Impacts of irrigation-climate interactions on irrigated soybean yields in the US Arkansas Delta from 2003 to 2017

Abstract

Irrigation has been widely implemented across the globe as a mitigation strategy to combat climate change and erratic rainfall. Irrigation in the confined geographic region like the Arkansas Delta of the US has likely affected heat and moisture fluxes at the land surface with possible effects on regional climate conditions. Irrigation unquestionably benefits crop yields with direct water supplies. However, the effect of irrigation-climate interactions on Arkansas Delta crop yields remains unclear. In this study, we applied multiple satellite and climatic datasets to assess the influence of soybean irrigation in the US Arkansas Delta on the regional climate from 2003 to 2017 and how the resultant climate variability has affected soybean yields. Our findings show that soybean irrigation in the Arkansas Delta leads to statistically insignificant precipitation change and significant daytime and night time cooling during the growing season from June to August over the period of 2003–2017. Using a statistical crop yield model, we further demonstrate that such surface temperature cooling due to irrigation could enhance soybean yields as much as ∼1.13 ( $\pm$ 0.87) bu/acre, accounting for 7.78% of total soybean yields gain due to irrigation. Our results highlight the important positive effects of irrigation-climate interactions on soybean yields, which may be more important in the Arkansas Delta, given the depletion of groundwater that farmers relied on most for irrigation.

Keywords

irrigation precipitation temperature soybean crop yields Arkansas Delta

Introduction

Fossil fuel combustion has warmed the planet by ∼ 1°C compared to the preindustrial era (IPCC, 2018). Negative effects of this warming on agriculture have already occurred and are projected to intensify further (Coffel et al., 2019). To combat climate change, irrigation has been widely implemented across globe as an important mitigation strategy. The influence of irrigation on climate has a long history of investigation (Mahmood et al., 2006). Irrigation induced land use transitions can significantly affect temperature, moisture, and circulation through biogeophysical and biogeochemical processes at the land-atmosphere interface (He and Lee, 2016; Liu et al., 2021). For example, irrigated cropland expansion in Northeast China could reduce average daily surface air temperature by increasing energy partitioning towards latent rather than sensible heat fluxes (He et al., 2020). In the US High Plains, a similar response has been documented, with the cooling effect of expanding irrigation leading to a 1.2°C decrease in the average daily near-surface temperature (Adegoke et al., 2003). In contrast, Chen and Jeong (2018) found a nighttime surface warming because of irrigation in the North China Plain, which was attributable to energy storage in the wetter soil during the daytime. The annual and seasonal precipitation have been affected as well. Boucher et al. (2004) suggested increased water vapor due to irrigation could increase local precipitation over irrigated areas in the Columbia Basin and the Snake River Plain in the US. Complicating this finding, however, is another work arguing that the enhanced surface pressure due to evaporative cooling of irrigation could sustain anticyclonic air circulation, thus inhibiting precipitation over irrigated areas in Gezira Scheme in East Africa (Alter et al., 2015). These reviews suggest that the impact of irrigation on local climates differs in different geographic regions.

The impacts of irrigation-induced climate variability on crop yields are an emerging and recent exploration. Over the last several decades, most attention focused on the water supply effects of irrigation on crop yields (Szilagyi, 2018). Irrigation increases crop yields by supplying an adequate amount of water to crops when rainfall is not sufficient or timely to meet the water demands of crops (Troy et al., 2015). For example, Kukal and Irmak (2020) found that irrigation could reduce crop yield variability by 41% across U.S. through meeting crop water demand in arid/semi-arid regions, buffering within-season precipitation variability in humid/sub-humid regions, or remedying other environmental variabilities. Crop yield is sensitive to changes in temperature and precipitation (Ray et al., 2015; Zhao et al., 2017; Li et al., 2019; Lesk et al., 2020), as such the precipitation or temperature changes due to irrigation can affect crop yield. The cooling effect of irrigation shifted the high temperature thresholds, reduced heat stress, increased photosynthesis, and thereby increased maize yield in the state of Nebraska in the US (Li et al., 2020). Similarly, the irrigation and agricultural intensification in the US Midwest reduced killing degree days for crops, thus benefiting maize and soy yields (Coffel et al., 2022). While most previous studies focus on the effects of irrigation resultant mean or extreme temperature variability on crop yields, few studies assessed how changing daytime and nighttime temperature influenced crop yield due to irrigation.

The US state of Arkansas, particularly in the Delta region (Figure 1) has experienced a tremendous increase in irrigation lands. The irrigated areas in Arkansas expanded by 121% from 1978 to 1997 and continued to increase to 4.9 million acres in 2017 (Evett et al., 2003; USDA, 2017). Because of the huge water demand for agriculture, Lonoke County in the Arkansas Delta drew off an average of 584.5 million gallons of groundwater per day for irrigation in 2015 (Walkenhorst, 2018). Such massive groundwater pumping has resulted in aquifer overdraft, particularly in the Arkansas Delta, leaving large areas of the Delta as critical groundwater zones (West et al., 2020). Given the continuing global warming, the reduction in the availability of groundwater, as well as erratic rainfall and frequent summer droughts, poses immense threats to agricultural sustainability in Arkansas. A quantitative understanding of the interactions of irrigation, climate variability, and crop yield is required for Arkansas farmers and policymakers to promote and implement sustainable irrigation practices.

Figure 1.

The Arkansas Delta Study Area. Graticule Datum: North American Datum 1983.

In this study, we filled these research gaps by investigating the effects of irrigation-induced climate variability on soybean yields in the Arkansas Delta by utilizing remote sensing, national inventory, and climate datasets. We address three research questions: (1) What are changes in the spatiotemporal dynamics of soybean areas in the Arkansas Delta from 2003 to 2017? (2) How does the irrigation of soybeans influence Delta climate during the growing seasons from June to August? (3) How does the changed climate affect soybean yields?

Materials and method

Study area

The Arkansas Delta located in eastern part of Arkansas (Figure 1), is extremely flat with little variations in local elevation. Most Arkansas Delta counties are wholly within the Mississippi Alluvial Plain ecoregion, but several counties also contain portions of the Mississippi Loess Valley (Clay, Greene, Craighead, Poinsett, Cross, St Francis, Lee, and Phillips) and South Central Plains (Jefferson, Lincoln, Drew, and Ashley) (Figure 1). Jackson and White counties (Mississippi Alluvial Plain and Arkansas Valley) and Lonoke County (Mississippi Alluvial Plain and Ouachita Mountains) also have mixed ecoregion compositions (Figure 1). The Arkansas Delta has annual precipitation of ∼1250 mm and temperature of ∼16° (Scott et al., 1998). There are abundant productive Sharkey clay soil series formed from Mississippi River slack water depositions in the region (Francis, 2022). Such flat topography, temperate climate, and fertile soil lead the Arkansas Delta as a home to various crops, such as soybean, rice, and cotton.

Data

To identify the soybean lands in the Arkansas Delta, we obtained the Cropland Data Layer (CDL) from the US Department of Agriculture-National Agricultural Statistics Service (USDA-NASS) (https://nassgeodata.gmu.edu/CropScape/). CDL with a spatial resolution of 30 m was generated using moderate resolution satellite imagery and extensive agricultural ground truth (Boryan et al., 2011). There are more than 140 crop types (soybean included), provided by CDL. The accuracy for major crop types is approximately 95% in most regions, according to CDL metadata, with some exceptions in the early years (USDA-NASS, 2021). For instance, CDL for Arkansas in 1999 has an overall accuracy of ∼70% (USDA-NASS, 2021). CDL from 1997 to 2020 covers the entire continental US, while data in some states in some years (e.g., Arkansas in 1997–1998) are missing.

To differentiate irrigated from rainfed lands, we used the newly developed, annual-scale, Landsat-based Irrigation Dataset (LANID) at a 30-m spatial resolution from 1997 to 2017 (https://zenodo.org/record/5548555#.YgGrMS-B2SU). LANID was created by Xie et al. (2021) using a random forest classifier based on an automatic training sample generation method and satellite-derived vegetation indices (e.g., normalized difference water index (NDWI)) and environmental variables (e.g., pressure of water vapor) (Xie and Lark, 2021). Validating against more than 10,000 ground-truthing samples, LANID has an overall accuracy of 90% across all the years and regions. We validated the accuracy of LANID in the Arkansas Delta with USDA-NASS harvested areas for irrigated and rainfed soybean.

The daily minimum and maximum temperature were derived from Moderate Resolution Imaging Spectroradiometer (MODIS) Land Surface Temperature (LST) product (i.e., MYD11A2 Version 6.1) (https://lpdaac.usgs.gov/products/myd11a2v061/) (Wan et al., 2015). We used LST as it can better capture canopy energy balance and improve crop yield prediction (Zhu and Burney, 2022). The Aqua (MYD) satellite rather than Terra (MOD) was chosen because its overpass times of 01:30 and 13:30 are closer to the times of normal daily morning minimum and afternoon maximum temperature (Botje et al., 2022). The MYD11A2 Version 6.1 product with a spatial resolution of 1 km and a temporal resolution of 8-days is available from July 2002 to current. The quality control information is provided along with LST data. We decoded the quality control information using NASA MODIS-VIIRS ArcGIS Python Toolbox (https://git.earthdata.nasa.gov/projects/LPDUR/repos/arcgis-modis-viirs-python-toolbox/browse). The resulting mandatory QA flags (i.e., 00 for good quality, 01 for other quality, 10 for unproduced due to cloud effects, and 11 for unproduced due to other reasons) were used to retain the pixels with good quality (i.e., QA flag = 00). The daily precipitation was obtained from Daily Surface Weather Data (Daymet) version 3 (https://daymet.ornl.gov/getdata). Daymet has a spatial resolution of 1 km and was derived from a collection of statistical modeling techniques to interpolate and extrapolate from daily meteorological observations (Thornton et al., 2016). The daily climatic variables covering North America are available from 1980 to present.

The annual yield data for irrigated and rainfed soybean are available on USDA-NASS Quick Stats Database (https://www.nass.usda.gov/Quick_Stats/). USDA-NASS provides survey yield data for irrigated and rainfed soybean at the US county level from 1972 to 2018 in the Arkansas Delta. The unit of soybean yield is bushels per acre (bu/acre, hereafter), and a bu/acre of soybean is equivalent to 67 kg ha⁻¹.

Considering the limited coverage and poor accuracy of CDL data in Arkansas before 2000 and the common years of the aforementioned datasets, the study period was set from 2003 to 2017. The crop yield and harvest area data from USDA-NASS for some counties may not cover the entire period; therefore, we used the available data for the following analyses.

Methods

We summarized data processing and statistical analyses in Figure 2 followed by the details of each step.

Figure 2.

The flowchart of the methods (green, golden, blue, and purple boxes denotes the input data, methods, intermediate results, and final results, respectively).

Extracting climatic variables over irrigated and rainfed soybean lands

To extract precipitation and temperature over irrigated and rainfed soybean lands in the Arkansas Delta, we firstly resampled CDL and LANID maps to 1 km from 2003 to 2017, a spatial scale that is consistent with that of MODIS and Daymet data, using a majority-aggregation method (He et al., 2018). We then used the Arkansas Delta boundary to resize CDL, LANID, MODIS, and Daymet data to obtain the data that only covers the Arkansas Delta region. After that, we extracted soybean areas using CDL and overlaid it with LANID data in the same year to identify Delta irrigated and rainfed soybean areas for each year from 2003 to 2017. We averaged the daily Daymet precipitation and 8-days MODIS temperature data in each month to obtain monthly average precipitation, and monthly average daily maximum and minimum temperature for the growing season from June to August during the investigated period. These climatic variables were subsequently overlaid with irrigated and rainfed soybean areas to acquire monthly average precipitation and temperature for irrigated and rainfed soybean lands, respectively.

As crop yield data from USDA-NASS are available only at the county level, precipitation and temperature for all irrigated and rainfed soybean pixels were averaged within a county to get county average values for irrigated and rainfed soybean, separately. In doing so, we obtained growing season monthly average precipitation and monthly average daily maximum and minimum temperature of irrigated and rainfed soybean for each county in the Arkansas Delta from 2003 to 2017.

Statistical analysis

Correlation analysis

To assess data accuracy, the irrigated and rainfed soybean areas from remote sensing products were aggregated to the county level to compare with the irrigated and rainfed soybean harvested areas from USDA-NASS from 2003 to 2017 using Pearson correlation coefficient (He et al., 2018). The significance of the correlation coefficient was tested by a Student’s t test. To confirm the results, we also computed the non-parametric Spearman’s rank and Kendall’s Tau correlation coefficients. This process was conducted using R 4.4.1 function cor.test.

Trend analysis

We conducted linear regression trend analysis ( $y = α x + ε$ ) for irrigated and rainfed soybean areas at the county level to determine the change in spatiotemporal patterns of soybean lands in the Arkansas Delta from 2003 to 2017 (Liu et al., 2021). The dependent variable ( $y$ ) in the linear regression is the aggregated county-level irrigated or rainfed soybean areas derived from remote sensing data and the independent variable ( $x$ ) is time (i.e., 2003–2017). We were interested in the slope ( $α$ ) as it reflects how soybean areas changed over time. We used Student’s t test (R 4.1.1 function lm) to assess the significance of the trend. We also applied non-parametric Mann–Kendall test coupled with Theil–Sen slope (R 4.1.1 package trend) to verify the trend of soybean areas in the Arkansas Delta.

Mean difference analysis

To reveal the impacts of irrigation on precipitation and temperature for each county in the Arkansas Delta, we applied mean difference analysis (Yang et al., 2021). We subtracted climatology mean monthly average precipitation, and monthly average daily maximum or minimum temperature values of rainfed soybean, averaged from 2003 to 2017 from that of irrigated soybean. The positive, negative, or zero values of the differences indicate that irrigation has positive, negative, or no effects on local precipitation or temperature for each county in the Arkansas Delta. A two-sample t test (R 4.1.1 function t.test) was used to test the significance of the difference for each individual county.

Crop yield model

We used a statistical quadratic model to predict soybean yields, following (Li et al., 2020) (Equation (1)). The soybean yield of all counties in the Arkansas Delta is used as the dependent variable, and the county-level monthly average precipitation, monthly average daily maximum and minimum temperature are used as independent variables

\hat{y i e l d} = a * y e a r + \sum_{m = J u n e}^{A u g u s t} (b_{m} * {T m i n}_{m} + c_{m} * {T m i n}_{m}^{2}) + \sum_{m = J u n e}^{A u g u s t} (d_{m} * {T m a x}_{m} + e_{m} * {T m a x}_{m}^{2}) + \sum_{m = J u n e}^{A u g u s t} (f_{m} * P_{m} + g_{m} * P_{m}^{2}) + c_{0}

(1)

where

a

b

c

, d,

e

f

, and

g

are the estimated coefficients.

T m i n

and

T m a x

denote monthly average daily minimum and maximum temperature, respectively.

P

denotes monthly average precipitation. The “year” predictor is the long-term increasing yield trends due to improvements in management and technology, and

c_{0}

is the residual that was not predicted by the model.

We built quadratic model for rainfed soybean (Equation (2)) (Li et al., 2020) to identify the irrigation-climate effects on crop yields

R a i n f e d m o d e l : {\hat{y i e l d}}_{r a i n f e d} = f_{r a i n f e d} ({T m i n}_{r a i n f e d}, {T m a x}_{r a i n f e d}, P_{r a i n f e d})

(2)

The irrigation-climate interactions on yields ${\hat{∆ Y i e l d}}_{i r r i g a t e d}$ is defined as the hypothetical yield change in rainfed lands if the same climate variability as irrigated lands was applied (Li et al., 2020). Thus, the effects of changed climate variability on crop yields ( ${\hat{∆ Y i e l d}}_{i r r i g a t e d}$ ) can be calculated as the predicted yields of irrigated precipitation and temperature with the rainfed model minus the predicted yields of rainfed precipitation and temperature with the rainfed model (Equation (3))

{\hat{∆ Y i e l d}}_{i r r i g a t e d} = f_{r a i n f e d} ({T m i n}_{i r r i g a t e d}, {T m a x}_{i r r i g a t e d}, P_{i r r i g a t e d}) - f_{r a i n f e d} ({T m i n}_{r a i n f e d}, {T m a x}_{r a i n f e d}, P_{r a i n f e d})

(3)

To further estimate the soybean yield change due to other effects (i.e., increasing water supply from irrigation), we also built quadratic model for irrigated soybean (Equation (4))

I r r i g a t e d m o d e l : {\hat{y i e l d}}_{i r r i g a t e d} = f_{i r r i g a t e d} ({T m i n}_{i r r i g a t e d}, {T m a x}_{i r r i g a t e d}, P_{i r r i g a t e d})

(4)

The water supply effect is embedded in the irrigated model $f_{i r r i g a t e d}$ , which was calculated by subtracting the predicted yields of the irrigated precipitation and temperature with rainfed model from the predicted yields of irrigated precipitation and temperature with the irrigated model (Equation (5))

{\hat{∆ Y i e l d}}_{w a t e r} = f_{i r r i g a t e d} ({T m i n}_{i r r i g a t e d}, {T m a x}_{i r r i g a t e d}, P_{i r r i g a t e d}) - f_{r a i n f e d} ({T m i n}_{i r r i g a t e d}, {T m a x}_{i r r i g a t e d}, P_{i r r i g a t e d})

(5)

The ${\hat{∆ Y i e l d}}_{i r r i g a t e d}$ and ${\hat{∆ Y i e l d}}_{w a t e r}$ were calculated for each county in the Arkansas Delta region.

Results and discussion

Spatiotemporal change patterns of soybean areas in the Arkansas Delta

The correlation coefficient value of irrigated soybean areas derived from CDL and LANID with that from USGS-NASS survey irrigated areas is 0.879, which is higher than that of rainfed soybean areas (r = 0.624) (Figure 3). The relatively poor agreement of rainfed areas may be because of the inaccurate classification of rainfed croplands during some high precipitation years (Xie and Lark, 2021). Nevertheless, both values are statistically significant at 1% level, indicating the reliability of remote sensing-derived irrigated and rainfed soybean data. This is largely consistent with the significant (p-value <0.1) non-parametric Spearman’s rank and Kendall’s Tau (shown in parentheses) correlation coefficients of 0.837 (0.656) and 0.666 (0.470) for irrigated and rainfed soybean lands, respectively.

Figure 3.

The correlation relationships of (a) irrigated and (b) rainfed soybean areas in each county in the Arkansas Delta derived from remotely sensed data and USDA-NASS survey data.

Figures 4 (a) and (b) present the averaged irrigated and rainfed soybean areas from 2003 to 2017 in the Arkansas Delta. The areas of irrigated soybean lands have much larger areas compared to that of rainfed soybean (Figures 4 (a) and (b)). Irrigated soybean lands are dominant in the eastern part of the Delta, such as Mississippi, Poinsett, Cross, Crittenden, Phillips, Arkansas, and Desha counties, with the areas ranging from ∼90,000 acres to ∼150,000 acres. The rainfed soybean areas in most of the counties range from ∼30,000 acres to ∼60,000 acres, with some counties having low areas, such as the Drew County.

Figure 4.

Mean (a) irrigated and (b) rainfed soybean areas for each county in the Arkansas Delta averaged over the period from 2003 to 2017, and temporal trends in (c) irrigated and (d) rainfed soybean areas estimated from 2003 to 2017. The un-stippled areas in (c) and (d) indicate significant trends at the 10% level.

Linear regression trend analysis reveals that irrigated soybean areas significantly increased in the eastern part of the Arkansas Delta from 2003 to 2017, with an increase rate ranging from 1840 acre/year to 9200 acre/year (Figure 4(c)). Among these counties, Mississippi and Desha counties have the largest increase. The irrigated areas in Lawrence County in the northern Arkansas Delta have significantly decreased. The rainfed soybean areas show insignificant changes in most of the counties in the Delta, while Randolph and White counties have significantly decreased areas (Figure 4(d)). The non-parametric trend analysis confirms the change pattern of soybean lands in the Arkansas Delta (Figure S1).

Impacts of soybean irrigation on precipitation and temperature

Figures 5(a)–(c) show the climatology mean difference of monthly average precipitation over irrigated soybean lands with that over rainfed lands, averaged from 2003 to 2017 for each month during the growing season of soybean. Irrigation has negative effects on precipitation over most counties, which is consistent with the fact that irrigated lands are generally drier than rainfed areas. Yet, these negative effects are statistically insignificant, indicating soybean irrigation has negligible impacts on precipitation variability in the Arkansas Delta. These limited impacts of irrigation on precipitation have been identified by several previous studies. For example, Puma and Cook (2010) used the Goddard Institute for Space Studies (GISS) ModelE to estimate irrigation effects on climate during the 20^th century and showed insignificant changes in precipitation during summer over several regions across the globe, including the State of Arkansas. By using an Advanced Research Weather Research and Forecasting (WRF) model, Yang et al. (2016) demonstrated that irrigation over the Huang-Huai-Hai Plain in China also had negligible effects on precipitation during summertime in this region, although its impacts over northern and southern parts were statistically significant. In a recent study, Qian et al. (2020) found negative but insignificant precipitation changes due to irrigation during summer in most parts of eastern Arkansas, which was attributable to the compound effects of increased Convective Available Potential Energy (CAPE) because of increased irrigation moisture and enhanced convective inhibition (CIN) due to irrigation-induced evaporative cooling.

Figure 5.

Irrigation impacts on monthly average (a–c) precipitation, (d–f) daily maximum temperature, and (g–i) daily minimum temperature in each county of the Arkansas Delta during the growing season of June to August. The unhatched regions indicate significant differences at the 10% level.

For most counties in the Arkansas Delta, irrigation shows significantly negative impacts on monthly average daytime (i.e., daily maximum) temperature (Figure 5(d)–(f)). These cooling effects were also revealed by Barnston and Schickedanz (1984), who reported irrigation appeared to lower the daily surface maximum temperature by ∼ 1°C–2°C during the warm season in the Southern Great Plains. Similarly, a recent study of Yang et al. (2020) found irrigation averagely decreased daytime land surface temperature by 1.15 K across irrigated lands in China. This temperature decrease may be attributable to a shift away from sensible and toward latent heat fluxes in the irrigated areas (Mahmood et al., 2004; Kueppers et al., 2007; He et al., 2020). Few counties also witness insignificant effects of irrigation on monthly average daytime temperature in the Arkansas Delta. These negligible impacts were consistent with the findings of Chen and Dirmeyer (2019), who used MODIS temperature data and stated that cooling effects of irrigation during summer were less evident in the areas of the Lower Mississippi River Basin, US.

As the impacts of irrigation on daytime temperature, irrigation also has significantly negative effects on monthly average nighttime (i.e., daily minimum) temperature in most of counties in the Arkansas Delta, although their magnitudes are much smaller than that of daytime temperature (Figure 5(g)–(i)). Such phenomena are also demonstrated in Yang et al. (2020) that in the night time, cooling was still the dominant effect in the majority of the irrigated areas, relative to non-irrigated areas, but with a much weaker magnitude. Yang et al. (2020) further claimed that the enhanced heat loss through higher evapotranspiration (ET) in the irrigated regions exceeded the extra energy gain induced by the lower albedo, leading to cooling effects of irrigation. Meanwhile, there are some counties showing significant nighttime warming effects of irrigation, especially during June (Figure 5(g) and (h)). Interestingly, the counties with significant warming effects mainly have relatively small areas of irrigated soybean (Figure 4(a) and Figure 5(g)). This largely supports Coffel et al. (2022)’s hypothesis that crops-induced cooling may fade in years with poor yields or a smaller harvested area. Nevertheless, these warming effects were also observed in the California Central Valley by a study using the Scripps Experimental Climate Prediction Center (ECPC) Regional Spectral Model (RSM) (Kanamaru and Kanamitsu, 2008). A similar finding was identified by an observational study that presented a dramatic warming in the daily minimum temperature in the irrigated California Central Valley compared to that in the non-irrigated adjacent highlands of Sierra (Christy et al., 2006). These warming effects are possible due to (1) the enhanced greenhouse warming because of increased water vapor concentrations in the atmosphere from soil evaporation and crop evapotranspiration into the boundary layer; (2) enhanced night time downward infrared flux due to swelling of aerosols as humidity increases toward morning; and (3) enhanced night time sensible heat flux from the surface due to the increased heat capacity of moist soil (Christy et al., 2006; Chen and Jeong, 2018).

Impacts of irrigation-climate interactions on soybean yield

We used equation (1) to build the crop yield models for rainfed and irrigated soybean, respectively. Due to the limited effects of irrigation impacts on monthly average precipitation as identified in the section 3.2, the average precipitation from irrigated and rainfed areas was used for both rainfed and irrigated crop yield models. The F statistics for the two models are 12.3 and 12.68, respectively. Both models have significant quadratic relationships with corresponding p values less than 0.01. The residuals of both models distribute evenly around the zero line, meaning that the average of the residuals is approximately zero (Figure 6(a) and (d)). The histograms of the residuals for both models suggest that the residuals are nearly normal with a zero mean (Figure 6(b) and (e)). The quantiles of the residuals show good fit with the quantiles of standard normal in the Q-Q plot (Figure 6(c) and (f)). All these statistical inspections support the normality of the residuals for both models.

Figure 6.

Normality test of the residuals for (a–c) rainfed and (d–f) irrigated crop yield models.

The adjusted R-squares of rainfed and irrigated models are 0.427 and 0.434, respectively (Table 1), indicating our selected independent variables could explain 42.7% and 43.4% of rainfed and irrigated soybean yields variability. To further assess the model performance, we compared the difference of yields between irrigated and rainfed soybean from our models with that derived from USDA-NASS data. The r value is 0.336 and significant at 1% level, indicating the reliability of our irrigated and rainfed models.

Table 1.

The yield model for rainfed and irrigated soybeans. Statistically significant p-values (p < 0.1) are bolded.

Rainfed
Variable	Coefficient	p -value	Adjust R ²
Year	0.46920	0.000	0.427
$T m a x$ in Jun	−0.08542	0.871
$(T m a x$ in Jun)²	0.00000	0.864
$T m a x$ in Jul	−1.53900	0.068
( $T m a x$ in Jul)²	−0.00005	0.069
$T m a x$ in Aug	−1.53400	0.028
( $T m a x$ in Aug)²	0.00005	0.029
$T m i n$ in Jun	0.61380	0.774
$(T m i n$ in Jun)²	−0.00002	0.769
$T m i n$ in Jul	1.13500	0.749
( $T m i n$ in Jul)²	−0.00004	0.751
$T m i n$ in Aug	−1.61700	0.456
( $T m i n$ in Aug)²	0.00006	0.458
$P$ In Jun	3.17900	0.000
$(P$ In Jun)²	0.19870	0.023
$P$ In Jul	1.52300	0.047
( $P$ in Jul)²	−0.13000	0.112
$P$ In Aug	2.02200	0.001
( $P$ in Aug)²	0.17620	0.004
Intercept	−22,370.00000	0.449
Irrigated
Variable	Coefficient	p-value	Adjust R ²
Year	0.82300	0.000	0.434
$T m a x$ In Jun	−0.29270	0.524
$(T m a x$ In Jun)²	0.00000	0.523
$T m a x$ In Jul	−1.36700	0.084
( $T m a x$ in Jul)²	0.00004	0.088
$T m a x$ In Aug	0.89450	0.149
( $T m a x$ in Aug)²	0.00003	0.147
$T m i n$ In Jun	−1.88600	0.331
$(T m i n$ In Jun)²	0.00006	0.335
$T m i n$ In Jul	0.21050	0.946
( $T m i n$ in Jul)²	−0.00000	0.949
$T m i n$ In Aug	−0.34100	0.856
( $T m i n$ in Aug)²	0.00001	0.858
$P$ In Jun	1.68000	0.020
$(P$ In Jun)²	0.12840	0.082
$P$ In Jul	0.73720	0.243
( $P$ in Jul)²	−0.18500	0.008
$P$ In Aug	1.41600	0.004
( $P$ in Aug)²	−0.15310	0.003
Intercept	32,860.00000	0.205

Equation (3) was used to assess how the irrigation-climate interactions influenced soybean yields for each year from 2003 to 2017. Figure 7(a) shows that irrigation-climate interactions increase soybean yields for all the years over the Arkansas Delta. The largest increase was in 2009 with a value of 1.27 ( $\pm$ 0.78) bu/acre. The spatial pattern of the effect shows that all counties in the Arkansas Delta experience positive effects of irrigation-climate interactions on soybean yields. The largest positive impacts are in the counties of Cross, Creene, Prairie, Randolph, and White (Figure 7(b)), with values of 1.13 ( $\pm$ 0.87) bu/acre, 1.02 ( $\pm$ 0.67) bu/acre, 1.17 ( $\pm$ 0.91) bu/acre, 1.04 ( $\pm$ 0.60) bu/acre, and 1.02 ( $\pm$ 1.09) bu/acre, respectively (Figure 7(b), accounting for 7.78%, 7.16%, 7.47%, 7.34%, and 6.80% of total crop yields gains due to irrigation (Figure 7(c)). Averagely, irrigation-climate interactions are responsible for 4.85% ( $\pm$ 1.90%) of irrigation yield benefit. The positive effects of irrigation-climate interactions on crop yields identified in this study support the recent findings that irrigation cooling could boost maize yields in Nebraska, U.S. (Li et al., 2020; Zhu and Burney, 2022) and soy yields in US Midwest (Coffel et al., 2022), although our magnitudes are smaller. This may be due to the fact that Nebraska is drier than Arkansas and crops there needs more irrigation water, as such the cooling effects are stronger, implying that irrigation effects also depend on background climate.

Figure 7.

Effects of (a–b) irrigation-climate interactions and (d–e) irrigation-water supply on soybean yields. (c) and (f) is the percentage of contribution of irrigation-climate interactions and irrigation-water supply to the total irrigation yield gain. The shaded areas in (a) and (c) are two standard deviations.

Based on equation (5), we also assessed how irrigation-water supply affected soybean yields from 2003 to 2017 in the Arkansas Delta region. The irrigation water increases soybean yields across all the years with a maximum value of 18.83 ( $\pm$ 2.09) bu/acre in 2012 (Figure 7(d)), which is much higher than the effects of irrigation-climate interactions. Spatially, the irrigation water also increases soybean yields for all counties in the Arkansas Delta, with a maximum increase of 16.36 ( $\pm$ 3.11) bu/acre in Phillips County (Figure 7(e)), accounting for 97.39% of crop yields gain due to irrigation. The average contribution of irrigation-water supply to irrigation crop yield benefit is 95.15% ( $\pm$ 1.90). The positive effects of irrigation-water supply on crop yields are expected, as irrigation water has a long history of being considered as an important factor for increasing crop yields (Okada et al., 2018; Kukal and Irmak, 2019).

There are some limitations in our analyses. First, the Daymet precipitation data are from interpolation or extrapolation of meteorological station data, which may mask out the small-scale irrigation-precipitation interactions. This maybe improved when high-resolution satellite-derived or reanalysis data are available. Second, there are some uncertainties in the input data (e.g., inconsistent reporting in USDA-NASS county-level yield data, inconsistency of CDL data between years 2003–2007 and 2008-2017, and relative low accuracy of rainfed data from LANID data). This may lead to uncertainties in the crop yield models, and thus over- or under-estimate the effects of irrigation-climate interactions and irrigation-water supply on soybean yields in the Arkansas Delta. Using training samples from Delta to improve the satellite-based crop types data and irrigated and rainfed data may enhance the data accuracy, thus reducing the uncertainties in yield model.

Conclusion

The Arkansas Delta, ranking third in the nation in terms of irrigated croplands, relies heavily on groundwater irrigation. The irrigated areas continue to increase in Arkansas with two million ha of irrigated croplands recently (Yaeger et al., 2018). Better understanding the irrigation-climate interactions impacts on crop yields is vital to promote sustainable irrigation practices and thereby maintain long-term agricultural profitability in Arkansas. Irrespective of the rapid increase of irrigated lands in the Arkansas Delta, the effects of irrigation and the consequent climate variability on crop yield in this region have not been a research priority relative to other parts of the United States like the Midwest. Such an unsolved issue becomes more important under the future climate change, where both water and heat stresses are expected to become severer (Coffel et al., 2019).

The study applied multiple remote sensing data products, a national inventory, and climate datasets to explore the effects of irrigation-climate interactions on soybean yields in the US Arkansas Delta in the early 21^st Century. The results show that the areas for irrigated soybean statistically significantly increase in the eastern Arkansas Delta, while the areas for rainfed soybean show insignificant change from 2003 to 2017. Soybean irrigation has limited impacts on monthly average precipitation, while the cooling effects of irrigation on monthly average daily maximum and minimum temperature are much stronger. By using statistical rainfed and irrigated crop yield models, we further demonstrated that in addition to the direct irrigation-water supply benefits, irrigation-climate interactions could boost soybean yields in the Arkansas Delta region, contributing ∼4.85% of total irrigation yield gain.

Our work serves as another important evidence that irrigation or increased water supplies to crop fields has both direct (i.e., meeting biological water demand) and indirect (i.e., cooling effect to reduce heat stress) contributions to irrigated crop yields gain. Together with previous studies (e.g., Zhu and Burney, 2022), this finding may be applied to other agricultural regions across the globe. Therefore, it could serve as the scientific basis to aid sustainable agricultural practices, thus benefiting local farmers in Arkansas and across the globe. However, continuous monitoring for the future of agriculture is needed. The global water supply crisis has occurred across the globe, including in the US, due to rapid population growth, fast urbanization, and drastic land-cover changes (He et al., 2020). The insignificant impacts of irrigation on local precipitation patterns we identified in this study could not guarantee surface reservoir or aquifer recharge in future, given the continuous warming and more frequent drought (Cook et al., 2020). In addition, groundwater in some regions (e.g., Arkansas Delta) has experienced critical declines (Reba et al., 2017). The limited groundwater may reduce irrigation intensity, as such cooling effects may fade in the future, exposing crops to more harmful heat stress that caused by climate change (Coffel et al., 2022).

Supplemental Material

Supplemental Material - Impacts of irrigation-climate interactions on irrigated soybean yields in the US Arkansas Delta from 2003 to 2017

Supplemental Material for Impacts of irrigation-climate interactions on irrigated soybean yields in the US Arkansas Delta from 2003 to 2017 by Yaqian He, Matthew H. Connolly, Rongting Xu, Xiao Huang, Zhuosen Wang, Marisol Filares Arreguin and Caden Rhodes in Progress in Physical Geography: Earth and Environment

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Faculty Development Competitive Grant at the University of Central Arkansas; NASA EPSCoR Research Infrastructure Development (RID21014); University Research Council Grant at University of Central Arkansas.

ORCID iD

Yaqian He

Supplemental Material

Supplemental material for this article is available online.

References

Adegoke

Pielke

Eastman

, et al. (2003) Impact of Irrigation on Midsummer Surface Fluxes and Temperature under Dry Synoptic Conditions: A Regional Atmospheric Model Study of the U.S. High Plains. Monthly Weather Review 131(3): 556–564.

Alter

E-S

Eltahir

EAB

(2015) Rainfall consistently enhanced around the Gezira Scheme in East Africa due to irrigation. Nature Geoscience 8: 763–767.

Barnston

Schickedanz

(1984) The effect of irrigation on warm season precipitation in the Southern Great Plains. Journal of Climate and Applied Meteorology 23(6): 865–888.

Boryan

Yang

Mueller

, et al. (2011) Monitoring US agriculture: The US department of agriculture, national agricultural statistics service, cropland data layer program. Geocarto International 26(5): 341–358. DOI: 10.1080/10106049.2011.562309

Botje

Dewan

Chakraborty

(2022) Comparing coarse-resolution land surface temperature products over western Australia. Remote Sensing 14(10): 2296.

Boucher

Myhre

(2004) Direct human influence of irrigation on atmospheric water vapour and climate. Climate Dynamics 22(6–7): 597–603. DOI: 10.1007/s00382-004-0402-4

Chen

Dirmeyer

(2019) Global observed and modelled impacts of irrigation on surface temperature. International Journal of Climatology 39(5): 2587–2600. DOI: 10.1002/joc.5973

Chen

Jeong

(2018) Irrigation enhances local warming with greater nocturnal warming effects than daytime cooling effects. Environmental Research Letters 13(2): 024005. DOI: 10.1088/1748-9326/aa9dea

Christy

Norris

Redmond

, et al. (2006) Methodology and Results of Calculating Central California Surface Temperature Trends: Evidence of Human-Induced Climate Change? Journal of Climate 19(4): 548–563.

10.

Coffel

Keith

Lesk

, et al. (2019) Future Hot and Dry Years Worsen Nile Basin Water Scarcity Despite Projected Precipitation Increases. Earth’s Future 7(8): 967–977. DOI: 10.1029/2019EF001247

11.

Coffel

Lesk

Winter

, et al. (2022) Crop-climate feedbacks boost US maize and soy yields. Environmental Research Letters 17(2): 024012. DOI: 10.1088/1748-9326/ac4aa0

12.

Cook

Mankin

Marvel

, et al. (2020) Twenty‐first century drought projections in the CMIP6 forcing scenarios. Earth's Future 8(6): e2019EF001461.

13.

Evett

Carman

Bucks

(2003) Expansion of Irrigation in the Mid-South United States: water allocation and research issues. In: 2nd International Conference on Irrigation and Drainage, Water for a Sustainable World – Limited Supplies and Expanding Demand, Phoenix, AZ, 12-15 May, pp. 247–260.

14.

Francis

(2022) Soils - Encyclopedia of Arkansas. Available at: https://encyclopediaofarkansas.net/entries/soils-5141/#:∼:text=IntheDelta%2CtheSharky,isaproductiveagriculturalsoil (accessed 4 May 2022).

15.

Harden

Liu

(2020) Comparison of water resources management between China and the United States. Geography and Sustainability 1(2): 98–108.

16.

Warner

McNeil

, et al. (2018) Reducing uncertainties in applying remotely sensed land use and land cover maps in land-atmosphere interaction: Identifying change in space and time. Remote Sensing 10(4): 506. DOI: 10.3390/rs10040506

17.

Lee

(2016) Empirical Relationships of Sea Surface Temperature and Vegetation Activity with Summer Rainfall Variability over the Sahel. Earth Interactions 20(6): 1–18. DOI: 10.1175/EI-D-15-0028.s1

18.

Lee

Mankin

(2020) Seasonal tropospheric cooling in Northeast China associated with cropland expansion. Environmental Research Letters 15(3): 034032. DOI: 10.1088/1748-9326/ab6616

19.

IPCC (2018) Global Warming of 1.5°C. An IPCC Special Report on the Impacts of Global Warming of 1.5°C above Pre-industrial Levels and Related Global Greenhouse Gas Emission Pathways the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty. Available at: https://www.ipcc.ch/sr15/ (accessed 10 July 2021).

20.

Kanamaru

Kanamitsu

(2008) Model diagnosis of nighttime minimum temperature warming during summer due to irrigation in the California Central Valley. Journal of Hydrometeorology 9(5): 1061–1072. DOI: 10.1175/2008JHM967.1

21.

Kueppers

Snyder

Sloan

(2007) Irrigation cooling effect: Regional climate forcing by land-use change. Geophysical Research Letters 34(3): L03703. DOI: 10.1029/2006GL028679

22.

Kukal

Irmak

(2019) Irrigation-limited yield gaps: Trends and variability in the United States post-1950. Environmental Research Communications 1(6): 061005. DOI: 10.1088/2515-7620/ab2aee

23.

Kukal

Irmak

(2020) Impact of irrigation on interannual variability in United States agricultural productivity. Agricultural Water Management 234: 106141.

24.

Lesk

Coffel

Horton

(2020) Net benefits to US soy and maize yields from intensifying hourly rainfall. Nature Climate Change 10(9): 819–822. DOI: 10.1038/s41558-020-0830-0

25.

Guan

Schnitkey

, et al. (2019) Excessive rainfall leads to maize yield loss of a comparable magnitude to extreme drought in the United States. Global Change Biology 25(7): 2325–2337. DOI: 10.1111/gcb.14628

26.

Guan

Peng

, et al. (2020) Quantifying irrigation cooling benefits to maize yield in the US Midwest. Global Change Biology 26(5): 3065–3078. DOI: 10.1111/gcb.15002

27.

Liu

Shen

(2021) Effects of cropland expansion on temperature extremes in Western India from 1982 to 2015. Land 10(5): 489. DOI: 10.3390/land10050489

28.

Mahmood

Foster

Keeling

, et al. (2006) Impacts of irrigation on 20th century temperature in the northern Great Plains. Global and Planetary Change 54(1–2): 1–18. DOI: 10.1016/j.gloplacha.2005.10.004

29.

Mahmood

Hubbard

Carlson

(2004) Modification of growing-season surface temperature records in the northern great plains due to land-use transformation: verification of modelling results and implication for global climate change. International Journal of Climatology 24(3): 311–327. DOI: 10.1002/joc.992

30.

Okada

Iizumi

Sakamoto

, et al. (2018) Varying Benefits of Irrigation Expansion for Crop Production Under a Changing Climate and Competitive Water Use Among Crops. Earth’s Future 6(9): 1207–1220. DOI: 10.1029/2017EF000763

31.

Puma

Cook

(2010) Effects of irrigation on global climate during the 20th century. Journal of Geophysical Research 115(16): D16120. DOI: 10.1029/2010JD014122

32.

Qian

Yang

Feng

, et al. (2020) Neglecting irrigation contributes to the simulated summertime warm-and-dry bias in the central United States. Npj Climate and Atmospheric Science 3(1): 31. DOI: 10.1038/s41612-020-00135-w

33.

Ray

Gerber

MacDonald

, et al. (2015) Climate variation explains a third of global crop yield variability. Nature Communications 6: 5989. DOI: 10.1038/ncomms6989

34.

Reba

Massey

Adviento‐Borbe

, et al. (2017) Aquifer depletion in the lower Mississippi River Basin: Challenges and solutions. Journal of Contemporary Water Research & Education 162(1): 128–139.

35.

Scott

, et al. (1998) Agricultural Water Management in the Mississippi Delta Region of Arkansas. Fayetteville: Report, University of Arkansas.

36.

Szilagyi

(2018) Anthropogenic hydrological cycle disturbance at a regional scale: State-wide evapotranspiration trends (1979–2015) across Nebraska, USA. Journal of Hydrology 557: 600–612. DOI: 10.1016/j.jhydrol.2017.12.062

37.

Thornton

, et al. (2016) Daymet: Daily Surface Weather Data on a 1-km Grid for North America. Oak Ridge, TN, USA: ORNL DAAC. Version 3. DOI: 10.3334/ORNLDAAC/1328

38.

Troy

Kipgen

Pal

(2015) The impact of climate extremes and irrigation on US crop yields. Environmental Research Letters 10(5): 054013. DOI: 10.1088/1748-9326/10/5/054013

39.

USDA (2017) United States Department of Agriculture National Agricultural Statistics Service. Available at: https://www.nass.usda.gov/Quick_Stats/CDQT/chapter/2/table/10/state/AR/year/2017 (accessed 10 August 2021).

40.

USDA-NASS (2021) CropScape and Cropland Data Layer-Metadata. Available at: https://www.nass.usda.gov/Research_and_Science/Cropland/metadata/meta.php (accessed 28 February 2022).

41.

Walkenhorst

(2018) Report: Arkansas farming a drain on water; again, irrigation cited in draw-off. Available at: https://www.arkansasonline.com/news/2018/jun/24/report-farming-a-drain-on-water-2018062-1/(accessed 28 February 2022).

42.

Wan

Hook

Hulley

(2015) MYD11A2 MODIS/Aqua Land Surface Temperature/Emissivity 8-Day L3 Global 1km SIN Grid V006. Available at: NASA EOSDIS Land Processes DAAC DOI: 10.5067/MODIS/MYD11A2.006 (accessed 2022 09 14).

43.

West

, et al. (2020) Arkansas Irrigation. Report. Fayetteville: University of Arkansas.

(2021) Landsat-based Irrigation Dataset (LANID): 30-m resolution maps of irrigation distribution, frequency, and change for the US, 1997-2017. Earth System Science Data 13: 5689–5710. DOI: 10.5281/zenodo.5003976

45.

Xie

Lark

(2021) Mapping annual irrigation from Landsat imagery and environmental variables across the conterminous United States. Remote Sensing of Environment 260: 112445. DOI: 10.1016/j.rse.2021.112445

, et al. (2018) Trends in the construction of on-farm irrigation reservoirs in response to aquifer decline in eastern Arkansas: Implications for conjunctive water resource management. Agricultural Water Management 208: 373–383. DOI: 10.1016/j.agwat.2018.06.040

47.

Yang

, et al. (2021) Human Mobility to Parks Under the COVID-19 Pandemic and Wildfire Seasons in the Western and Central United States. GeoHealth 5(12): e2021GH000494. DOI: 10.1029/2021GH000494

, et al. (2016) Climatic effects of irrigation over the Huang-Huai-Hai plain in China simulated by the weather research and forecasting model. Journal of Geophysical Research: Atmospheres 121(5): 2246–2264. DOI: 10.1002/2015JD023736

(2020) Irrigation cooling effect on land surface temperature across China based on satellite observations. The Science of the Total Environment 705: 135984.

(2020) Global assessment of the impact of irrigation on land surface temperature. Science Bulletin 65: 1440–1443.

, et al. (2017) Temperature increase reduces global yields of major crops in four independent estimates. Proceedings of the National Academy of Sciences of the United States of America 114(35): 9326–9331. DOI: 10.1073/pnas.1701762114

52.

Zhu

Burney

(2022) Untangling irrigation effects on maize water and heat stress alleviation using satellite data. Hydrology and Earth System Sciences 26(3): 827–840.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.32 MB