Integrated Foliar Spraying Effectively Reduces Wheat Yield Losses Caused by Hot–Dry–Windy Events: Insights from High-Yield and Stable-Yield Winter Wheat Regions in China

Abstract

Integrated foliar spraying has been proposed as an effective measure to mitigate the increasingly severe impacts of hot–dry–windy (HDW) events on winter wheat yield under ongoing climate change, and its physiological effectiveness has been mechanistically validated. However, there are still few quantitative assessments of the application of this technology at the regional scale. First, hourly meteorological data from the ERA5-Land reanalysis (1981–2020) were matched to the centroids of 599 counties within China’s major winter wheat-producing regions, allowing precise alignment with county-level yield data. Subsequently, spatial and temporal trends of sub-daily HDW events were analyzed. These HDW events were classified according to daily duration into three categories: short-duration (HDW_sd1, 1 h d⁻¹), moderate-duration (HDW_sd2, 2–3 h d⁻¹), and prolonged-duration (HDW_sd3, 4–8 h d⁻¹). Finally, a difference-in-differences (DiD) approach combined with panel matching methods was employed to quantitatively assess the effectiveness of integrated foliar spraying technology—comprising plant growth regulators, essential nutrients, fungicides, and insecticides—on wheat yield improvements under varying irrigation conditions. The results indicate that HDW is a major compound event threatening high-yield and stable-yield regions within the main winter wheat production areas of China, and in the study area, the annual average number of HDW days ranges from 3 to 13 days, increasing by 1–4 days dec⁻¹. While HDW events continue to intensify, the integrated foliar spraying technology effectively mitigates yield losses due to HDW stress. Specifically, yield increases of up to 18–20% were observed in counties with sufficient irrigation infrastructure since the large-scale implementation began in 2012, particularly in regions exposed to more than 2 days of HDW stresses annually. However, the effectiveness of integrated foliar spraying was notably compromised in areas lacking adequate irrigation infrastructure, highlighting the necessity of reliable irrigation conditions. In these poorly irrigated areas, yield improvements remained limited and inconsistent, typically fluctuating around negligible levels. These findings underscore that robust irrigation infrastructure is pivotal to unlock the yield benefits of integrated foliar spraying technology, while also highlighting its transformative potential in advancing climate-smart agriculture globally—particularly in regions grappling with intensifying compound stress events driven by climate change, where this innovation could foster resilient and adaptive food systems to counter escalating environmental extremes.

1. Introduction

Sub-daily hot–dry–windy (HDW) events [1] consist of short bursts of high temperature, low humidity, and moderate wind. Even a few hours of HDW events desiccate the flag leaf, collapse stomatal conductance, and curtail starch deposition, cutting final yield more than any full-season heatwave or drought alone, because they strike during grain-set and accelerate canopy desiccation and thermal injury [1,2,3,4,5,6]. Compared with single meteorological hazards, compound extreme events—such as simultaneous occurrences of extreme heat, drought, and strong winds—often impose substantially greater negative impacts on crops [7,8]. For example, compound dry–hot conditions, characterized by high temperatures and low humidity, significantly disrupt grain filling in wheat by impairing starch accumulation, ultimately reducing grain yield [9,10]. When heat and drought coincide without wind, plants partially mitigate stress by closing stomata and raising leaf temperature. Wind further amplifies heat and drought stress by raising transpiration demand and stripping leaf water.

Historical evidence underscores the severity of such multivariate climate hazards globally. For instance, the combined occurrence of persistent high temperature, drought, and high winds across the Great Plains region in 2014 led to record-low soil moisture levels, severely compromising winter wheat yields in states such as Kansas, Oklahoma, and Texas, the lowest recorded yields since 1995 [10]. Similarly, research indicates that the likelihood of extreme dry–hot years, characterized by record-breaking high temperatures and drought conditions exceeding critical enzyme thresholds in wheat plants, has notably increased from 1% to 6% in China’s primary winter wheat production regions between 1981 and 2020 [11,12]. During such heat stress events, plants fail to adequately accumulate starch in grains [9], further decreasing wheat yields and exacerbating economic losses. Furthermore, in these regions—including the Huang-Huai-Hai plain and provinces such as Shanxi, Shaanxi, and Hubei, which collectively contribute about 88% of the nation’s wheat output—HDW events pose a serious threat to wheat production and cause annual yield losses from 5% to 30% [13].

To address the critical impacts of HDW on agricultural productivity, standardized criteria have been established, such as China’s Meteorological Industry Standard (QX/T82-2019 [14], revised from QX/T82-2007 [15]). This standard classifies HDW severity into mild, moderate, and severe categories, based primarily on daily maximum temperatures, relative humidity, wind speeds, and soil moisture conditions at 14:00. However, daily classifications alone may inadequately capture the rapid dynamics of HDW events. Assessing HDW at finer temporal scales, particularly hourly durations within a day, is essential for accurately characterizing the full impact of these climatic stressors on crop growth. Recent studies within China, based on these standards, have identified declining trends in daily-scale HDW severity due to slight decreases in peak temperatures and reductions in wind speeds from 1961 to 2015, indicating potentially beneficial conditions for wheat grain filling and yield stability [16,17,18,19]. Nevertheless, globally, definitions of HDW vary widely, creating difficulties in comparative analyses. For instance, studies in the United States have adopted varying temperature thresholds (29–37 °C and above 38 °C), low humidity conditions (<30%), and higher wind speeds (≥7 m s⁻¹) to characterize HDW [3,20,21]. Zhao et al. (2022) [1] reported that exposure to HDW for merely 10 h during wheat maturity resulted in yield losses of approximately 4% (0.09 t ha⁻¹) in the U.S. Great Plains. In Australia, controlled sub-daily experiments revealed that 6 h daily HDW exposure (≥35 °C, relative humidity 20–25%) for five consecutive days before flowering reduced wheat yield by 0.15% [22]. These variations underscore the urgent need for systematic characterization of environmental parameters at finer temporal scales to precisely quantify the compound stress effects of sub-daily HDW events on wheat and to develop process-based mechanistic adaptation measures.

Recognizing these limitations, our study innovatively advances HDW assessment by shifting from conventional tri-element threshold classification to a duration-centric approach at sub-daily resolution. Building on the meteorological industry standard QX/T82-2019 [14], we establish hourly HDW events using the minimum tri-element thresholds (T ≥ 31 °C, RH ≤ 30%, U ≥ 3 m s⁻¹) and integrate them with daily exposure duration to capture their sub-daily scale dynamics utilizing high-frequency hourly meteorological observations across 599 counties in China’s winter wheat belt (1981–2020), we develop a novel hierarchical classification system based on cumulative stress duration within a day: short-duration (HDW_sd1: 1 h per day), moderate-duration (HDW_sd2: 2–3 h per day), and prolonged-duration (HDW_sd3: 4–8 h per day). This temporal disaggregation reveals critical patterns obscured by daily-scale analyses; our method captures more transient stress episodes that meet agricultural impact thresholds but escape daily-averaged detection.

Integrated foliar spraying—which typically includes a combination of foliar fertilizers, fungicides, insecticides, and plant growth regulators—has been shown to effectively enhance wheat yields by alleviating multiple stresses simultaneously. At a physiological level, foliar fertilizers containing osmoprotectants such as glycinebetaine significantly improve crop resilience to abiotic stresses, including drought and heat stress [23]. Glycinebetaine acts as a compatible osmolyte that stabilizes cellular structures, maintains membrane integrity, and preserves photosynthetic activity under drought and heat stress by protecting critical physiological processes, thereby increasing grain number per spike and overall grain yield by approximately 18% under field conditions [23]. Furthermore, foliar application of macro- and micro-nutrients (including nitrogen, phosphorus, zinc, and iron) enhances grain nutritional quality and prolongs flag leaf photosynthesis during critical growth stages, significantly delaying leaf senescence and thus ensuring continued assimilate supply during grain filling [24,25,26]. Specifically, Blandino and Reyneri (2009) [24] demonstrated that integrated foliar application at anthesis consistently increased winter wheat yields by prolonging flag leaf greenness, thus maintaining photosynthetic capacity under stress. Meanwhile, fungicides integrated into foliar spraying programs further enhance this protective effect by reducing disease pressures—particularly from Fusarium spp., rust, and septoria—thereby sustaining canopy health and photosynthetic efficiency during grain filling [24]. The application of fungicides, especially triazoles, has been reported to significantly decrease disease severity and mycotoxin contamination, indirectly promoting yield stability and grain quality [24]. Thus, the integrated application of fungicides alongside foliar nutrients can synergistically stabilize grain yields under stress conditions.

Although mechanistic studies show that integrated foliar spraying enhances wheat resilience by tightening stomatal control, boosting antioxidant activity, and prolonging assimilate supply [23,24,25], its operational efficacy at regional and landscape scales remains insufficiently quantified. Given the intensifying sub-daily HDW episodes across China’s winter-wheat belt and beyond, our study systematically quantifies yield sensitivities associated with HDW events using high-resolution (hourly) ERA5-Land reanalysis data. Specifically, to clearly identify sub-daily HDW trends across China’s major winter-wheat-producing areas, we categorize HDW exposure into sub-daily durations (1–8 h day⁻¹). We posit that a single, well-timed integrated foliar spraying application—delivering rapidly absorbed nutrients, plant-growth regulators, and broad-spectrum pest protection—can alleviate part of this damage, but only where irrigation keeps soil moisture high enough to permit foliar uptake. To rigorously evaluate whether integrated foliar spraying can mitigate HDW-induced yield damage, we utilize a quasi-experimental difference-in-differences (DiD) design across 23,960 county-year observations (599 counties, 1981–2020), controlling for freezing, extreme precipitation, and high-temperature stress. We hypothesize that integrated foliar spraying will significantly reduce yield losses from sub-daily HDW events, but only in counties with irrigation fractions that exceed a critical threshold, where soil moisture is sufficient for leaf uptake and activation of antioxidant defenses [27]. Further, by stratifying the sample according to county-level irrigation fractions, we explicitly test the physiological hypothesis that sufficient soil moisture is a prerequisite for foliar nutrient uptake and stress mitigation. Finally, spatially explicit estimates of HDW-yield elasticities and conditional integrated foliar spraying effects are derived to support tailored adaptation strategies that integrate water-saving irrigation and timely spraying interventions under increasingly frequent HDW events globally.

2. Materials and Methods

2.1. Overview of the Study Area

The study area encompasses key wheat-growing regions in Northern China (Figure 1A), including Hebei (HE, 109 counties), Shanxi (SX, 24 counties), Shandong (SD, 116 counties), Jiangsu (JS, 61 counties), Shaanxi (SN, 67 counties), Anhui (AH, 52 counties), Henan (HA, 114 counties), and Hubei (HB, 56 counties). The study area experiences a predominantly temperate monsoon climate, with regions such as HE and SD characterized by hot, humid summers and cold, dry winters, while SX, SN, and parts of HA have a continental monsoon climate with greater temperature fluctuations. AH, JS, and HB, located further south, have a subtropical monsoon climate. The study area receives annual precipitation ranging from 400 mm to 700 mm, with the majority concentrated in the summer months (June to August). However, wheat is typically harvested in early June, meaning that the critical growth stages, such as heading and grain filling, occur before the main rainy season. During this period, the winter wheat is highly vulnerable to HDW events, which can cause significant damage by accelerating water loss and reducing grain yield. This climatic mismatch between rainfall and wheat growth highlights the region’s susceptibility to environmental stressors, especially during the late growth stages.

2.1.1. Production

The yield levels vary significantly between provinces, with HB, SD, and JS generally showing higher yields, where the majority of counties have yields between 4 and 6 t/ha. In contrast, provinces such as SX, SN, and parts of HE exhibit lower yields, with many counties producing below 4 t/ha (Figure 1B). The histograms highlight the distribution of wheat productivity, showing a concentration of higher-yielding counties in the eastern plains, which benefit from favorable climatic conditions and lower elevations, while the western and southern provinces, characterized by higher elevations and more rugged terrain, tend to have lower yields.

2.1.2. Irrigation Fraction

Irrigation fraction is defined as the proportion of each county’s winter wheat area that is irrigated. We calculated this by overlaying the global gridded irrigation map [29] on our winter-wheat mask, summing irrigated wheat area, and dividing by total wheat area (average 2000–2015). Values range from 0.0 (no irrigation) to 0.8 (intensive irrigation) (Figure 1C). Higher values are predominantly observed in the plains of northern HA, southwestern SD, and central-southern HE, whereas hilly and mountainous regions such as northern SN and western SX exhibit relatively low irrigation fractions.

2.1.3. Soil Type

Figure 1D further shows that soils across the region are heterogeneous: Cambisols, Luvisols, and Phaeozems dominate the North China Plain (HE, SD, JS, eastern HA), providing medium texture and moderate water-holding capacity, whereas Kastanozems, Calcisols, and Chernozems prevail on the Loess Plateau and piedmont areas of SX and SN, where lower organic matter and higher bulk density can exacerbate moisture stress. Acrisols and Alisols occur sporadically in the southern margins (AH, HB), reflecting stronger weathering and acidity. These soil differences influence root development, available water capacity, and nutrient cycling, thereby modulating the crop’s physiological response to HDW. Incorporating both climatic and pedological contexts is, therefore, essential for interpreting spatial patterns of wheat yield sensitivity and for designing integrated mitigation strategies.

2.2. Data Preparation

2.2.1. Winter Wheat County-Level Yield Data

County-level winter wheat yield data were gathered from two sources: data from 1981 to 2000 were provided by the Ministry of Agriculture and Rural Affairs of the People’s Republic of China (http://www.zzys.moa.gov.cn, accessed on 10 May 2024), while the 2001–2020 figures were extracted from the China County Statistical Yearbook (https://www.stats.gov.cn/, accessed on 10 May 2024).

2.2.2. Climate Data

Climate data were derived from the ERA5-Land dataset [30], a global reanalysis product by the European Centre for Medium-Range Weather Forecasts (ECMWF), with a horizontal resolution of 0.1° × 0.1°. Hourly data from 1981 to 2020 included 2m air temperature, total precipitation, relative humidity, wind speed, and near-surface radiation. Because near-surface relative humidity was not directly available, it was calculated from 2m air temperature and dew point temperature using the Clausius–Clapeyron relationship between saturation vapor pressure and temperature [31]. County-level climate data were then extracted from the nearest grid points to each county centroid.

2.2.3. Irrigation Fraction Data

The irrigation data were obtained from https://zenodo.org/record/4392826#.YkEM_ehByUk (accessed on 25 May 2024) [29] and represent global irrigation intensity for 2000–2015. The dataset consists of three discrete values—0, 1, and 2—indicating no irrigation, moderate irrigation, and high irrigation, respectively, at a resolution of 10 × 10 km. In this study, we clipped and resampled these data to our winter-wheat cultivation map, then computed each county’s irrigation fraction as the irrigated winter-wheat area divided by total winter-wheat area, averaged over 2000–2015.

In the study area, irrigation is delivered primarily via surface (furrow and border-strip) and tube-well systems—with sprinkler and drip irrigation limited to research/demo sites—and an average seasonal irrigation depth of 80–180 mm applied in 1–3 events, depending on rainfall [32,33]. Fertigation is not practiced for winter wheat; all nutrient inputs are applied via soil basal/topdressing and foliar sprays.

2.2.4. Phenology Data

To allocate weather stresses to physiologically relevant growth stages, we reconstructed county-year crop calendars following the global rule-based approach of Minoli et al. (2022), implemented in the R package cropCalendars (version 0.1.0; R version 4.3.2) [34]. For each county, we first distinguished temperature versus precipitation-seasonal climates and then generated a rule-based crop calendar. Spring-type winter-wheat sowing was assigned to either the first day of the wettest 120-day window (precipitation-driven regions) or the first day on which mean air temperature at 2 m above ground level exceeded 12 °C (temperature-driven regions). Physiological maturity was timed so that grain filling occurred under the least-stressful conditions—at the onset or end of the warmest month when mean Tmax ≤ 25 °C in temperature-seasonal areas, or earlier to avoid terminal heat stress; and just before the P/PET ratio dropped below 1.0 in precipitation-seasonal areas. Thermal time was tracked as growing-degree-days (base 0 °C) accumulated from 1 January [35]. Within each simulated season, jointing and heading were fixed at 35% and 57% of total GDD to maturity, respectively, producing three analysis windows—sowing-to-jointing (SW-JT), jointing-to-heading (JT-HD), and heading-to-harvest (HD-HV)—as summarized in

Table 1. Winter wheat development stage dates in different provinces.

Province	Sowing (SW, Date ± Days)	Jointing (JT, Date ± Days)	Heading (HD, Date ± Days)	Harvest (HA, Date ± Days)
Hebei (HE)	10/4 ± 4	4/19 ± 9	5/3 ± 10	6/20 ± 13
Shanxi (SX)	9/25 ± 6	4/1 ± 10	4/29 ± 12	6/23 ± 15
Jiangsu (JS)	10/12 ± 1	4/1 ± 6	4/23 ± 5	6/13 ± 5
Anhui (AH)	10/10 ± 0	3/20 ± 6	4/10 ± 5	5/28 ± 7
Shandong (SD)	10/19 ± 2	3/29 ± 11	4/16 ± 12	6/20 ± 15
Henan (HA)	10/14 ± 5	3/19 ± 9	4/12 ± 10	6/4 ± 13
Hubei (HB)	10/30 ± 4	3/4 ± 10	4/5 ± 10	5/31 ± 11
Shaanxi (SN)	10/1 ± 11	4/8 ± 11	4/28 ± 13	6/20 ± 17

.

2.2.5. Soil Type Data

Soil type data were obtained from the SoilGrids v2 global soil database, which provides dominant WRB 2015 soil groups at a 250 m resolution [28].

2.3. Definition and Measurement of a Sub-Daily HDW Event

We first defined HDW events using threshold values aligned with China’s Meteorological Industry Standard (QX/T82-2019): temperature ≥ 31 °C, relative humidity ≤ 30%, and wind speed ≥ 3 m/s. Building on this standardized definition, we further developed a novel duration-based classification system to more accurately capture sub-daily HDW dynamics, specifically categorizing these events by their intraday duration to better reflect their biological and agronomic impacts. For each county-year observation, we first quantified the hourly occurrences of meteorological conditions meeting or exceeding these elemental thresholds during the critical heading-to-harvest growth stage. These episodes were then categorized into three distinct classes based on their intraday persistence: short-duration HDW (HDW_sd1, 1 h/day), moderate-duration HDW (HDW_sd2, 2–3 h/day), and prolonged-duration HDW (HDW_sd3, 4–8 h/day). By focusing on stress duration rather than binary threshold exceedances, this classification accounts for both transient heat pulses and cumulative exposure effects, thereby providing a more biologically relevant characterization of HDW impacts on wheat reproductive processes. As demonstrated in Figure 2, the first 31 °C exceedance for all duration classes consistently occurred between 10:00 and 16:00 local solar time, indicating that our analysis exclusively captures post-dawn HDW events and excludes physiologically distinct pre-dawn conditions.

2.4. Implementation of Integrated Foliar Spraying Technology

Integrated foliar sprays were applied using tractor-mounted or self-propelled boom sprayers, delivering 450–600 L ha⁻¹ as a fine, low-pressure mist between 09:00 and 17:00 LT under dry-leaf conditions (avoiding rain and morning dew). Aerial application was not employed except in limited research trials. This technology—combining plant growth regulators, nutrients, fungicides, and insecticides—was first piloted in HA Province, China, in 1986, with the goal of mitigating yield losses caused by HDW events. Initially tested across limited areas, the treated acreage expanded rapidly due to its demonstrated effectiveness in improving grain yield, quality, and prolonging leaf senescence [24]. After early adoption in HA Province (covering 1970 kha by 1996), the practice gradually expanded into neighboring provinces, including HE, AH, and SX. By 2004, following recognition by China’s Ministry of Agriculture and leading agricultural researchers, the area applying this integrated technology experienced rapid growth, expanding from 4660 kha in 2004 to more than 13,333 kha by 2012. Since 2012, the widespread adoption of integrated foliar spraying across China’s primary winter wheat production regions has played a pivotal role in stabilizing wheat yields, effectively mitigating yield losses caused by HDW events, particularly during critical growth stages [36].

Integrated foliar spraying consists of a single tank-mix applied once between full heading and early milk (Zadoks 55–73, about 7–10 days) that delivers four functional ingredients simultaneously. The working solution includes 0.2–0.3% potassium dihydrogen phosphate (KH₂PO₄) (about 450–600 L solution ha⁻¹) to enhance potassium and phosphorus nutrition and improve leaf water status; 10–30 mL/ha of plant-growth regulators or immune inducers (typically 24-epibrassinolide, thidiazuron, or amino-oligosaccharins) to delay senescence and enhance stress tolerance; 100–150 g of active ingredient/ha of broad-spectrum triazole fungicides (tebuconazole, propiconazole, or prothioconazole) to suppress diseases such as stripe rust, powdery mildew, and Fusarium head blight; and 8–15 g/ha of insecticides (λ-cyhalothrin or thiamethoxam) to control aphids and other piercing–sucking pests [36]. Sprays were delivered at 450–600 L/ha using tractor-mounted or self-propelled boom sprayers, avoiding application during rainfall or morning dew. Solutions were thoroughly mixed prior to application to ensure uniform concentration. National extension guidelines permit a second spray 5–7 days later only when prolonged rainfall or heavy disease pressure persists, a contingency that occurred in fewer than 8% of county-years in our panel, thus treated as part of the same “treated” category in our DiD analysis.

2.5. Method for High-Yield and Stable-Yield Association of Winter Wheat

To evaluate the high-yield and stable-yield characteristics of different eras, we used the actual winter wheat production in different periods (1981–1990, 1991–2000, 2001–2010, and 2011–2020) to reflect the high and low yield levels with the mean of winter wheat yield per county in a certain period, and the corresponding inter-annual coefficient of variation to reflect the stable yield level [37]. The calculation formula is as follows:

$${\mu }_{\mathit{ip}}={\displaystyle \frac{{{\displaystyle \sum }}_t^{t+n-1}{\mu }_{\mathit{it}}}{n}}$$

(1)

$${\mathit{CV}}_{\mathit{ip}}={\displaystyle \frac{{\sigma }_{\mathit{ip}}}{{\mu }_{\mathit{ip}}}}$$

(2)

where μ_ip is winter wheat yield (kg/ha) in county i in year t; n is the number of years; μ_ip and σ_ip are the mean and standard deviation of winter wheat yield in county i at different periods p, respectively; CV_ip is the interannual variation coefficient of winter wheat yield in county i at a specific period p.

According to the median yield μ_mp and the median interannual coefficient of variation CV_mp of the entire region in a specific period p, the four association types of high-stable-yield, high-unstable-yield, low-stable-yield, and low-unstable-yield are identified according to the method shown in

Table 2. Classification of high-yield and stable-yield association of winter wheat.

Type	Condition
High yield and stable yield	μi ≥ μmt and CVi < CVmt
High yield and unstable yield	μi ≥ μmt and CVi ≥ CVmt
Low yield and stable yield	μi < μmt and CVi < CVmt
Low yield and unstable yield	μi < μmt and CVi ≥ CVmt

, and the type to which each county belongs is determined.

2.6. Method for the Trend Analysis of Sub-Daily HDW Events

Sub-daily HDW events were examined using hourly meteorological data spanning 1981–2020 for each county. To evaluate temporal changes in the frequency and duration of these events, we applied ordinary least squares regression to the time series data for each event category (HDW_sd1, HDW_sd2, and HDW_sd3). The significance of the identified trends was rigorously tested using the Mann–Kendall analysis at a 95% confidence level, ensuring robust detection of monotonic changes. In addition, we quantified the magnitude of these trends using Sen’s nonparametric slope estimator, which effectively determines the rate of change in cases where the trend is presumed linear.

2.7. Method for Estimation Using Panel Matching with Difference-in-Differences

To quantify the effectiveness of integrated foliar spraying in reducing wheat yield losses induced by HDW stress, we implemented a difference-in-differences (DiD) analytical framework combined with panel matching. This approach compared wheat yield changes in counties frequently exposed to HDW events (defined as experiencing two or more HDW days annually since 2004) against those with fewer HDW occurrences (less than two days per year).

However, the standard DiD approach relies on the assumption of comparability between treated and control groups, which may be violated if counties differ systematically. To address this issue, we employed panel matching methods using propensity score weighting, which matched counties based on relevant agricultural and climatic covariates. Specifically, the matching process accounted for freezing stress, defined as cumulative degree-hours below 0 °C (Frez) [38], hourly extreme heat stress (EH), calculated as accumulated hourly degrees above 31 °C, cumulative HDW exposure during heading to harvest, total precipitation during critical growth stages, and the irrigation fraction (Table S1). Matching was conducted with a four-year lag to accurately capture the delayed impacts of climatic stressors and technology adoption on wheat yields.

Following panel matching, the PanelEstimate function was applied to quantify yield impacts over time. The econometric specification is presented as follows:

$$Y_{it}={\beta }_0+{\beta }_1·trea{t}_{it}+{\beta }_2·X_{it}+{\beta }_3·Irrigatio{n}_{it}+Soil typ{e}_i+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(3)

where

• Y_it is the outcome variable, representing wheat yield in county i at time t,
• treat_it is the treatment indicator, which takes the value of 1 for treated counties (HDW days greater than or equal to 2 days post-2004) and 0 for control counties.
• X_it includes covariates capturing meteorological conditions, such as temperature extremes, precipitation, and HDW exposure, Frez was estimated as the sum of degree hours <0 °C to define freezing stress [38] and extreme heat (EH) was estimated to the accumulation of temperatures above 31 °C, which, together, capture the nonlinear yield response to temperature, and different development period rainfall predictors (Prcp). HDW is accumulated over the heading harvest period.
• Irrigation_it is the irrigated fraction, defined as the irrigated area divided by the total winter wheat planting area for county i at time t.
• Soil type_it is the soil type for county i.
• μ_i represents county fixed effects, which control for time-invariant characteristics specific to each county.
• γ_t represents year fixed effects, which capture common time-varying factors across all counties.
• ε_it is the error term.

To validate our panel matching identification, we assessed covariate balance using absolute standardized mean differences before and after matching. All covariates achieved post-matching SMD < 0.2, with reductions ranging from 60.8% to 96.4%, confirming adequate balance.

We further classified counties into two groups based on irrigation fraction: areas with irrigation fractions greater than 0.37 (the 75th percentile) were defined as well-irrigated regions, while those below this threshold were classified as less-irrigated. We then estimated the differential impacts of integrated foliar spraying between these two groups.

This rigorous analytical approach allows us to reliably estimate how integrated foliar spraying mitigates wheat yield losses due to intensified HDW stress, providing valuable evidence for adaptive agricultural management strategies under climate change.

2.8. Robustness Checks

To verify that our results are not driven by model specification or spurious trends, we implemented five sensitivity analyses.

(1)

Mixed placebo—For 1000 iterations, we simultaneously randomized (a) the treated-county set (same size as the actual treated group) and (b) a pseudo-treatment start year drawn from 1996 to 2004, re-estimating the full fixed-effects DiD each time.

(2)

HDW-threshold variation—The treatment indicator was re-defined with HDW cutoffs of 1–3 days.

(3)

HDW-duration variation—The treatment indicator was re-defined with the three mutually exclusive duration counters HDW_sd1, HDW_sd2, and HDW_sd3.

(4)

Cutoff-year variation—The post period was shifted between 2003 and 2005.

(5)

Irrigation-fraction definitions—“Well-irrigated” counties were re-classified using the 25%, 50%, and 75% percentile cutoffs of the irrigation fraction distribution.

All robustness regressions retained the same meteorological and soil controls and county/year fixed effects as the baseline model; standard errors were clustered at the county level.

3. Results

3.1. High-Yield and Stable-Yield Association of Winter Wheat

From the 1980s to the 2010s, counties achieving high yields consistently demonstrated stability, indicating a strong correlation between high yield and yield stability. Over these decades, countries progressively shifted toward higher and more stable yields, reducing yield uncertainty (Figure 3 and Figure 4). High-yield and stable-yield counties initially concentrated in China’s eastern and central plains but gradually expanded both northward and southward, creating larger contiguous clusters. Conversely, low-yield and unstable-yield counties persisted mainly in the western and southern regions, areas typically constrained by drought or limited irrigation, though some localized improvements were noted. Counties classified as high-yield but unstable or low-yield but stable remained relatively dispersed, indicating potential transitional states influenced by evolving agricultural practices or climatic conditions.

Heat maps corroborated these observations by revealing the evolution in the number of counties in each category over several decades (Figure 3). Notably, SD, HE, JS, and HA emerged as the primary high-yield and stable-yield zones. In the 1980s, despite these provinces achieving relatively high yields, the distribution of counties in the high-yield and stable-yield quadrant was more scattered. By the 1990s, scatter plots indicated more pronounced clustering, with counties in SD, HE, and HA beginning to merge into continuous clusters. This trend intensified in the 21st century and was most evident in the 2010s, where the concentration of data points in high-yield and stable-yield areas markedly increased, as reflected by the corresponding heat maps.

Overall, the winter wheat high-yield and stable-yield association regions exhibited an ‘eastward shift plus expansion’ trend. The spatial differentiation of stability and yield was driven by both geographic conditions and human interventions. The primary spatial pattern—characterized by the continuous expansion of high-yield and stable-yield clusters alongside the persistent presence of low-yield and stable-yield clusters—underscores the need for sustained agricultural investment, enhanced water resource management, and technological innovation in less advantaged regions, while reinforcing the stability and high productivity of core production areas.

3.2. Spatial and Temporal Outline of the Sub-Daily HDW Events

Spatial and temporal analysis of sub-daily hot-dry wind (HDW) events from 1981 to 2020 indicated pronounced regional variability and an overall intensification trend across China’s major winter wheat production regions (Figure 5, Figure 6 and Figure 7,

Table 3. Mean slopes of HDW_sd1, HDW_sd2, and HDW_sd3 events in different provinces from 1981 to 2020.

Province	HDWsd1 (Days/a)	HDWsd2 (Days/a)	HDWsd3 (Days/a)
Hebei (HE)	0.106 ± 0.047	0.076 ± 0.037	0.026 ± 0.020
Shanxi (SX)	0.051 ± 0.034	0.039 ± 0.033	0
Jiangsu (JS)	0.004 ± 0.010	0	0
Anhui (AH)	0	0	0
Shandong (SD)	0.035 ± 0.037	0.03 ± 0.030	0.001 ± 0.005
Henan (HA)	0.015 ± 0.026	0.007 ± 0.018	0
Hubei (HB)	0	0	0
Shaanxi (SN)	0.01 ± 0.017	0.008 ± 0.020	0

and

Table 4. Mean days and proportion of counties of HDW_sd1, HDW_sd2, and HDW_sd3 events from 1981 to 2020.

	HDWsd1 (%)		HDWsd2 (%)		HDWsd3 (%)
	n	n/Total	n	n/Total	n	n/Total
1	147	24.54	176	29.38	136	22.70
2	77	12.85	54	9.02	31	5.18
3	38	6.34	42	7.01
4	30	5.01	62	10.35
5	48	8.01	22	3.67
6	28	4.67	1	0.17
7	1	0.17
Total	369	61.60	357	59.60	167	27.88

). The frequency and intensity of HDW events varied significantly by province, reflecting a clear north-to-south gradient (Figure 5).

3.2.1. Spatial Variability of Sub-Daily HDW Events

Spatially, northern provinces consistently experienced higher annual frequencies of HDW across all duration categories. The HE province, in particular, showed the highest occurrence rates, averaging 9.09 days annually for HDW_sd1, 4.42 days for HDW_sd2, and 3.49 days for HDW_sd3 (Table S2). SD and SX followed, with average occurrences of HDW_sd1 events at 4.59 and 3.45 days per year, respectively. In contrast, southern provinces such as AH and HB experienced significantly fewer HDW events, with annual averages rarely exceeding 0.5 days. Spatial mapping further highlighted this distributional disparity, with counties in Hebei and Shandong frequently experiencing annual total HDW days exceeding 7 days, and localized clusters of prolonged-duration events (HDW_sd3) identified in northern areas (Figure 5A–D).

County-level analysis revealed a significant proportion of counties experiencing frequent HDW occurrences (Table 4). For instance, 61.6% of all counties (369 counties) experienced HDW_sd1 events at least annually, with nearly 25% encountering more than one event annually. Moderate-duration HDW_sd2 events affected approximately 59.6% of counties, and prolonged-duration HDW_sd3 events occurred in approximately 27.9% of counties. Importantly, counties frequently exposed to prolonged-duration HDW events—although fewer in number—represent critical vulnerability zones requiring targeted adaptation measures. The stacked bar chart (Figure 7C) reinforced these results, clearly illustrating that northern provinces (HE, SD, SX, and parts of HA) were disproportionately affected, particularly by moderate and prolonged-duration HDW events.

3.2.2. Temporal Dynamics of Sub-Daily HDW Events

The temporal trends revealed marked regional differences in HDW event intensification over the past four decades (Figure 6). The HE province exhibited the most pronounced increasing trend, with short-duration HDW_sd1 events significantly rising by approximately 1 day per decade (0.106 days/a, p < 0.05). Moderate (HDW_sd2) and prolonged events (HDW_sd3) in HE also exhibited consistent upward trends (0.076 and 0.026 days/year, respectively). Similarly, SX showed statistically significant increases, particularly for HDW_sd1 (0.051 days/a, p < 0.05) and moderate-duration HDW_sd2 (0.039 days/a). SD and HA exhibited moderate increases primarily in shorter-duration HDW events (0.035 and 0.015 days/a, respectively), while prolonged-duration HDW events showed negligible trends in these provinces. In southern regions, including AH, JS, and HB, HDW frequencies remained stable or exhibited negligible changes, further highlighting the uneven impacts across regions (Table 4).

Segmented linear regression analyses at regional scales showed a moderate yet accelerating trend of HDW intensification over recent decades (Figure 7A). Between 1981 and 2004, the trend in annual total HDW days increased slightly but insignificantly (0.034 days/a, R² = 0.21, p = 0.33). From 2004 onward, the intensification became somewhat more pronounced, though still statistically weak (0.077 days/a, R² = 0.26, p = 0.34), indicating a gradual worsening of HDW conditions. Year-to-year variability analysis further demonstrated that shorter-duration events (HDW_sd1 and HDW_sd2) occurred more frequently, with annual averages of 3.00 and 2.49 days, respectively, whereas prolonged-duration events (HDW_sd3) occurred less frequently but persisted consistently, averaging 0.75 days annually (Figure 7B).

3.2.3. Implications for High-Yield and Stable-Yield Zones

Combining the spatial distribution of HDW days (Figure 5) with the evolving patterns of high-yield and stable-yield zones depicted in Figure 3 indicates that many core high-yield and stable-yield counties—particularly those in the northern and north-central portions of the study region—are more frequently exposed to higher HDW incidence. SD, HE, and parts of HA, which have emerged as the dominant stable and high-yield areas over time, overlap substantially with zones experiencing moderate to high HDW days. While these provinces maintain robust wheat yields and production stability, the prevalence of HDW events underscores the need for continued investment in irrigation infrastructure, heat-tolerant varieties, and region-specific management practices to mitigate potential yield losses. By contrast, counties farther south, though showing increasing trends in yield and stability, are generally subject to fewer HDW days, based on ERA5-Land reanalysis, suggesting a comparative advantage in mitigating heat stress yet still requiring attention to sustain and enhance production amid changing climate conditions.

3.3. Variation in Estimated Treatment Effects of Integrated Foliar Spraying

The results presented in Figure 8 demonstrate that integrated foliar spraying markedly improved yields, especially after large-scale adoption post-2012. Covariate balance diagnostics confirm that our matching reduced selection bias (Figure 9). Before 2012, yield differences between treated (areas adopting integrated spraying) and control practices remained modest, generally fluctuating with limited consistency. However, following widespread adoption after 2012, yield differences progressively increased, reaching approximately 12–18% by 2020. This highlights the substantial role of integrated foliar spraying technology in mitigating the adverse impacts of HDW events on wheat productivity. Physiologically, however, the technology acts very differently on short- versus long-duration HDW events. Brief heat-dry pulses (HDW_sd1) often elicit a mild “heat-priming” response that can strengthen antioxidant activity and osmotic adjustment; therefore, their yield impact is usually small or even neutral. By contrast, prolonged episodes (HDW_sd3) exceed the plant’s protective capacity, accelerating canopy desiccation and photosynthetic decline; here, the foliar spray supplies readily absorbed nutrients and protective fungicide/insecticide coverage, helping leaves stay green and functional, and therefore curbing the otherwise severe losses. This distinction is consistent with controlled-environment experiments in which wheat exposed to 35–42 °C for 1–5 days (~6 h d⁻¹) five days before anthesis suffered grain-number and yield reductions of 0.16% and 0.15% per °C h, respectively, underscoring the decisive role of exposure duration in determining damage versus acclimation [22].

The analysis further reveals pronounced differences in the technology’s effectiveness depending on regional irrigation conditions. In well-irrigated areas, the technology demonstrated greater efficacy, consistent with initial pilot results from Zhoukou City in Henan province, a highly irrigated region (Figure 1C). Driven by the substantial success of the pilot trials, integrated foliar spraying was widely adopted across major wheat-producing regions after 2004, with significant yield benefits rapidly becoming evident, particularly post-2012. By 2020, yield increases in these well-irrigated regions consistently ranged between 18% and 20% compared to control practices. Such pronounced yield improvements underline a positive synergy between sufficient irrigation availability and integrated foliar spraying, enabling plants to maximize physiological benefits from nutrient supplementation, enhanced water-use efficiency, and improved resistance to HDW-related stress.

Conversely, areas characterized by insufficient irrigation infrastructure exhibited limited and inconsistent yield responses to integrated foliar spraying. In poorly irrigated counties, the differences between treated and control yields frequently fluctuated around zero, and the improvements seldom exceeded 5%, indicating marginal effectiveness under constrained water conditions. This limited yield response emphasizes the critical role of adequate irrigation infrastructure, as water availability significantly influences crop uptake efficiency, foliar nutrient absorption, and physiological resilience during stress periods, all crucial to realizing the full potential of integrated foliar applications.

These findings indicate that while integrated foliar spraying technology effectively reduces wheat yield losses induced by HDW stress, maximizing its benefits strongly depends on adequate irrigation management. Figure 10 extends these results with four robustness exercises. The mixed-placebo test (panel A) places the observed coefficient far outside 99% of random county–year assignments, and panels B–E confirm that the effect is stable under alternative HDW thresholds, sub-daily events, treatment cutoff years, and irrigation-fraction definitions. Together, these tests reinforce the credibility of the estimated yield gains. Consequently, investments in irrigation infrastructure coupled with integrated agronomic management are essential for sustainably enhancing wheat production resilience under increasingly severe climatic stress.

4. Discussion

The increasing frequency of HDW events [1] is largely driven by the impacts of climate change, which heightens atmospheric instability and triggers more frequent, more severe compound extreme weather patterns [39,40]. HDW events are a critical concern for agriculture, particularly in wheat-growing regions where these weather conditions can cause significant yield losses [1]. In the U.S. Great Plains, compound HDW events are commonly defined at 32 °C, ≤30% RH, and winds ≥7 m s⁻¹; recent analyses show these events have risen markedly since 1982 and now cut winter-wheat yields by 4% for every additional 10 h during heading-to-maturity, translating to losses of up to 0.09 t ha⁻¹ per decade [1]. By contrast, Australia’s rain-fed wheat belt adopts a still-harsher threshold of ≥35 °C and RH < 25% (usually with ≥3 m s⁻¹ winds); controlled screens at 36 °C for three consecutive 8 h days reduced grain set and final yield by >15% compared with 28 °C controls [41]. These cross-country contrasts illustrate two key points. First, hotter and drier baseline climates demand higher HDW thresholds; yet, even so, prolonged events (our HDW_sd2–3) remain the dominant yield-loss driver in all regions examined (China, U.S., Australia) [42]. Second, the Great Plains and Australian findings corroborate our conclusion that duration, not just frequency [43], dictates damage: short one-hour pulses may induce acclimation, whereas multi-hour episodes overwhelm protective capacity—an effect our integrated foliar spraying most effectively mitigates. Incorporating such region-specific HDW definitions into global risk assessments is therefore essential for accurately gauging climate change-related threats to wheat production.

The regions with the highest wheat yield in the main winter wheat production area of China, such as HE, SD, and HA provinces, also experience more frequent HDW events, as shown in Figure 4 and Figure 5. This spatial overlap underscores the critical significance of the present study. Although these regions achieve high productivity due to favorable agro-environmental conditions—including fertile soils, advanced agricultural practices, and substantial irrigation infrastructure—their vulnerability to HDW significantly threatens yield stability. Consequently, understanding and mitigating the adverse impacts of HDW in these key wheat-producing provinces is essential for ensuring national wheat production stability and safeguarding food security.

One widely adopted adaptive management practice to enhance wheat resilience against environmental stressors, such as HDW events, is the integrated foliar spraying approach. This comprehensive technology involves the combined application of foliar fertilizers, plant growth regulators, fungicides, and insecticides, which collectively contribute to crop resilience and improved productivity under stress conditions. Fungicides play a critical role in protecting wheat plants from diseases such as powdery mildew and rust [44], meanwhile improving the health and longevity of the plant’s foliage, allowing for more effective photosynthesis during critical stages of grain filling, which can otherwise lead to significant yield losses. This enhanced leaf health directly contributes to greater grain density (GD) and improved nitrogen use efficiency (NUE) [45]. The use of fungicides supports nitrogen management in wheat by preventing the diversion of nitrogen to non-grain parts of the plant, thereby ensuring more nitrogen is available for grain development [45]. Previous research also showed that effective disease control with fungicides can increase grain yield by up to 35% by maintaining a larger active leaf area and reducing competition for nutrients between vegetative and reproductive growth stages [46]. Blandino et al. [24] found that adding a foliar feed containing macro- and micro-nutrients during critical growth stages, such as anthesis, increased the number of fertile flowers, improved grain protein content by enhancing nitrogen remobilization, and helped extend the duration of green leaf area in wheat. This delay in leaf senescence is crucial because the flag leaf contributes significantly to photosynthesis during grain filling, ultimately supporting better grain development. By maintaining a longer photosynthetically active canopy, the plants can better sustain growth under conditions of stress, such as those induced by HDW events. The application of foliar fertilizers as part of this technology has been shown to mitigate these effects by maintaining plant vigor and improving nutrient utilization during stress periods. Our analysis further confirmed these physiological and agronomic benefits. The results showed substantial yield improvements following extensive implementation of integrated foliar spraying beginning in 2004, with markedly pronounced benefits emerging after 2012. By 2020, counties employing integrated foliar spraying exhibited yield improvements of 12–18% compared to control practices, clearly demonstrating its effectiveness in stabilizing wheat productivity under increasing HDW intensities. These outcomes strongly align with earlier experimental research emphasizing the necessity of maintaining optimal leaf health and nutrient management to achieve sustainable high wheat yields in the face of escalating climate-induced stress.

Irrigation plays a critical role in maximizing the benefits of this technology, particularly in mitigating the effects of HDW events. Liu et al. (2020) found that irrigation significantly increased nitrogen partial factor productivity (NPFP) by 24%, demonstrating that adequate water supply is a prerequisite for efficient foliar-nutrient uptake [47]. Water availability enhances the dissolution of foliar-applied nutrients and their movement into plant tissues, supporting physiological processes during critical growth stages such as grain filling. Furthermore, irrigation helps maintain optimal soil moisture levels, reducing water stress during periods of HDW and allowing the protective effects of the technology to be fully realized. Adequate irrigation improves soil moisture, which, in turn, supports microbial activity and nutrient mineralization, ensuring that nutrients remain available to the plant throughout its growth cycle [48,49]. This interaction between irrigation and nutrient availability is particularly important in the North China Plain, where water resources are often limited, and periods of HDW can cause severe moisture deficits. Since full implementation across the study area in 2012, yield gains in counties that experienced two or more days of HDW events have risen from 12 to 15% to approximately 18–20% relative to those counties with fewer than two such days. Conversely, in poorly irrigated areas, the lack of water limits the effectiveness of the technology, as the protective mechanisms it offers cannot fully counterbalance the negative effects of HDW without adequate moisture. These results emphasize the need to couple precision irrigation with foliar spraying—integrating water and nutrient management—to maximize climate resilience in wheat systems.

Although our quasi-experimental design controls for time-invariant county characteristics, we did not include explicit soil-quality covariates in the DiD regressions. Recent meta-evidence drawn from 12,115 site-years across China shows that high-quality soils (medium texture, high SOM, and adequate Olsen-P) raise mean cereal yield by about 10% and cut inter-annual yield variability by about 16%, largely by buffering short-term climate stress through superior water retention and nutrient-supply capacity [50]. Figure 1D already reveals strong pedological contrasts within our study area—Cambisols, Luvisols, and Phaeozems with moderate water-holding capacity dominate the high-yield North-China Plain, whereas drier Kastanozems and Calcisols prevail on the Loess Plateau, where yields and stability are lower. Therefore, we attribute part of the north–south productivity gradient to HDW, and irrigation may, in fact, be mediated by underlying soil differences. Future work should merge county-level panel data with gridded SoilGrids-v2 attributes—e.g., volumetric water content at field capacity, SOM, and bulk density—to (i) test interaction terms between soil quality and integrated foliar spraying and (ii) quantify whether “soil improvement + integrated foliar spraying” delivers synergistic gains, especially in HDW-prone but soil-limited regions of SX and SN. Such an extension would strengthen the generality of our conclusions and help formulate site-specific “improve soil–secure water–spray once” adaptation packages for climate-smart wheat production.

5. Conclusions

(1)

The primary high-yield and stable-yield winter wheat production areas in China significantly overlap with regions frequently exposed to sub-daily HDW events, creating substantial climatic challenges for yield stability.

(2)

Integrated foliar spraying technology, involving plant growth regulators, essential nutrients, fungicides, and insecticides, effectively mitigated wheat yield losses from HDW events. Since its large-scale adoption in 2012, counties with reliable irrigation have enjoyed yield gains of 18–20%, especially when seasonal HDW exposure ≥2 days.

(3)

However, the effectiveness of integrated foliar spraying was notably compromised in areas lacking adequate irrigation infrastructure, where yield responses remained small (often <10%) and highly variable, highlighting the necessity of a dependable water supply.

(4)

To sustainably mitigate climatic risks and maintain wheat yield stability amid increasingly severe HDW stress, future adaptation strategies should integrate technological measures (such as foliar spraying and heat-tolerant varieties) with upgraded agricultural infrastructure. In water-scarce districts, priority should be given to water-saving irrigation systems to raise irrigation efficiency.

(5)

Nevertheless, agronomic consensus now recognizes that integrated foliar spraying should be embedded in a broader, multi-pronged mitigation strategy that also includes (i) soil-health interventions—conservation tillage, organic amendments, and balanced macro-/micro-nutrient management—to improve water retention and root functioning, and (ii) deployment of HDW-tolerant cultivars with superior heat-shock protein expression and stay-green traits. The synergy of healthy soils, resilient varieties, and timely foliar nutrition offers the most robust defense against compound HDW stresses.

These integrated findings provide robust scientific evidence and practical guidance for agricultural policymakers, underscoring the urgency of coordinated technological and infrastructural interventions to sustain China’s wheat productivity under evolving climate stress conditions.