (2008). A Negative Soil Moisture–Precipitation 519 Relationship and Its Causes. Journal of Hydrometeorology, 9(6), 1364–1376. 520 https://doi.org/10.1175/2008JHM955.1 521

Westra, D., Steeneveld, G. J., & Holtslag, A. A. M. (2012). Some Observational Evidence for 522 Dry Soils Supporting Enhanced Relative Humidity at the Convective Boundary Layer Top. 523 Journal of Hydrometeorology, 13(4), 1347–1358. https://doi.org/10.1175/JHM-D-11-524 0136.1 525

Xie, S., McCoy, R. B., Klein, S. A., Cederwall, R. T., Wiscombe, W. J., Jensen, M. P., … 526 Turner, D. D. (2010). CLOUDS AND MORE: ARM Climate Modeling Best Estimate Data. 527 Bulletin of the American Meteorological Society, 91(1), 13–20. 528 https://doi.org/10.1175/2009BAMS2891.1 529

Yin, J., Albertson, J. D., Rigby, J. R., & Porporato, A. (2015). Land and atmospheric controls on 530 initiation and intensity of moist convection: CAPE dynamics and LCL crossings. Water 531 Resources Research, 51(10), 8476–8493. https://doi.org/10.1002/2015WR017286 532

Zeng, X., Barlage, M., Castro, C., & Fling, K. (2010). Comparison of Land–Precipitation 533 Coupling Strength Using Observations and Models. Journal of Hydrometeorology, 11(4), 534 979–994. https://doi.org/10.1175/2010JHM1226.1 535

Zhang, Y., & Klein, S. A. (2010). Mechanisms Affecting the Transition from Shallow to Deep 536 Convection over Land: Inferences from Observations of the Diurnal Cycle Collected at the 537 ARM Southern Great Plains Site. Journal of the Atmospheric Sciences, 67(9), 2943–2959. 538 https://doi.org/10.1175/2010JAS3366.1 539

540 541 542 543 544 545 546 547 548 549 550 551