Zambia lies in Southern Africa and has no seacoast, lying between latitudes −8˚ and −18.5˚ South and Longitudes 22˚ and 34˚ East (Figure 1), with a classified climate that is tropical wet and dry with minimal stretches of semi-arid climate to the Southwestern part and around the Zambezi valley. Elevation ranges from ≤371 m in low-lying Southern and eastern areas to >1729 m in the northern and northeastern highlands. This pronounced topographic gradient influences the regional atmospheric circulation, rainfall distribution, and Agro-climatic conditions. These are critical for understanding spatial variability in maize yield anomalies and their sensitivity to the Indian Ocean Dipole-related climate variability ([18]). Located in the south, it boasts a wide range of natural features and ecological processes that support its abundant wildlife spread over 752,618 square kilometers. Knowing these characteristics of Zambia is clearly crucial for addressing conservation issues and promoting sustainable land management, considering the increasing environmental challenges. With these topographic gradient influences, Zambia’s geology, geography, temperature, plants, and biodiversity are affected ([8]). Its topography modulates its rainfall distribution as the start of the season is October to March, with the austral summer being the peak of the rainy season.

Figure 1. Zambia’s boundaries in Southern Africa.

This study used various datasets, which include weather observations and Maize production, where meteorological yields were computed from data collected at the start of the task from the Zambia Statistics Agency, Ministry of Agriculture. Weather observations data sets were collected from the Zambia Meteorological Department headquarters in Lusaka. We used the provincial-level maize yield data for the last climatological period as of the 2024 crop marketing season in the post-rain season. The climatic upper and lower limits are computed by adopting some methods where similar studies were conducted at a provincial level on rice yield anomalies in China, based on the same climatic suitability methods ([32]).

The reanalysis data sets were drawn from various websites with different resolutions, including the interim https://cds.climate.copernicus.eu/datasets/derived-era5-single-levels-daily-statistics?tab=download with a resolution of 0.25˚ × 0.25˚. Precipitation data were downloaded from CHIRPS V2, which is the Climate Hazards Group infrared precipitation available at the data portal for the International Research Institute for Climate and Society. The dataset offers a high resolution of 0.05˚ × 0.05˚ ([9]; [10]). The Sea Surface temperature is the Hadley Centre global Sea Surface Temperature and Sea Ice ([25]). The other dataset was downloaded from NOAA’s website (https://www.cpc.ncep.noaa.gov/products/wesley/reanalysis2/) with a high resolution of 2.5˚ × 2.5˚. The National Center for Environmental Prediction (NCEP)/National Centre for Atmospheric Research (NCAR) has a reanalysis of a global observation network that uses a data assimilation system of meteorological variables ([12]). Climate change on the Earth’s surface is commonly quantified by the variations in the observed Sea Surface Temperature that drive other climate systems ([11]). For the total cloud amount, it was the vertical integral of clouds as classified as high, medium, and low clouds ([32]). The data used in the analysis spans from 1993 to 2024.

2.3.1. Calculation of the Indian Ocean Dipole

The Indian Ocean Dipole index was defined as the difference between the western Indian Ocean and the eastern Indian Ocean ([27]). The difference in the sea surface temperature anomalies is twofold, with positive and negative in a seesaw-like pattern. In this work, we selected the positive and Negative years by using ± 0.5, dealing with the full rainy season from October to March. The IOD selected years the paper is working with are Positive IOD events being (1995, 1998, 2007, 2019, 2020, and 2024) and negative events being (1997, 1999, 2005, and 2006) respectively. The SVD was applied to area-weighted monthly anomalies to account for grid cell convergence to the poles. Following [3], √cos (φ) weighting was applied to precipitation and SST fields prior to SVD decomposition. This ensures the cross-covariance matrix represents actual surface area rather than equal weighting at grid points. All other steps are the same. climatological monthly anomalies (1993-2024: no detrending), ONDJFM season, precipitation domain: 18.0˚S - 8.3˚S, 22.0˚E - 33.7˚E; SST domain: 30.0˚S - 20.0˚N, 30.0˚E - 120.0˚E. The formula used for the IOD is as follows.

2.3.2. Calculation of the Annual Maize Yield Anomaly

We followed some of the approaches from [32] to calculate the annual maize yield anomalies. For a given year, provincial Agrometeorological Centre P, the percentage anomaly that deviates from a normal yield based on the country’s production mean, whereby a defined 5-year running mean for the interval t − 2 to t + 2 was calculated. The 5-year running average means reducing technological trend bias and isolating climate-driven variability. The 5-year running mean interval was calculated as follows, aggregated as a country running mean based on the medium maturity maize variety.

2.3.3. Precipitation Anomaly Percentage

Precipitation Anomaly percentage was calculated as follows.

where P ¯ is the average precipitation for the research period in Zambia, calculated for the entire country; P t is the precipitation amount in a particular month, and P t ' is the anomaly percentage for a particular month in the agricultural season for the crop.

2.3.4. Climatic Suitability of Maize Yield Anomalies-Temperature Based

Earlier research ([32]) shows that climatic appropriateness has been demonstrated to be closely tied to crop yield. Hence, identifying climatic conditions is an excellent way of predicting crop yields and adaptability.

In a given year under a specific climate scenario ([29]), Climatic suitability is the best way to forecast crop yields as heat stress affects these crops in both sub-tropical and temperate regions. Most studies examine three forms of climatic suitability, and that is in accordance with temperature, duration of sunshine on the crop, and daily rainfall at early stages to maturation, with each stage varying ([32]).

The temperature is calculated using fuzzy mathematics ([7]) as a nonlinear equation. The β-function can fairly depict the nonlinear relationship between the maize crop’s development and temperature. The ideal temperature range varies above and below the mean suitability, with B being the temperature-dependent factor, and T₀ the time-dependent optimum observational temperature. The temperature suitability (ST) ranges from 0 to 1. When the temperature (T) is less than or equal to the lower limit or greater than or equal to the upper limit, the crop stops growing as temperature suitability is zero ([32]). defined as:

During growth stages of any crop, sunshine duration plays a pivotal role within a certain range, owing to the process of photosynthesis depending on the crop type and variety, with 70% daily duration being a critical value (S₀) and suitability is one ([28]). The crop response to total sunshine hours on a daily basis explains the suitability of the growth state upon attaining and exceeding this value. The value b is an empirical parameter that varies with maize growth by stage for the study period ([32]). Sunshine suitability (S_s) on maize was calculated as:

Precipitation suitability (S_p) is based on 10 days of precipitation meteorological Observations in millimeters (mm), with P being the actual precipitation. However, when precipitation is excessive, it harms the crop during the growth stages, and on the other hand, a lack of water balance at each stage results in crop failure, and the crop dies. In increased water vapor transport, the precipitation suitability is affected negatively towards the crop growth ([32]). The calculation was as follows:

To arrive at the relationship between the Indian Ocean’s SSTs and the maize yield, we employed the statistical method of Singular Value Decomposition (SVD) because it is selected as powerful in identifying coupling modes that are dominant between two (2) data set time series ([3]).

The dominant SVD mode represents the primary coupled variability of 82.7% between Indian Ocean SST anomalies and precipitation (Figure 2(a) andFigure 2(b)) over Zambia, consistent with the maximum covariance framework of ([3]). The precipitation pattern shows a clear dipole structure across Zambia, with positive rainfall anomalies dominating the southern and central regions, while negative anomalies appear over the northern sector ([27]). The results show that the SVD1 is associated with spatially contrasting rainfall responses over Zambia, suggesting a redistribution of moisture rather than uniform wet or dry conditions. The SST pattern displays predominantly negative SST anomalies across much of the equatorial and central Indian Ocean, with localized warm anomalies toward the Southwestern Indian Ocean. This pattern resembles an Indian Ocean Dipole-like structure where cooling over the central-eastern basin is linked to circulation changes that favor enhanced moisture transport toward southern Zambia. There is a variation (Figure 2(c)) coherence over time with a moderate positive correlation of r = 0.48, indicating a close relationship. This shows divergence between two series, implying that other climate drivers like ENSO also influence variability that is dominated by large-scale oscillations over the region ([21]). The spatial loading is uniformly positive across most of the country, demonstrating that both wet and dry years tend to occur synchronously across Zambia rather than exhibiting strong regional contrasts. The coherence is suggested to be spatially controlled by large-scale general circulation rather than either mesoscale or local processes, with variance that is attributed to basin-scale circulation anomalies linked to tropical SST forcing ([15]). The consistency of these results in Southern Africa highlights the robustness of large-scale oceanic-atmosphere teleconnection in shaping the entire regional hydroclimate.

Table 1.Parameters used to determine the climatic variables S_T, S_P, and S_s during maize growth and development suitability stages were determined by 9 agrometeorological observations in Zambia.

Figure 2. Coupled variability between the Indian Ocean (SST) anomalies and precipitation revealed by Singular Value Decomposition (SVD). (A) Precipitation anomaly pattern over Zambia and (B) Associated SST anomaly pattern over the Indian Ocean. The correlation coefficient of the (TCC) of the precipitation (blue line) and IOD-like SST (red) is 0.48. The stripling indicates statistically significant regions at the 99% confidence level. SVD1 describes 82.7% variance.

The IOD index is defined based on the SST anomaly as the difference between the western and the southeastern region within 10˚S - 10˚N, 50˚E - 70˚E, and 10˚S to the equator, 90˚E - 110˚E regions, respectively. Attention should be drawn to the central, southern, and some parts of eastern Zambia (Figure 3), where there are more maize farmers, indicating a higher percentage in the total maize yield anomaly for the study area. Large-scale circulation anomalies associated with the IOD phases are illustrated. During the negative IOD events, anomalous cool SSTs in the western Indian Ocean and warm SSTs in the Eastern basin modify the zonal SST gradient. This alters the Walker circulation, thereby strengthening the easterly low-level inflow towards southern Africa ([27]; [1]).

Figure 3. (A) Relative maize cultivated area intensity (% of maximum area) and (B) Pearson correlation coefficient between IOD index and maize yield anomalies in Zambia. Black dots in the panel indicate statistically significant correlations at a 95% confidence level from 1993 to 2024.

Figure 4. Composite anomalies during positive (left) and negative(right) Indian Ocean Dipole (IOD) phases: wind vectors and shaded anomalies (a, b) 200 hPa; (c, d) 500 hPa; and (e, f) 850 hPa. Shading indicated anomaly magnitude. Units consistent with the color bar and reference vector length corresponds to13 m/s.

At 850 hPa, these conditions modulate enhanced moisture transport from the Southwest Indian into the Sub-continent over Africa, complemented by Northerly inflow from the Congo basin. The convergence of these moisture streams over the study area results in a strong low-level moisture convergence that creates (Figure 4) conditions that are favorable for an unstable atmosphere through enhanced convective developments following the rising motion. Vertically integrated moisture flux diagnostics confirm further moisture availability during negative IOD years, favoring wet conditions. At level 500 hPa, the negative phase is associated with anomalous cyclonic circulation and vertical motion supporting deep convection by facilitating moisture transport and cloud formation. At 200 hPa, enhanced divergence aloft indicates efficient mass evacuation from the upper troposphere, reinforcing sustained convective activity. These connections at low, mid, and high level upward coherence upper level divergence represent a dynamical configuration for wet austral summers over southern Africa classically ([31]; [20]). During the positive phase, in contrast, a weakened easterly moisture transport reduces inflow from the Congo Basin, and dominant low-level divergence over the study area is accompanied by anticyclonic circulation at mid-levels and pronounced subsidence that suppresses convection and limits rainfall. This underscores the sensitivity of Zambia’s hydroclimate to the Indian Ocean SST gradients.

Figure 5. Time series of standardized maize yield anomalies (top orange) and agroclimatic suitability indices (bottom) for temperature (TMP_Suit, red), precipitation (Precip_Suit, blue), and sunshine duration (Sun_Suit, yellow). All variables are expressed in standard deviation from their long-term means.

Among the temporal evolution of the proposed climatic suitability variables based on precipitation, temperature, and sunshine duration, precipitation suitability exhibits the strongest and most immediate response to the general circulation (Figure 5) represented by the IOD. The negative phase is characterized by sustained improvement in precipitation over Zambia, particularly during the tasseling and grain-filling maize growth stage when moisture demand is at peak. Physiologically, water stress during these stages reduces kernel number and increases gain weight. This leads to substantial crop yield losses even when earlier growth conditions are analyzed as favorable ([14]; [5]). The strong alignment between precipitation suitability and yield anomalies, therefore, reflects both climatic control and crop sensitivity with a correlation coefficient of 0.45, p = 0.01, and −0.21 for negative and positive IOD years, respectively. Temperature suitability shows a weaker but still notable influence on yield variability. Beyond the upper limit (Table 1), the flowering stage can exacerbate moisture stress and reduce pollen viability ([23]). In this study, temperature suitability remains within an optimal threshold due to enhanced Total Cloud Cover and evaporative cooling conditions that reinforce the beneficial effects of increased rainfall. Sunshine duration on maize decreases during wet years because cloud amount increases; however, it does not offset the crop from improved soil moisture and reduces heat stress. These results are consistent with previous studies in rainfed agricultural tropical systems, which identify moisture availability as the primary limiting factor for maize production ([6]; [24]).

The compound effect of the El Niño Southern Oscillation (ENSO) and the Indian Ocean Dipole interactions demonstrates that the impacts of the IOD on maize yield anomalies cannot be considered in isolation. Instead, it operates within a broader climate system where interactions with ENSO modulate both the magnitude and spatial expression. These interactions intensify, leading to severe droughts and widespread crop failures, while on the other hand, they improve yield potential. However, excessive rainfall may also increase risks of waterlogging, nutrient leaching, and crop disease, highlighting the nonlinear nature of climate-crop interactions ([2]; [9]).

The IOD exerts a statistically significant and spatially coherent control on Zambian precipitation, producing a meridional dipole characterized by enhanced rainfall in northern Zambia during the positive phase and suppressed rainfall in the southern part during negative IOD phases (Figure 6), with the opposite pattern during the negative phase ([27]). With the negative phase breaking the barrier, the country is considered saturated during the peak months, hence favorable for yield. This response reflects the large-scale reorganization of moisture transport and vertical motion (Figure 7) associated with Indian Ocean convection anomalies. This demonstrates that Zambia‘s hydro climate variability is strongly modulated by the tropical Indian Ocean circulation. For instance, the 1997-1998 El Niño and favorable IOD event resulted in a 50% drop in Zambian crop output ([13]) and attracted a severe drought. Lanina is, and IOD are capable of boosting rainfall, therefore increasing flood hazards as seen during the 2010/2011 farming season ([33]). Changes in atmospheric circulation, including the subtropical jet stream and local low-pressure systems, mediate these interactions. Notwithstanding their importance, the combined impacts of ENSO and IOD on Zambian rainfall and maize yields are still understudied.

Figure 6. Indian Ocean Dipole influence on precipitation over Zambia: (A) Correlation coefficient between IOD index and seasonal precipitation; (B) Climatological mean precipitation from 1993-2024; (C) Composite anomalies for precipitation (%) during negative IOD phases, and (D) Composite precipitation anomalies (%) during positive IOD phases. Dots indicate statistically significant correlations at a 95% confidence level.

A key contribution of this study is the explicit demonstration of a physically consistent pathway linking general circulation, represented by the Indian Ocean Dipole, to maize yield anomalies in the study area. The negative Indian Ocean Dipole phase modifies the Walker circulation, enhances easterly moisture transport, and promotes vertically coherent ascent over southern Africa, resulting in improved precipitation during critical maize growth stages.

The water vapor transport is seen in the asymmetry in the circulation response over Zambia, with both the positive and the negative IOD years exhibiting a distinct pattern of moisture transport variability. The confidence levels differ markedly (Figure 8) between phases. During positive IOD years, significant anomalies are more widespread and coherent along moisture transport pathways into central Southern Africa. This suggests that suppressed moisture inflow and enhanced subsidence associated with positive IOD phases constitute a robust and dynamically consistent circulation anomaly, leading to a reliable reduction in moisture deliver into Zambia while the negative phase shows enhanced moisture flux anomalies toward the region, the corresponding significant points are relatively sparse indicating that the wet circulation response is less stable and more influenced by interannual variability and interactions with other climate drivers like the ENSO and the Angola low pressure system. This asymmetric response implies that the drying influence of positive IOD is more systematic and reproducible than the wet phase of the negative IOD, thereby consistent with the previous studies emphasizing the dynamical impact of the tropical Indian Ocean variability on Southern African climate ([27]; [22]; [17]).

Figure 7. (A) Mean of vapour flux in the 1993-2024 rain season; (B) Relative bias of water divergence in positive IOD years; (C) Relative bias in IOD negative years. Water vapour flux divergence is indicated by depletion of more moisture in Zambia; Colours and the vectors indicate the pathway of water vapor flux. Dotted areas are significant points in a two-tailed t-test at a 95% confidence level.

Figure 8. (A) Sunshine suitability relative bias in positive IOD years; (B) Negative IOD years; (C) Average sunshine duration in October-March from 1993-2024 in Zambia; (D) - (F) is for temperature, and (G) - (I) is for precipitation. The precipitation suitability has a close relationship with maize yield anomaly in Zambia.

The relationship between monthly precipitation anomalies and annual maize yield anomalies was assessed by computing the correlation coefficient for a large range of precipitation anomaly percentages (0% - 200%) for the period from October to April (Figure 9). We see precipitation during the mid-late growing season (December through March) exhibits positive correlations with maize yield, with the strongest association appearing in February and March. Specifically, correlation coefficients in February increased progressively from moderate to high values (r = 0.52 - 0.71) with correlations statistically significant at the 95% confidence level. This indicates that above normal rainfall in this period is associated with positive yield anomalies. By contrast, early-season precipitation shows weak correlation with yield, and many of these correlations were not significant, suggesting limited effects of early rainfall on yield. April correlations were generally low and non-significant; hence, the study period and main season ended in March. In crop culture ([32]), there is a threshold in percentages, and only when a crop hits that threshold in extreme years will it have yield anomalies. In rice crop culture, an anomaly of 40% around July affects the crop.

Figure 9. The annual precipitation anomaly in October-April, respectively, is compared with the annual maize yield anomaly. Boxes with (*) passed the significant test at the 95% level of significance.