The research, led by Chinyere P. Okechukwu of Lovely Professional University together with colleagues at Nnamdi Azikiwe University and Ogbonnaya Onu Polytechnic, assembled 43 years of quality-controlled monthly rainfall records, spanning 1981 to 2023, from 38 stations across the country. Rather than treating Nigeria as a single statistical unit, the team first asked whether the data themselves reveal natural rainfall regions. Using ensemble clustering on a 15-dimensional feature space capturing seasonal, magnitude and variability characteristics, they identified four homogeneous rainfall zones, with a Silhouette coefficient of 0.375 and a Davies–Bouldin index of 0.696 quantifying the coherence of the resulting groups. The zones include a Coastal Humid region, a Transitional Sub-Humid belt, a Northern Arid/Sahelian zone and a smaller cluster of anomalous or outlier stations that behave unlike their neighbours.

With the regional map in hand, the authors turned to trend detection. Mann–Kendall tests, a standard non-parametric tool for identifying monotonic change in noisy hydrological series, revealed statistically significant wetting trends in two zones that might surprise observers accustomed to narratives of Sahelian desiccation. The Transitional Sub-Humid zone showed a trend statistic of Z = 3.516 with a p-value of 0.0004 and a slope of 0.607 millimetres per year, while the Northern Arid/Sahelian zone recorded Z = 5.065, p < 0.0001, and a slope of 0.529 millimetres per year. Crucially, the increases were concentrated in the June-to-September monsoon season, the window when most of the region’s crops depend on rain and when saturated soils are most vulnerable to flooding.

Trends in the mean, however, tell only part of the story. Extreme rainfall events, not averages, destroy infrastructure and displace communities. To characterise those extremes, the team fitted Generalized Extreme Value (GEV) models, the workhorse of hydrological risk analysis, and allowed the distribution’s parameters to vary through time rather than assuming a stationary climate. Model comparison using the Bayesian Information Criterion showed that non-stationary formulations were consistently preferred, with ΔBIC values ranging from 7.52 to 189.07 across zones, evidence that the statistical behaviour of rainfall extremes has genuinely shifted over the four-decade record. Interestingly, when it came to estimating return levels, a stationary LogNormal distribution provided the best overall fit, a reminder that model choice in extreme-value analysis remains a careful balancing act between flexibility and parsimony.

The return-level estimates carry direct engineering consequences. The 100-year rainfall, the depth expected to be exceeded on average once per century, ranged from 382.6 millimetres in the Northern Arid zone to 643.8 millimetres in the Coastal Humid zone. Because these figures are rising against a background of non-stationarity, drainage systems, culverts and flood defences designed decades ago using historical stationarity assumptions are increasingly likely to be undersized. The authors argue that their zone-specific return levels offer a more defensible basis for climate-resilient infrastructure design than national averages or legacy design manuals.

The predictive phase of the framework compared a battery of machine-learning models across the four zones, and the results underline a central lesson of modern applied machine learning: no single algorithm wins everywhere. Random Forest achieved the strongest performance in the Outlier, Transitional and Northern Arid zones, with Nash–Sutcliffe Efficiency (NSE) values ranging from 0.654 to 0.894, while Support Vector Regression took the lead in the Coastal Humid zone with an NSE of 0.735. To test whether the models had genuinely learned transferable rainfall physics rather than station-specific quirks, the team employed Leave-One-Cluster-Out cross-validation, training on three zones and predicting the held-out one, which yielded NSE values from 0.220 to 0.766, a more sobering but honest picture of spatial generalisation.

Deep learning added a further twist. Across repeated runs, standard Long Short-Term Memory (LSTM) networks consistently outperformed their Bidirectional counterparts in the Coastal Humid, Transitional and Northern Arid zones, posting mean NSE values of 0.625, 0.778 and 0.870 respectively, against 0.465, 0.368 and 0.353 for the Bidirectional LSTM. The two architectures performed comparably only in the Outlier/Anomalous zone. Because bidirectional models read the input sequence forwards and backwards, they are often assumed to extract richer temporal context; here, that extra capacity conferred no consistent advantage, suggesting that for monthly rainfall prediction the relevant information is largely causal, carried by antecedent conditions rather than future context.

What elevates the study beyond a leaderboard of algorithms is its commitment to explainability. Using SHapley Additive exPlanations (SHAP), a game-theoretic method that attributes each prediction to individual input features, the authors found that Lag-1 rainfall, the previous month’s total, was the dominant predictor in every zone, with mean absolute SHAP values between 51.0 and 92.8 millimetres. More strikingly, trend-related features carried meaningful attribution weight, revealing that the models had absorbed the long-term intensification signal embedded in the record, particularly in the Sahelian north. In other words, the networks were not merely memorising seasonality; they were detecting the slow drift of a changing climate.

Equally notable is what the models did not find. Indices of the El Niño–Southern Oscillation (ENSO), the Atlantic Multidecadal Oscillation (AMO) and the Intertropical Convergence Zone (ITCZ) proved non-significant across all cluster stations, indicating that, at the monthly scale and within these zones, local antecedent rainfall and embedded trends dominate predictability over large-scale teleconnections. For forecasters in West Africa, that is an actionable result: it suggests that skilful monthly outlooks can be built primarily from local observation networks, even where global climate-index-based forecasting underperforms.

The framework’s broader significance lies in its integration. Regionalisation, trend analysis, non-stationary extreme-value modelling, deep learning prediction and SHAP-based attribution are usually pursued as separate exercises; here they form a single pipeline that connects data-driven rainfall zoning to physically interpretable, zone-specific insight. The authors position the work as a foundation for climate-resilient infrastructure design, flood risk management and adaptive water resource planning in Nigeria and in other monsoon-driven regions facing similar gradients. As rainfall extremes continue to drift away from the stationary past, tools that are simultaneously accurate and transparent may prove as important as the rain gauges themselves.

Subject of Research: Explainable machine learning for non-stationary rainfall modelling and extreme value analysis in Nigeria

Article Title: A non-stationary and explainable machine learning framework for adaptive rainfall modelling

Article References: Okechukwu, C. P., Anabike, I. C., Ubaoji, K. I., Ugorji, C. C., Paul, R. U., & Obulezi, O. J. (2026). A non-stationary and explainable machine learning framework for adaptive rainfall modelling. Theoretical and Applied Climatology, 157(10), Article 617. https://doi.org/10.1007/s00704-026-06583-3

Image Credits: AI Generated

DOI: 10.1007/s00704-026-06583-3

Keywords: rainfall modelling, machine learning, Nigeria, extreme value analysis, GEV distribution, LSTM, SHAP explainability, Mann-Kendall trend test, monsoon, climate non-stationarity, flood risk, Sahel

The data reveal a climate that refuses to sit still. Relative humidity in the study area fluctuated between a low of 42.46 percent in 2006 and a high of 65.09 percent in 2022, a range wide enough to alter evaporation rates, disease pressure, and the water balance of paddy fields. Maximum temperatures ranged from 36.88 degrees Celsius in 2022 to a scorching 42.27 degrees Celsius in 2010, while minimum temperatures displayed an even more dramatic spread, dipping to 5.48 degrees Celsius in 2015 and climbing to 11.96 degrees Celsius in 2017. Such nocturnal temperature swings matter enormously for rice, because cool nights slow crop development while warm nights accelerate respiration and erode the biomass a plant has spent the day building.

Rainfall, the single most consequential variable for rainfed and irrigated agriculture alike, proved the most volatile of all. Annual totals climbed from 454.9 millimeters in 2018 to an extraordinary 3,101.6 millimeters in 2021, before collapsing back to 1,505.9 millimeters in 2022. To put those numbers in perspective, the 2021 figure was nearly seven times the 2018 total, a swing that no farming calendar can comfortably absorb. The authors found that such excessive annual rainfall induced flooding, particularly during critical growth stages of the rice crop, and that these flood events translated directly into reduced yields. For a region where rice is both a staple and a commercial crop, the implications are sobering.

Methodologically, the study leaned on a battery of statistical tools designed to extract reliable signals from noisy environmental data. Descriptive statistics such as means and percentages established the baseline climatology, while Spearman’s rank correlation, tested for significance with Student’s t-test, quantified the strength and direction of associations between each climatic variable and rice yield. To detect whether the underlying climate variables were trending over time rather than merely fluctuating, the researchers applied the Mann-Kendall non-parametric trend test, a workhorse of hydro-climatological research that is robust to outliers and does not assume normally distributed data. Sen’s Slope estimator complemented the Mann-Kendall test by providing a robust measure of the magnitude of any trend, expressed as the rate of change per year.

The correlation results upended some intuitive expectations. Maximum temperature showed a positive correlation with rice yield, with a Spearman coefficient of 0.311, and minimum temperature was similarly positive at 0.301. Rainfall, by contrast, was negatively correlated with yield at -0.455, and relative humidity was also negative at -0.255. In other words, in this irrigated setting, hotter years were associated with better harvests while wetter, more humid years were associated with worse ones. The most plausible explanation lies in the fact that Kadawa’s rice is grown under the Kano River Irrigation Project, where water supply does not depend primarily on rainfall. In such systems, abundant sunshine and warmth can boost photosynthesis and crop growth, whereas torrential rainfall overwhelms drainage infrastructure and floods fields at sensitive stages such as flowering and grain filling.

This finding carries a broader lesson for climate impact research: the direction of climate-yield relationships is not universal but depends on the farming system in question. In rainfed rice systems across West Africa, rainfall and yield typically move together, and drought is the dominant threat. Kadawa’s irrigated regime inverts that logic. The negative rainfall-yield correlation of -0.455 is the strongest association in the dataset, suggesting that water excess, not water scarcity, is the principal climatic hazard facing farmers there. It is a reminder that adaptation strategies must be tailored to local hydrological realities rather than copied wholesale from other regions, and that irrigation schemes themselves can create new vulnerabilities when drainage lags behind water delivery.

The marked temporal instability documented in the study echoes a wider body of evidence on climate and agriculture. Global analyses have estimated that climate variation explains roughly a third of worldwide crop yield variability, and West Africa has repeatedly been flagged as a hotspot where rainfall variability and hydroclimatic extremes intersect with food insecurity. Within Nigeria specifically, recent studies have documented warming trends in major cities using the same non-parametric methods employed at Kadawa, and modeling work with the APSIM crop simulation platform has projected climate change effects on rice production across the wider Kano River Irrigation Scheme. The Kadawa study grounds these larger-scale findings in a concrete, thirty-year local record, showing that the variability long predicted for the Sahel is already inscribed in the region’s yield statistics.

The physiological mechanisms linking the observed climate extremes to rice performance are well understood from plant science. Rice is particularly sensitive to temperature during its reproductive phase: daytime temperatures above roughly 35 degrees Celsius can cause spikelet sterility by disrupting anther dehiscence and pollen viability, while high night temperatures increase respiration losses and shorten the grain-filling period. Cold minima, such as the 5.48 degrees Celsius recorded in 2015, can delay seedling establishment and slow early vegetative growth. Flooding, meanwhile, deprives roots and soil of oxygen, promotes toxic reduced compounds, and can fully submerge plants, with complete submergence for more than a few days often proving fatal to intolerant varieties. Humidity shapes the microclimate of the canopy and influences the incidence of fungal diseases such as sheath blight, which thrives under warm, moist conditions.

Against this backdrop, the authors argue that rice cultivation in Kadawa is growing steadily more vulnerable to climatic variability, and they outline a portfolio of adaptive responses. Chief among these are climate-resilient rice varieties, including submergence-tolerant cultivars. Breeding programs have made notable progress here: the SUB1A quantitative trait locus, originally identified in the Indian landrace FR13A, confers tolerance of complete submergence and has been introgressed into popular varieties across South and Southeast Asia, with ongoing research extending these gains and elucidating the underlying metabolic and physiological mechanisms. Deploying such genetics in West African irrigation schemes could buffer harvests against the flood years that the Kadawa record shows are increasingly disruptive.

Equally important, the study emphasizes, are engineering and water-management measures. Improved drainage infrastructure would allow excess rainfall to be evacuated from fields before it reaches damaging depths, directly addressing the negative rainfall-yield relationship at the heart of the findings. Enhanced water-management practices, from regulated irrigation scheduling to field-level water control, would help farmers exploit Kadawa’s warmth and sunshine while shielding the crop from moisture extremes. The authors frame these interventions within the broader agenda of climate-smart agriculture, which seeks to sustain high and stable yields under changing climatic conditions rather than simply maximizing output in good years. As the 2021 to 2022 rainfall collapse demonstrates, the cost of inaction is not hypothetical: a farming system calibrated to one year’s climate can be blindsided the next. With Nigeria among West Africa’s largest rice producers and demand for the grain still rising, the thirty-year Kadawa record offers both a warning and a practical roadmap for keeping harvests stable on a warming, whipsawing climate frontier.

Subject of Research: Long-term climate variability and its effects on rice yield in Kadawa, Kano State, Nigeria

Article Title: Assessing long‑term climate variability and its effects on rice yield in the Kadawa Area, Kano State, Nigeria

Article References: Assessing long‑term climate variability and its effects on rice yield in the Kadawa Area, Kano State, Nigeria. (n.d.). https://doi.org/10.1007/s00704-026-06578-0

Image Credits: AI Generated

DOI: 10.1007/s00704-026-06578-0

Keywords: climate variability, rice yield, Nigeria, Kano State, Kadawa, rainfall, temperature, flooding, irrigation, Mann-Kendall trend test, climate-smart agriculture, submergence tolerance

The Ångström–Prescott model, extended to practical use by John Prescott in 1940, estimates global solar radiation, the total shortwave energy arriving at the surface, from the ratio of actual bright sunshine hours to the maximum possible daylight hours. Its two coefficients, conventionally labelled a and b, translate the fraction of clear-sky sunshine into a fraction of extraterrestrial radiation. The Food and Agriculture Organization’s influential FAO-56 handbook, which underpins crop water requirement calculations worldwide, suggests default values that assume a relatively clean, stable atmosphere. India’s atmosphere, as the new study demonstrates, is anything but that.

Led by Ashutosh Kumar Misra of the India Meteorological Department in Pune, with colleagues from the department’s regional centres in Guwahati and Chennai, the team derived station-specific coefficients from collocated daily radiation and sunshine-duration records. The network-wide means came out at a = 0.336 with a standard deviation of 0.079, and b = 0.364 with a standard deviation of 0.122. Those averages alone would not be alarming, but the spatial pattern is striking: the largest departures from FAO-56 defaults occurred precisely at stations burdened with heavy aerosol loading across the Indo-Gangetic Plain, Rajasthan, and the western coast, where dust, smoke, and industrial pollution scatter and absorb incoming sunlight in ways the default coefficients never anticipated.

To judge how well the calibrated model performed, the authors deployed a demanding seven-metric evaluation framework rather than relying on a single goodness-of-fit statistic. They computed the coefficient of determination, mean absolute error, root mean square error, normalised RMSE, Willmott’s index of agreement, mean bias error, and the Kolmogorov–Smirnov integral, then synthesised all seven into a Global Performance Index. This composite approach matters because different metrics capture different failure modes: a model can correlate well with observations yet carry a systematic bias, or track the shape of the seasonal cycle while missing its amplitude. Collapsing the picture into one number risks hiding exactly the flaws that matter for irrigation scheduling or solar farm siting.

The headline numbers are respectable. Across the network, the calibrated model achieved a mean coefficient of determination of 0.75, a root mean square error of 1.75 megajoules per square metre per day, a normalised RMSE of 9.95 percent, and Willmott’s index of agreement of 0.92, a value close to the theoretical maximum of one. For a model that requires nothing more than a sunshine recorder and a table of extraterrestrial radiation, that level of skill explains why the Ångström–Prescott formulation has survived a hundred years of competition from far more elaborate schemes, including neural networks and satellite retrievals.

But the seasonal breakdown tells a more nuanced story, and it is here that the study becomes genuinely useful for practitioners. The pre-monsoon months of March through May produced the weakest correlations, with a network-mean R² of just 0.50, even though the RMSE of 1.75 megajoules per square metre per day matched the annual figure. The culprit is day-to-day aerosol variability superimposed on transient convective cloud development. In the hot dry season, dust storms, crop residue burning, and the first thunderstorms of the year can transform the atmospheric transmission of sunlight within hours, decoupling the sunshine-duration record from the radiation actually measured at the surface. A pyranometer responds to every attenuating process; a sunshine recorder, by contrast, only registers when direct beam irradiance exceeds a threshold, so the two instruments can disagree most dramatically exactly when the atmosphere is most turbulent.

The monsoon season posed a different problem. There the model recorded its largest root mean square error, 2.02 megajoules per square metre per day, alongside a systematic positive bias of 0.88 megajoules per square metre per day, meaning the model overestimated the radiation reaching the ground. During the southwest monsoon, thick stratiform and cumuliform cloud decks, elevated humidity, and aerosol-cloud interactions suppress surface radiation far below what the sunshine fraction alone would suggest. For hydrologists computing evapotranspiration during the very season when Indian agriculture depends most critically on accurate water balance estimates, that bias is not a statistical curiosity but a potential source of systematic error in crop water demand calculations.

Beyond calibration, the study turned its attention to the long-term trajectory of surface solar radiation itself, applying the Mann–Kendall nonparametric trend test to annual mean values at each qualifying station. The verdict was sobering: statistically significant downward trends emerged at the majority of stations examined. This finding aligns India with the broader global phenomenon of solar dimming documented in the early 2000s, when researchers first showed that declining surface radiation across many continents was linked to increasing anthropogenic aerosols. Over the Indo-Gangetic Plain, where aerosol optical depths are among the highest in the world and continue to be fed by urbanisation, industrial emissions, and seasonal biomass burning, the dimming signal appears firmly entrenched in the observational record.

The implications ripple outward across several domains. Agricultural water management in India relies heavily on FAO-56 evapotranspiration methods, which use solar radiation as a key input; using uncalibrated default coefficients in aerosol-laden regions risks misestimating irrigation demand by meaningful margins. Solar energy resource assessment, a sector expanding rapidly as India pursues ambitious photovoltaic deployment targets, likewise depends on accurate historical radiation records to rank candidate sites and forecast plant output. And climate modellers have long noted that global models tend to underestimate the direct radiative effects of aerosols over India, a discrepancy that ground-based calibration studies like this one help to quantify and ultimately correct.

What makes the study particularly valuable is its provenance. The data come from the India Meteorological Department’s own operational network, the same institution whose 150-year history of observation underpins much of South Asian climatology, and the daily radiation records are available through the department’s data service portal in Pune for researchers who follow the prescribed procurement procedure. By grounding a century-old empirical relationship in four decades of modern measurements and testing it with a rigorous multi-metric framework, the authors have delivered both a practical toolkit, regionally tuned coefficients that any Indian water manager or solar planner can adopt, and a warning written in the language of statistics: the sunlight reaching India’s surface is fading at many stations, and the models we use to track it must be recalibrated to see that clearly.

Subject of Research: Calibration of the Ångström–Prescott solar radiation model and long-term surface solar radiation trends across India

Article Title: Calibration and long-term trends of the Ångström-Prescott model for surface solar radiation across India

Article References: Misra, A. K., Ghosh, K., Karale, M., & Kumar, A. (2026). Calibration and long-term trends of the Ångström-Prescott model for surface solar radiation across India. Theoretical and Applied Climatology, 157(10), Article 639. https://doi.org/10.1007/s00704-026-06567-3

Image Credits: AI Generated

DOI: 10.1007/s00704-026-06567-3

Keywords: Ångström–Prescott model, solar radiation, India, sunshine duration, aerosols, solar dimming, FAO-56, Mann–Kendall trend test, India Meteorological Department, evapotranspiration, Indo-Gangetic Plain, Global Performance Index

Mozambique holds approximately 300,000 hectares of mangroves, the third largest extent in Africa and the thirteenth largest in the world. These salt-tolerant forests, dominated in the south by species such as Avicennia marina, Rhizophora mucronata, Bruguiera gymnirrhiza, Ceriops tagal and Xylocarpus granatum, buffer shorelines against storms, sustain fisheries, cycle nutrients, and store substantial quantities of carbon. Yet the southern provinces have not been spared. Between 2008 and 2015, an estimated 5,898 hectares of mangrove were lost in the region, with the heaviest impacts recorded in Maputo, Inhambane and Gaza provinces. Timber extraction, coastal population growth, agricultural expansion, aquaculture, and intensifying climate impacts have all contributed to the pressure.

In response, Mozambique has woven mangrove conservation into its national policy framework. The country’s commitments under Sustainable Development Goal 14 were formalized through Decree 103/2019, which established eight marine protected areas, all of which contain mangroves. A National Mangrove Restoration Strategy and Action Plan covering 2015 to 2024 remains in effect, and it identifies systematic assessment and monitoring as central pillars of successful restoration. The new study speaks directly to that need, providing exactly the kind of long-term, spatially explicit evidence the strategy calls for.

Technically, the research is a tour de force of open satellite data. The team combined imagery from Landsat 7 ETM+, Landsat 8 OLI/TIRS, and Sentinel-2 MSI, harmonized to a common 30-meter resolution and processed on the Google Earth Engine platform. Cloud screening with a threshold of 20 percent or less ensured radiometric reliability, and Sentinel-2 data were resampled to match the Landsat record. The researchers then deployed a Random Forest classifier, configured with 100 trees, to sort every pixel into seven land-cover classes: water, mangrove, forest, habitation, cultivation, prairie, and exposed soil. Between roughly 209 and 896 training points were collected per class in each region, totaling close to 3,000 samples per site, with validation points collected independently to avoid inflated accuracy estimates.

The classification performed robustly. Overall accuracy exceeded 0.80 in all four regions and reached a maximum of 0.96, while Kappa coefficients ranged from 0.76 to 0.95, indicating strong to excellent agreement with reference data. Water and mangrove classes were the most reliably mapped, with producer’s and user’s accuracy consistently above 0.94 in most regions. Forest also performed well. The greatest confusion arose in the heterogeneous landscapes of Tembe and Limpopo, where settlements, cultivation, and savanna-like vegetation overlap spectrally, a well-known challenge in coastal land-cover mapping.

Beyond the classified maps, the team computed a battery of spectral indices to probe vegetation health and hydrology. These included the Normalized Difference Vegetation Index and Enhanced Vegetation Index as general vigor measures, the Mangrove Vegetation Index as a mangrove-specific probe, and water- and soil-related indices such as NDWI, MNDWI, and NDSI. Temporal trends were tested with the non-parametric Mann-Kendall test, quantified with Kendall’s tau and 95 percent confidence intervals. The results showed statistically significant upward trends in vegetation indices across all regions, while water-related indices consistently indicated hydrological reduction or retraction. The mangrove-specific indices confirmed the same modest improvement, though the confidence intervals included zero, signaling uncertainty in effect size even where the direction of change was clear.

The spatial patterns are where the story becomes genuinely compelling. Tembe recorded a net mangrove gain of 41 hectares relative to its base year, with the water class stable throughout. Limpopo and Belavista also registered slight increases, which the authors link to restoration measures including hydrological rehabilitation and the restoration of tidal dynamics, alongside mild temperatures favorable to mangrove growth. In Belavista, declining cultivation and prairie classes appear to have been colonized by forest and mangrove vegetation, suggesting that abandoned or degraded land is being reclaimed naturally where pressures have eased. Limpopo’s estuary, with its sediment accretion and community-based management, has previously been identified as a zone of natural recovery, and the new analysis is consistent with that picture.

Inhambane stands out as the exception. Mangrove cover there declined by more than 2,000 hectares despite years of relative stability, a trajectory the researchers attribute to the late start of local restoration efforts, which only began in 2020, combined with anthropogenic pressure, forest encroachment into mangrove zones, and reduced rainfall. The contrast with the other three sites underscores a central lesson of restoration ecology: timing matters, and interventions launched after degradation has gained momentum face a steeper uphill battle. It also echoes national assessments that have linked urban expansion and economic activity to mangrove decline in some coastal systems even as others expand under favorable geomorphological and hydrological conditions.

Climate emerges as a silent but powerful actor throughout the study period. Although temperatures remained within the normal range for mangrove growth, roughly between 22 and 35 degrees Celsius, rainfall was persistently scarce, and the period coincided with severe and moderate El Niño events in southern Mozambique. Previous research has shown that precipitation and mangrove cover follow a sigmoid relationship, with a reflection point near 1,368 millimeters, and that rainfall below 1,000 millimeters can induce physiological stress. Comparable El Niño episodes have driven catastrophic mangrove dieback in northern Australia and limited restoration success elsewhere in Mozambique, including massive dieback documented at the Maputo River estuary. The modest gains detected in this study are best understood against that climatic backdrop: restoration can work, but drought and hydrological stress can throttle seedling survival and canopy recovery even where human pressures are controlled.

The study is candid about its limits. It rests entirely on remote sensing, without field or socioeconomic data, and the 30-meter resolution of the imagery means mixed pixels can blur the boundary between mangroves and adjacent vegetation, a recognized source of uncertainty in mangrove classification. Tidal variation can also distort spectral indices when images are not acquired under comparable conditions, though multi-temporal image extraction was used to mitigate this. The authors recommend that future assessments incorporate full hydrological and socioeconomic datasets to strengthen causal interpretation.

Even so, the implications for policy are clear. Restoration alone, the findings suggest, is not a guarantee of recovery; it must be paired with systematic monitoring, coordination between government and scientific institutions, and genuine participation by local communities. The mangroves of southern Mozambique occupy narrow intertidal zones where every hectare counts, and the satellite record now shows precisely where recovery is taking root and where it is slipping away. As climate variability intensifies, that kind of evidence will be indispensable for deciding where to plant, where to protect, and where to let the tide do the work.

Subject of Research: Spatiotemporal monitoring of mangrove cover change in restoration areas of southern Mozambique using satellite remote sensing from 2009 to 2025

Article Title: Mangrove spatiotemporal changes in four areas with restoration projects in Southern Mozambique from 2009 to 2025

Article References: Mangrove spatiotemporal changes in four areas with restoration projects in Southern Mozambique from 2009 to 2025. (n.d.). https://doi.org/10.1016/j.envc.2026.101669

Image Credits: AI Generated

DOI: 10.1016/j.envc.2026.101669

Keywords: mangroves, Mozambique, remote sensing, restoration, Random Forest classification, Google Earth Engine, spectral indices, El Niño, coastal ecosystems, land cover change, blue carbon, Mann-Kendall trend test

The research, led by Hiwot Neway of Hawassa University together with colleagues in Ethiopia and Austria, stitched together meteorological station observations and the ERA5 reanalysis dataset to build a continuous hydroclimatic record from 1990 to 2024. Using the modified Mann-Kendall trend test, a non-parametric method robust to outliers and serial correlation, the team found that annual precipitation in both districts showed only weak, statistically insignificant positive tendencies, with Sen’s slope estimates of roughly 1.1 to 1.8 millimeters of additional rain per year. On paper, the water supply is holding steady. But the same analysis revealed something far more consequential: mean annual temperature rose significantly in both districts at nearly identical rates of about 0.017 degrees Celsius per year, and potential evapotranspiration climbed in lockstep at approximately 1.5 millimeters per year.

Potential evapotranspiration, or PET, is the amount of water that would evaporate from the soil and transpire from plants if water were freely available. It is the atmosphere’s demand side of the water ledger. As temperatures rise, PET rises with them, meaning crops and soils lose moisture faster even when every millimeter of rainfall stays the same. This is the thermodynamic mechanism the researchers believe explains the long-standing puzzle of perception-instrument divergence in East Africa, where smallholders routinely report worsening conditions that precipitation-only analyses fail to detect. Where the water balance between rainfall and evaporative demand tightens, effective moisture for crops declines, and the farmers feel it in wilting leaves and delayed planting, even if the annual rainfall statistics look reassuringly flat.

To capture drought in this warming context, the team computed the Standardized Precipitation-Evapotranspiration Index, or SPEI, at three-, six-, and twelve-month accumulation periods, fitting monthly water-balance series to a log-logistic distribution. Unlike rainfall-only indices, SPEI incorporates both supply and demand, making it far more sensitive to temperature-driven drying. The results revealed a landscape of dramatic swings: SPEI values ranged from extreme drought to severe wet conditions across the 35 years, with synchronized severe drought events standing out in 2009 and 2022-2023, the latter coinciding with the devastating El Nino-linked crisis that gripped the Horn of Africa. Yet, strikingly, no long-term drying trend reached statistical significance at any timescale. Drought in these districts, the study concludes, is a story of savage interannual variability layered on top of a slowly tightening water budget, not a steady march toward aridity.

The survey side of the study brought the human dimension into focus. Between September and December 2024, the researchers interviewed 204 households across both districts, selected through multistage stratified random sampling and limited to household heads aged thirty or older with long-term residence. The numbers were unambiguous: 89.7 percent of households reported perceiving shifts in rainfall patterns, 82.9 percent reported rising temperatures, 76.5 percent said droughts had become more frequent, and 72.5 percent observed a shortened rainy season. When asked about specific shocks, flooding topped the list at 67.6 percent, followed by land degradation at 48.5 percent and resource-based conflict at 43.5 percent. Focus group participants described maize yields collapsing to 60 to 70 percent below normal in shock years, with the drought events visible in the SPEI record matching the worst crop failures farmers remembered.

The paradox of flooding being the most-reported shock in a region without increasing rainfall resolves itself when one considers how a warming atmosphere behaves. Warmer air holds more moisture, concentrating precipitation into fewer, heavier events. Combined with severe land degradation, which strips soils of the organic matter and structure needed to absorb water, these intense downpours translate directly into flash flooding and erosion. The farmers’ accounts, in other words, describe precisely what climate physics predicts: not less rain overall, but rain that arrives more violently, soaks in less effectively, and evaporates faster afterward. Annual totals smooth all of this away, which is why the study’s authors argue that monitoring systems calibrated to precipitation deficits alone are structurally blind to the stressors smallholders actually experience.

The analysis also delivered a surprise on the question of who is most exposed. Using a Climate Perception Index built from ten binary indicators of climate-related shocks, and applying Poisson regression with robust standard errors, the researchers found that medium-wealth households reported 47 percent higher cumulative shock exposure than poor households, while district alone explained a 32 percent difference in shock rates. Bilate Zuria households in the lowland zone reported dramatically higher odds of flooding, land degradation, and conflict compared with their midland counterparts in Boricha, though they were far less likely to report erratic rainfall or famine. The authors suggest a provocative explanation for the wealth pattern: medium-wealth households own more assets worth losing, such as livestock and cropland, but lack the savings and off-farm income buffers of richer households, leaving them simultaneously more exposed and less protected. The finding challenges the reflexive assumption that the poorest always bear the greatest climate burden.

Amid these warnings, the study also surfaced a quieter source of resilience. In the midland enset-maize-coffee farming system, enset, a drought-tolerant banana relative often called Ethiopia’s tree against hunger, consistently buffered food security during maize failures. Households in the region harvest enset flexibly, drawing on its carbohydrate-rich corm and pseudostem in bad years, effectively storing food in the ground. Recent research elsewhere in Ethiopia shows that farmers expand enset cultivation after severe droughts and that enset-growing households display greater resilience to climatic shocks. The authors suggest that this indigenous crop functions as a low-cost, locally embedded biological buffer that could complement formal instruments such as weather-index insurance, and they call for formal cost-effectiveness comparisons between the two strategies across wealth groups.

The broader implication reaches well beyond two Ethiopian districts. If rising evaporative demand can erode agricultural water security while rainfall statistics remain flat, then drought early-warning systems, crop insurance schemes, and climate adaptation plans built on precipitation thresholds are measuring the wrong thing in warming regions. The study’s authors recommend sub-seasonal, growth-stage-specific drought indices to capture the dry-spell timing that farmers actually experience, panel datasets to disentangle wealth and exposure effects, and high-resolution water-balance mapping to locate hidden hotspots of evaporative stress. As global temperatures continue to climb, the gap between what the gauges record and what the fields endure will only widen, and this study makes a compelling case that closing it requires listening to both the instruments and the people who live under the same warming sky.

Subject of Research: Temperature-driven increases in potential evapotranspiration and farmer perceptions of drought in southern Ethiopia

Article Title: Increasing potential evapotranspiration despite non-significant rainfall trends: a 35-year hydroclimatic analysis and assessment of farmer perceptions in Southern Ethiopia

Article References: Neway, H., Mekuyie, M., Melka, Y., Abrha, H., & Abdi, A. T. (2026). Increasing potential evapotranspiration despite non-significant rainfall trends: a 35-year hydroclimatic analysis and assessment of farmer perceptions in Southern Ethiopia. Theoretical and Applied Climatology, 157(10), Article 677. https://doi.org/10.1007/s00704-026-06553-9

Image Credits: AI Generated

DOI: 10.1007/s00704-026-06553-9

Keywords: potential evapotranspiration, drought, Ethiopia, climate change, smallholder agriculture, SPEI, rainfall variability, water balance, farmer perceptions, Sidama, food security, enset