Accurate estimation of evapotranspiration is crucial for enhancing real-time irrigation scheduling and decision making in water resource planning. Traditionally, empirical methods are used to calculate reference evapotranspiration (ET0) using available meteorological data. However, in many areas, such data are limited or unavailable for ET0 estimation. Hence, this study aims to explore data-driven models such as machine learning (ML) and deep learning (DL) for estimating ET0 with minimal meteorological data. In this study, five ML models, including linear regression (LR), random forest (RF), support vector regression (SVR), XGBoost, KNN regression, and two deep learning methods, such as feedforward neural networks and long-term short-term memory (LSTM), were used to estimate the reference evapotranspiration (ET0) over the Phulnakhara canal command area, Odisha, India using various combinations of meteorological variables. The results of these models were compared with the Penman‒Monteith-based ET0. The Penman-Monteith based ET0 is significantly (p < 0.01) positively correlated with sunshine hour and maximum temperature, with correlation coefficients of 0.8 and 0.6, respectively, whereas maximum relative humidity and minimum humidity are negatively correlated. The findings revealed that when all climate data (maximum temperature (Tmax), minimum temperature (Tmin), maximum relative humidity (RHmax), minimum relative humidity (RHmin), wind speed and sunshine hour) are available, the coefficient of determination (R2) increases to 0.98. However, when data are limited, it decreases to 0.78. The SVR model outperformed the other ML models with all the input combinations. However, KNN emerged as the most reliable model for estimating ET0 with input data of maximum and minimum temperature. The results revealed that using only three variables (temperature, wind speed, and relative humidity) or even two-parameter combinations (temperature with either relative humidity or wind speed) yielded R² values in the range of 0.78–0.79. These findings indicate that ML and DL can effectively estimate ET0 even under sparse meteorological conditions within the canal command area.The findings of this study offer valuable insights for estimating ET0 in regions with limited climate data, which is crucial for effective agricultural water management.
Evapotranspiration (ETo) estimation plays a crucial role in management of water resources, agricultural planning and environmental monitoring besides providing essential insights into water usage and irrigation scheduling. The FAO-56 Penman–Monteith (PM) method offers accurate ETo estimates based on meteorological data. However, this method often requires precise and comprehensive data, which may not always be readily available. Moreover, in recent years, artificial intelligence (AI) techniques have gained prominence as innovative solutions for ETo estimation, potentially addressing some of the limitations associated with traditional methods. The scope of this review is to evaluate and synthesize recent advancements in AI-based approaches for ET₀ estimation, emphasizing model performance, key input variables, advantages, limitations, and their comparison with traditional methods. To achieve this, a systematic review was conducted in accordance with the PRISMA 2020 guidelines. This review explores the application of various AI techniques viz. machine learning algorithms (e.g., Support Vector Machines, Neural Networks, and Random Forests), ensemble models, and optimization methods for ETo estimation during the period 2012–2024. By evaluating these AI-driven approaches and comparing them with the standard FAO-56 PM method, the review aimed to assess their effectiveness in improving estimation accuracy, computational efficiency and adaptability to different conditions. This study systematically reviewed applicability of AI techniques to ensure accuracy of ETo estimation, focusing on their ability to leverage diverse data sources, handle missing or incomplete information and adapt to varying climatic conditions. Moreover, the potential of hybrid models that combine AI with traditional methods to enhance predictive capabilities is also being discussed. By synthesizing current research and identifying gaps in the literature, this study provides insights into future directions for ETo estimation using AI techniques leading to judicious irrigation scheduling for improving water productivity and ensuring sustainable agricultural practices.
Tropospheric ozone (O3), an emerging climate change-induced stressor, enters leaf tissues via stomata, triggering reactive oxygen species (ROS) generation and suppressing key N assimilation enzymes like nitrate reductase and glutamine synthetase impairing nitrogen metabolism and reducing grain yield and nitrogen use efficiency (NUE) in rice. This study evaluated the impact of elevated O3 (e[O3]) on nitrogen (N) uptake, NUE components, and yield attributes across two contrasting seasons using Open Top Chambers (OTCs) with four treatments: UC (ambient, open field, 30 +/- 5 ppb), CC (ambient, OTC, 30 +/- 5 ppb), EO40 (40 +/- 5 ppb), and EO60 (60 +/- 5 ppb), across three N levels. Partial Least Squares Path Modeling (PLS-PM) was used to analyze trait interrelationships affecting grain yield. Results showed that e[O3] significantly reduced total N uptake (17-28%), agronomic NUE (27-35%), N recovery efficiency (22-28%), physiological NUE (6-10%), and partial factor productivity of N (14-23%) compared to CC. Grain yield declined by 14-23%, with greater reductions observed during Kharif season, likely due to higher stomatal conductance facilitating increased O3 uptake. Higher N application only partially mitigated O3-induced NUE impairment. PLS-PM identified spikelet fertility as the strongest direct yield determinant under O3 stress.
The United Nations Sustainable Development Goal-15.3 (SDG-15.3) has conveyed a message of emergent connotation of restoration strategy and policy for degraded land ecosystems globally, and advocated for innovative and collaborative basic and applied research programs internationally. Trenching is an economically viable and practicable option for the conservation of green-water and sustainable yield of production systems in the degraded land ecosystem. An optimum trenching density that has maximum net-benefit per unit cost of production system is necessary for management and restoration of degraded/bad land ecosystems. Developed a concept for estimation of optimum trenching density by maximizing the net-benefit per unit cost of the production system in land ecosystem, and demonstrate field applicability of the proposed concept for determining the optimum density of trenching in degraded ravine ecosystem. Evolved optimum trenching density concept consists of development of models for net-benefit per unit cost of the production system. The two-steps curve-fitting approach was used for the development of models. An optimization model was formulated for optimum trenching density based on maximum net-benefit per unit cost of the production system. The proposed concept was field-tested for the staggered contour trenching (SCT) in an experimental horti-silvi-pastoral system in the degraded ravine ecosystem. The developed optimization model for the SCT was solved employing the AMPL optimization software. The optimum density for the SCT in the horti-silvi-pastoral production system in the degraded ravine ecosystem was worked out to be 357 trenches ha-1 for getting maximum net-benefit per unit cost of the horti-silvi-pastoral production system. The optimized density of the SCT (357 trenches ha-1) can successfully be implemented in the horti-silvi-pastoral or other similar production systems in degraded land ecosystems anywhere in India and world. Derived concept could be used by the field functionaries and watershed managers, researchers, academicians, and policy-and decision-makers for developing suitable micro-level management plan in watershed development and/or crop production improvement programs, and evaluating the effectiveness of green-water management and conservation practices in a specific production system in a given degraded land ecosystem that gives maximum net-benefit per unit cost of the production system.
A field experiment was conducted during the kharif season of 2022 at Acharya Narendra Deva University of Agriculture & Technology, Ayodhya, eastern Uttar Pradesh, India, to evaluate the effects of sowing date and cultivar on phenology, growth, biomass production and yield of mungbean (Vigna radiata (L.) Wilczek). Three sowing dates (5, 15 and 25 July) were evaluated as main-plot treatments and three cultivars (Pusa Baisakhi, Narendra Moong-1 and SML-668) as subplots in a split-plot design with three replications. Delaying sowing progressively shortened crop duration: maturity occurred at 80, 75 and 70 days in Pusa Baisakhi and at 77, 72 and 67 days in SML-668 for the three sowing dates, respectively. Mean grain yield declined from 1260 kg/ha with 5 July sowing to 1180 and 1110 kg/ha with 15 and 25 July sowing, respectively. Pusa Baisakhi produced the highest grain yield (1330 kg ha−1), dry matter accumulation (384.3 g m−2) and biological yield (3840 kg ha−1), whereas SML-668 produced the lowest values for these traits but reached maturity earliest. Harvest index remained comparatively stable across sowing dates (32.0–33.6%), although Pusa Baisakhi recorded the highest cultivar mean (34.7%). The results indicate that, under the conditions of the present one-season field experiment, early July sowing and selection of a higher-biomass cultivar such as Pusa Baisakhi improved mungbean productivity. These findings provide a locally relevant basis for optimising kharif mungbean establishment in eastern Uttar Pradesh.