Journal of Agrometeorology

Field Spectroscopy-Assisted Machine Learning Framework for Assessing Agroclimatic Effects on Paddy

2026-05-14T06:23:17+00:00

Assessment of yield-limiting factors in paddy production should be carried out properly in order to improve productivity under varying agro-climatic conditions. This research paper introduces a combined geospatial and machine learning system to estimate stage-based paddy yield stress during Kuruvai 2023 in the Thanjavur region using vegetation indices and agrometeorological variables. Five different satellite-derived vegetation indices were calibrated using spectroradiometer data during the vegetative, reproductive, and ripening stages. The field- and satellite-derived indices showed good agreement in calibration (R²=0.85-0.92). Correlation analysis showed that there were significant associations between rainfall_vegetation, humidity_reproductive, and yield variability, whereas the canopy vigor represented by ripening_EVI, SAVI correlated positively with grain filling and final productivity. Yield prediction before and after calibration was performed using machine learning models such as Multiple Linear Regression (MLR), K-Nearest Neighbors (KNN), Random Forest (RF), and eXtreme Gradient Boosting (XGBoost). Among these, XGBoost resulted in high performance after calibration (R²=0.908, RMSE=179.728 kg/ha, MAE=143.487 kg/ha). Feature importance revealed that reproductive-stage humidity, rainfall, and ripening-stage soil moisture were the most important yield drivers. This model can be used to provide a powerful decision-support tool for spatial stress mapping, accurate irrigation regulation, and climate-resilient crop strategies in irrigated rice environments.

A Modified Genetic Algorithm-Based Deep Learning Approach for Rainfall Category Prediction

2025-09-30T17:44:57+00:00

The Rainfall prediction is helpful in different fields like water management, energy supply, agriculture and others. The primary objective of rainfall prediction model is to accurately predict daily rainfall and assist in making significant decisions to tackle several challenges. In this paper, deep learning methodology has been coupled with a modified genetic algorithm to develop a Deep Learning Modified Genetic Algorithm (DLMGA) model. The goal of the model is to predict daily rainfall based on 28 years (1992 to 2020) of historical rainfall data. The performance of the DLMGA model is compared with that of the SDL (Simple Deep Learning) and DLGA (Deep Learning with Genetic Algorithm) models. An overall performance comparison of the SDL, DLGA and DLMGA models in terms of accuracy, precision, recall and f1-score reveals that the DLMGA model performs remarkably well in terms of prediction and has an impressive level of accuracy.

Influence of El Niño-Southern Oscillation on Rainfall Variability and Cereal Crop Yields in Northwestern Nigeria

2026-04-27T10:01:46+00:00

This study assessed the influence of the El Niño-Southern Oscillation (ENSO) on rainfall variability and cereal crop yields using monthly rainfall data (1983–2024) sourced from the NASA Prediction of Worldwide Energy Resources; annual rice, millet and sorghum yield data (2000–2024) from the National Agricultural Extension and Research Liaison Services (NAERLS); and ENSO indices (1983–2024) obtained from NOAA. Rainfall anomalies were computed to characterise interannual variability, while linear regression and one-way analysis of variance (ANOVA) with eta-squared (η²) effect size were used to evaluate crop yield trends and ENSO effects. Annual rainfall averaged 862.26 mm (CV = 37.14%) and varied significantly among ENSO phases (F = 5.32, p = 0.0052), with pronounced wetter conditions during some La Niña years (2021 and 2022). Rice (p = 0.013) and millet (p < 0.001) yields exhibited significant increasing trends, whereas sorghum (p = 0.002) showed a slight significant decline. However, ENSO had no significant influence on all the yields. These findings indicate that ENSO is a significant driver of rainfall variability but a weak predictor of annual cereal yields, highlighting the need to integrate ENSO-based seasonal climate forecasts with climate-smart agricultural practices to strengthen agrometeorological advisory services.

Multi Index Climate Drought Assessment (MICDA) Using Hybrid ML Weightings – A Spatio-temporal analysis for the Semi-Arid Watershed in Eastern Ghats, India (2000–2024)

2026-06-12T04:22:20+00:00

Drought is a severe climatic hazard that affects the ecology and environment. It occurs when there is a prolonged period of low precipitation, increased evapotranspiration demand, poor vegetation health, low soil moisture, and limited surface water availability. Similar observations were noticed in the watersheds of Eastern Ghats, India. A comprehensive Combined Drought Index (CDI) for the ARG watershed (1,186.23 km²) using Principal Component Analysis (PCA), ML models, and multi-decadal remote-sensing data (2000-2024) was carried out. Six drought indicators, SPI3, ETA, VHI, SMAI, MNDWI, and FAPAR were used for CDI generation. CDI weights were calculated using PCA driven Random Forest (RF) and Multiple Linear Regression (MLR) model. RF showed superior performance (R = 0.830; r² = 0.701). Spatiotemporal analysis shows 6 of 25 years with drought conditions, associated with high temperatures (34-36 °C) and unpredictable monsoon precipitation (<1000 mm). The drought analysis in the catchment and command area along with trend analysis shows recovery patterns. The long-term satellite data was used for scalable drought monitoring for Multi Indicator Climate Drought Assessment (MICDA). The PCA–RF-based CDI supports Sustainable Development Goals SDG-6, SDG-13, and SDG-15 by utilizing data to plan for watershed resilience and monitor drought in real time.

Impact of Climate Dynamics and Agricultural Inputs on Agricultural Greenhouse Gas Emissions in India: An ARDL Cointegration Approach

2026-06-19T06:45:27+00:00

In India, where agriculture is both a foundation of the economy and a major source of greenhouse gas (GHG) emissions, rising temperatures due to climate change are accelerating environmental degradation while simultaneously undermining the resilience of the agrarian sector. In this context, the present study examines the long-run relationship between agricultural greenhouse gas emissions, climate variables and agricultural inputs in India. The Autoregressive Distributed Lag (ARDL) model technique is applied to time series data on six key variables from 1990 to 2021. The findings of the ARDL bound F-test confirmed the significant long-run cointegration among variables under consideration. Furthermore, area under paddy crop (0.257) and manure nitrogen content (0.842) exert significant and positive long-run effects on greenhouse gas emissions. Also, maximum temperature (0.341) and rainfall (0.086) positively contribute to the rising of greenhouse gases in the long run. With an adjusted R² value of 0.947, the ARDL model exhibits strong explanatory power and confirms the impact of climate and agricultural drivers on agricultural emissions in India. These findings highlight the need to consider the climate variables in agricultural GHG mitigation efforts and policy strategies.

Monitoring Wheat Vegetation Health and Moisture-Related Spectral Variation Using Landsat 8-Derived Indices in Punjab

2026-05-29T12:08:54+00:00

Efficient monitoring of wheat growth and moisture-related stress is critical for improving irrigation management in Punjab, where groundwater depletion and climate variability threaten crop productivity. This study used Landsat 8 Surface Reflectance data processed in Google Earth Engine to assess vegetation health, phenological dynamics, and moisture-related spectral variation during the rabi seasons of 2022–23 and 2023–24 in Phagwara, Punjab. The Normalized Difference Vegetation Index (NDVI), Land Surface Water Index (LSWI), and False Colour Composite (FCC) were analysed at regional and selected field scales. FCC imagery clearly captured seasonal changes in vegetation cover and crop development. NDVI showed a consistent wheat growth pattern, increasing from sowing and early establishment to peak vegetative–reproductive stages, followed by decline during maturity due to senescence and canopy drying. Field-level NDVI trends corresponded well with observed phenological stages and crop growth indicators, including plant height, leaf area index, and relative water content under contrasting irrigation conditions. NDVI-based classification further distinguished wheat and non-wheat pixels within the selected study area. LSWI provided supplementary information on moisture-related spectral variation associated with rainfall, crop stage, and canopy condition. Overall, the integrated use of NDVI, FCC, and LSWI provides a practical framework for monitoring wheat growth and supporting irrigation decision-making.

Application of MCDM Techniques in Ranking of CMIP6 GCMs for Extreme Temperatures over the Periyar River Basin

2026-04-11T08:40:20+00:00

Climatic extremes like temperature extremes are susceptible to climate change (CLC) warming impacts. Thus, it is essential to study the effect of CLC on extreme temperatures (minimum temperature, i.e., T_min and maximum temperature, i.e., T_max) of the Periyar River Basin (PRB), which has several human-induced constraints. Choosing the best general circulation models (GCMs) from a set through ranking is needed for precise projections of future climatic variables. None of the reviewed studies has executed GCMs ranking for 6th phase of Coupled Model Intercomparison Project (CMIP6) corresponding to extreme temperatures of the PRB. Thus, in the current study, ranking of 27 CMIP6 GCMs is performed for extreme temperatures at every grid point of the PRB by using four performance indicators, entropy method and Compromise Programming (CP) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS). The aggregated ranking of GCMs corresponding to the whole study area for extreme temperatures is also executed by employing the Group Decision-Making Approach (GDMA). Top six ranked GCMs obtained through GDMA for T_max are FGOALS-g3, BCC-CSM2-MR, CanESM5, GFDL-CM4-gr2, GISS-E2-1- G and CNRM-CM6-1 corresponding to TOPSIS and CP techniques while that for T_min are KIOST-ESM, CanESM5, CNRM-CM6-1, INM-CM5-0, INM-CM4-8 and GISS-E2-1-G.

A Methodological Approach for Selecting Climate Datasets for Drought Prediction Using ROC Analysis and Youden’s Index

2025-10-31T04:57:17+00:00

This study presents a methodological approach for evaluating and selecting climate datasets for drought prediction in Guyana, based primarily on the combination of Receiver Operating Characteristic (ROC) analysis and Youden’s Index (J). Traditional hydrological metrics—Nash–Sutcliffe Efficiency (NSE) and Kling–Gupta Efficiency Modified (KGEM) were applied only as comparators to highlight their limitations for event detection. Ten climate datasets available in CARiDRO were assessed at two Standardized Precipitation Index (SPI) time scales—3 months (SPI3) and 12 months (SPI12) using records from two contrasting stations: Georgetown (coastal) and Timehri (inland). ROC–Youden rankings were stable across stations and time scales and revealed clear leaders. At Georgetown, the highest J values were obtained by RCP 4.5 for both SPI3 (J = 0.44) and SPI12 (J = 0.48), followed by aexsk (J = 0.22 and 0.34, respectively). At Timehri, aexsk achieved the highest performance for both SPI3 (J = 0.21) and SPI12 (J = 0.35), while RCP 4.5 showed strong but second-best performance only at SPI12 (J = 0.32). In contrast, NSE and KGEM yielded inconsistent rankings between SPI3 and SPI12, underscoring the limitations of continuous-value error metrics for classifying drought versus non-drought events. By prioritizing classification skill (true/false event discrimination) over aggregate fit, the ROC–Youden framework provides an operationally relevant basis for dataset selection in agricultural drought monitoring. The SPI3-focused findings are directly applicable to short-term crop-water decisions, whereas SPI12 results inform longer-term water allocation.

An Assessment of Climatic Risk and Temporal Trend of Rainfed Crop Potential in the Sub-Himalayan Plains of West Bengal

2026-04-29T05:59:44+00:00

This study assessed water availability, climatic risk and temporal trends for Kharif transplanted rice and rainfed Rabi crops in six blocks of Alipurduar district of Sub-Himalayan West Bengal, viz, Alipurduar I & II combined, Falakata, Madarihat, Kalchini and Kumargram, using IMD gridded rainfall (1921–2020) and temperature (1951–2020) data. Trends in water availability parameters were analyzed using the non-parametric Mann–Kendall test and Sen’s slope estimator, while crop water balance components and length of growing period (LGP) were estimated following the Thornthwaite and Mather (1955) approach. Kharif rainfall exhibited a significant declining trend (47-87 mm/Decade) at three of the five locations. Only 20–24% (20–26 Standard Meteorological Week [SMW]) and 29–33% (28–30 SMW) of seasonal rainfall was stable and effective for transplanted rice, with lower values at Madarihat. Assured LGP exceeded 200 days in three out of four years, indicating low risk for rainfed double cropping, although LGP variability was relatively higher at Kumargram and Madarihat. For rainfed yellow sarson sown at 42 SMW, stored soil moisture met about 50% of crop water requirement, declining by 20–22 mm per fortnight with delayed sowing up to 48 SMW. Overall, water availability conditions for rainfed crops, in terms of LGP, showed improvement (1.7-2.4 days /Decade) across the study locations.

Spatio-Temporal Assessment of Land Use/Land Cover Change, Land Surface Temperature, and Groundwater Fluctuations Using Remote Sensing and GIS: A Case Study of Ghaziabad District

2026-08-03T04:47:47+00:00

Urbanization has been increasing at a rapid rate, which has led to transformation in the land use/land cover (LULC) and thus has a significant impact on groundwater resources and land surface temperature (LST) in rapidly developing cities. The present study aims to identify the spatio-temporal dynamics of LULC, LST, and the variability of groundwater in Ghaziabad District, Uttar Pradesh, India, from 2013 to 2023 using Remote Sensing and GIS techniques. Multi-temporal Landsat data were classified using the Maximum Likelihood Supervised Classification algorithm for the LULC maps that included 5 classes: built-up area, agriculture, vegetation, barren land, and water bodies. The accuracy of the classifications was assessed by the Kappa coefficient, and the Ordinary Kriging (OK) method was applied to spatially interpolate the groundwater depth in ArcGIS 10.8. Between 2013 and 2023, substantial increases in built-up areas and loss of vegetation cover were observed in the findings. Over the same time period, groundwater depth continued to decline in a progressive manner, and high LST areas became larger, especially in dense urban settings. The integrated analysis shows that surface thermal conditions and groundwater stress have intensified due to LULC transformation. The results highlight the importance of sustainable urban planning, groundwater recharge initiatives, and green infrastructure to reduce thermal risks and promote environmental sustainability in rapidly urbanizing areas.

Smart Weather Station for Real-Time Climate Monitoring and Prediction Using IoT and Machine Learning Technologies

2026-06-19T10:05:26+00:00

Accurate hyper-local climate prediction is essential for decision-making in agriculture, urban planning, and environmental management. This study addresses the scarcity of accessible monitoring solutions by developing an autonomous, IoT-enabled smart weather station that integrates high-frequency data acquisition with real-time Deep Learning forecasting. The hardware architecture utilizes an Arduino™ for multi-sensor data collection—covering temperature, humidity, pressure, wind dynamics, and solar radiation—coupled with LoRa™ for robust wireless transmission. A Raspberry® Pi 3 functions as an edge computing platform for localized processing and inference. Predictive performance was optimized through a multivariate Long Short-Term Memory (LSTM) network designed to capture non-linear temporal dependencies. Operational validation confirmed high data integrity, comparable to high-end commercial stations. The LSTM model significantly outperformed traditional architectures, yielding a Coefficient of Determination (R²) of 0.957 and a Mean Squared Error (MSE) of 0.0434. This research provides a reliable, low-latency solution for localized forecasting by successfully synthesizing high-precision models for edge deployment. The scalable architecture represents a significant advancement toward actionable, data-driven decision-making in climate-sensitive sectors, offering a low-uncertainty alternative for sustainable resource management and smart initiatives in resource-constrained environments.

Impact of Urban Heat Islands on Rainfall Patterns in Coastal Cities of South India: A Case Study of Mangaluru

2026-06-02T04:32:21+00:00

Rapid urbanization along the southwestern coast of India has significantly altered local microclimatic conditions, with complex feedbacks between the Urban Heat Island (UHI) effect and monsoon precipitation dynamics. This study investigates the influence of UHI intensification on rainfall distribution and extreme precipitation events in Mangaluru — a fast-growing coastal city in Karnataka, India, situated between the Arabian Sea and the Western Ghats. Multi-decadal Land Use and Land Cover (LULC) data derived from Landsat satellite imagery (1994–2024), India Meteorological Department (IMD) gridded rainfall records (0.25° × 0.25°), and MODIS Land Surface Temperature (LST) datasets were used to quantify spatiotemporal shifts in surface thermal anomalies and correlate them with changes in monsoon precipitation behavior. Urban built-up cover expanded from 5.70% to 27.67% between 1994 and 2024, while mean Land Surface Temperature in the urban core rose by 4.89°C. Surface Urban Heat Island Intensity (SUHII) reached 5.21°C by 2024. Mann-Kendall trend analysis indicates a statistically significant increase in extreme rainfall events (≥150 mm/day), from 3.1 events/year in the 1980s to 7.9 events/year in 2020–2024. Weather Research and Forecasting (WRF-ARW) model simulations suggest an estimated 8–12% increase in Convective Available Potential Energy (CAPE) associated with urbanization-induced surface changes, though large-scale climatic drivers cannot be excluded. The interaction of UHI dynamics with Arabian Sea moisture supply and Western Ghats orography creates a unique precipitation amplification mechanism in this coastal setting, with significant implications for agricultural water availability, monsoon-dependent crop planning, and food security in coastal Karnataka.

Drought Stress Mitigation through Antioxidant Defence Response in Sugar Beet Under Various Foliar Applications

2026-06-10T07:44:56+00:00

The rapid increase in climate change is leading to serious changes in global weather patterns, influencing overall plant responses, where heat waves, cyclones, floods, and flash or long-term droughts become more frequent. Drought stress intensified atmospheric moisture demand which, accelerated by rising temperatures, led to serious soil water depletion. This climate- driven cycle remains a major environmental concern. A two-year study was conducted during “winter-summer” season, evaluating the effect of three levels of irrigation and seven foliar treatments on tropical sugar beet under Odisha condition. A significant impact of water scarcity was observed in the S₂ stressed plants, where MDA increased from 5.05 to 13.8 nmol. g^-1. Confronted with the harmful effect of drought, sugar beet induces its defence mechanisms, when the stress intensity increased from moderate(7days) to severe(14days) for prolonged period. As a result, it induced increase in antioxidant enzyme activities in the leaves like CAT, SOD, APX, and POD were recorded 1.97, 205.29, 2.15, and 1.51 u.g^-1.min^-1respectively, under 14 days of severe stress cycles after 4^th spray. Stress impact was also reflected in phenolics accumulation as well as proline. Foliar application of various treatments showed a significant ability to protect cell metabolism through up-regulating defence pathways against oxidative stress. Among all effective foliar treatments, SA, N-Si, Chitosan and JA exhibited as promising in alleviating prolonged drought stress in sugar beet.

Machine Learning-Based Multi-Decadal Analysis of Paddy Field Dynamics in the Cauvery Delta, India: Implications for Sustainable Agriculture

2026-06-12T07:32:49+00:00

Paddy cultivation forms the backbone of agriculture in the Cauvery Delta of Tamil Nadu, India; however, long-term monitoring of its spatial dynamics remains limited. This study employed multi-temporal Landsat imagery and machine learning techniques to analyse the spatiotemporal dynamics of paddy cultivation in Thanjavur district over a 23-year period (1995–96, 2007–08, and 2018–19), while an additional reference year (2015–16) was used for independent validation. Three machine learning classifiers, namely Random Forest (RF), Gradient Tree Boosting (GTB), and Support Vector Machine (SVM), were evaluated using Recall, Precision, Overall Accuracy (OA), F1 Score, and Kappa Coefficient (KC). The RF classifier achieved the highest classification accuracy during 1995–96 (OA = 0.84, KC = 0.68), 2007–08 (OA = 0.90, KC = 0.81), and 2015–16 (OA = 0.90, KC = 0.80), whereas GTB showed superior performance during 2018–19 (Recall = 0.85, OA = 0.82, F1 Score = 0.83, KC = 0.65). The classification results were validated using independently sourced Ground Control Points (GCPs) for 2015–16 and official Agriculture Department acreage statistics, demonstrating strong agreement with the reference data. The results revealed a 6.72% decline in paddy cultivation area (approximately 228 km²) between 1995–96 and 2018–19, accompanied by an expansion of non-paddy and urban land uses. Spatially, paddy cultivation became increasingly concentrated in the western and central parts of the district along the Cauvery River, primarily due to groundwater depletion, irrigation constraints, rapid urbanisation, and increasing climate variability. The findings demonstrate the effectiveness of ensemble-based machine learning approaches for large-scale agricultural monitoring and provide a transferable framework for assessing long-term land-use changes, thereby supporting food security, climate adaptation, and sustainable water resource management in monsoon-dependent deltaic regions.

Rainfall Analysis for Crop Planning for Paddy Grown in Cauvery Delta Zone of Tamil Nadu

2025-12-16T00:57:24+00:00

The Cauvery Delta Zone (CDZ) of Tamil Nadu is highly dependent on rainfall variability for paddy production. The long-term daily rainfall data (1991–2024) from three representative districts (Thanjavur, Tiruvarur, and Nagapattinam) were analysed to evaluate the rainfall trends, regime shifts, and assured rainfall for crop planning. Non-parametric Mann–Kendall and Sen’s slope tests were used to detect monotonic trends, while Innovative Trend Analysis (ITA) and Pettitt’s change-point test identified non-linear trends and structural shifts. Weekly rainfall probabilities were estimated using the incomplete gamma distribution. Results indicated a statistically significant increase in Northeast Monsoon rainfall (Sen’s slope: +3.1 to +3.7 mm yr⁻¹; p ≤ 0.05) and a weak decline in Southwest Monsoon rainfall (−0.9 to −1.4 mm yr⁻¹). Pettitt’s test detected a significant rainfall regime shift around 2005 (p < 0.05), marking increased interannual variability. District-wise probability analysis showed that Standard Meteorological Weeks (SMW) 42–48 provide assured rainfall ≥75 mm at 75% probability, suitable for Samba paddy transplanting, whereas Navarai season rainfall was unreliable (<10 mm week⁻¹). Regression analysis revealed that ENSO and IOD together explained 82% of interannual rainfall variability (R² = 0.82; p < 0.01). The study provides statistically robust, rainfall-responsive insights for paddy crop planning in the Cauvery Delta Zone.

Temporal Variations of PM₂.₅ and Its Meteorological Determinants in an Urban Background Site of Butuan City, Philippines

2026-03-05T10:19:35+00:00