Acta Geophysica最新文献

{"title":"Spatiotemporal drought dynamics in the Southeastern Anatolia, Türkiye: a CEEMDAN-LSTM hybrid deep learning approach with SHAP-based explainability and uncertainty quantification","authors":"Ramazan Acar,&nbsp;Ergun Akbas,&nbsp;Musa Esit","doi":"10.1007/s11600-026-02030-z","DOIUrl":"10.1007/s11600-026-02030-z","url":null,"abstract":"<div><p>Drought has become one of the most critical hydroclimatic hazards under climate change with increasing temperature, enhanced atmospheric evaporative demand and increasing hydroclimatic variability. This research explores the spatiotemporal behavior and predictability of meteorological drought in the Southeastern Anatolia Region by means of the Standardized Precipitation Index (SPI) and Standardized Precipitation Evapotranspiration Index (SPEI) at 3-, 6-, and 12-month timescales for the period 1965–2025. The variability of drought, long-term trends, spatial patterns and forecasting performance were analyzed using the monthly precipitation and temperature data of 15 meteorological stations. A hybrid Complete Ensemble Empirical Mode Decomposition with Adaptive Noise–Long Short-Term Memory (CEEMDAN-LSTM) framework was applied and compared with Persistence, Random Forest, XGBoost, and standalone LSTM models. Model interpretability was evaluated with SHapley Additive exPlanations (SHAP), and forecast uncertainty was measured in terms of quantile regression-based prediction intervals. The results reveal a significant regional drying trend after 2000, with the regional mean SPEI-12 shifting from near-normal conditions to persistent negative values. The drought episode of 2020–2025 is the most extreme and longest drought episode in the 60-year observational record. Compared with the SPI, the SPEI exhibited stronger and more spatially coherent drying trends, which revealed the amplification role of temperature-driven evapotranspiration under warming conditions. CEEMDAN decomposition successfully identified dominant interannual and multi-decadal oscillatory modes, while residual components showed a persistent long-term drying trend. The CEEMDAN-LSTM model displayed the best predictive performance for most index-timescale combinations across all forecasting approaches, with <i>R</i>² values of 0.990 for SPEI-6 and 0.981 for SPI-12. SHAP analysis revealed that seasonal evapotranspiration variability strongly controls SPEI predictability, while SPI forecasts are mainly driven by lagged precipitation persistence. Uncertainty analysis further revealed decreased prediction interval reliability during the unprecedented 2022–2025 drought period, highlighting the effects of climatic no stationarity and the limitations of historical training data under intensifying drought conditions. The results, generally, demonstrate the high potential of hybrid deep learning frameworks for drought forecasting and offer valuable scientific support for climate adaptation and sustainable water resource management in the Euphrates–Tigris Basin.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 6","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149189496","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Surface thermal dynamics and air quality stressors in a rapidly urbanising semi-arid city: a multi-sensor satellite assessment of Greater Irbid, Jordan (2000–2025)","authors":"Bushra Obeidat,&nbsp;Mohammed J. Alshayeb,&nbsp;Javed Mallick","doi":"10.1007/s11600-026-02054-5","DOIUrl":"10.1007/s11600-026-02054-5","url":null,"abstract":"<div><h3>Background</h3><p>Semi-arid cities of the Middle East face compounding pressures from rapid urban growth, rising temperatures and persistent air pollution, yet long-term evidence for secondary cities remains scarce.</p><h3>Methods</h3><p>This study examined twenty-five years of daytime surface thermal change and the present-day air quality regime of Greater Irbid, Jordan, using an integrated multi-sensor satellite framework. Land surface temperature (LST) was retrieved from Landsat 5, 8 and 9 Collection 2 Level-2 archives for six epochs between 2000 and 2025, and daytime surface thermal island intensity was computed against an elevation-controlled rural reference derived from epoch-matched Global Human Settlement Layer (GHSL) built-up surfaces. Five air quality stressors, namely nitrogen dioxide (NO2), sulphur dioxide (SO2), carbon monoxide (CO), formaldehyde (HCHO) and the ultraviolet aerosol index (AI), together with two contextual variables, ozone (O3) and methane (CH4), were characterised from Sentinel-5P TROPOspheric Monitoring Instrument (TROPOMI) records for 2019–2025.</p><h3>Results</h3><p>The city exhibited a daytime surface urban cool island typical of drylands that weakened from − 6.26 °C to − 4.14 °C between the study endpoints, though the contrast was not monotonic across the intervening epochs. This weakening was strongly associated with the 2.2-fold growth of the dense urban core (<i>r</i> = 0.977, 95 per cent confidence interval 0.80–1.00) and is consistent with densification-driven relative warming. NO2 peaked over the urban core with a winter to summer ratio of 1.93, HCHO and aerosols peaked in summer and CH4 tracked the global background rise, confirming retrieval consistency.</p><h3>Conclusion</h3><p>The results identify an east–south belt where elevated surface heat visually corresponds with pollution enhancements and provide a reproducible evidence base for cooling, greening and emission management in dryland municipalities.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 6","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149189579","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Research on transformer-Kolmogorov–Arnold neural network open-pit mine slope deformation prediction model based on temporal interferometric synthetic aperture radar","authors":"Yufan Li,&nbsp;Chengyu Jiang,&nbsp;Chen Wang,&nbsp;Ting Lu,&nbsp;Shan Jiang,&nbsp;Shunyu Yang,&nbsp;Na Zhao","doi":"10.1007/s11600-026-02057-2","DOIUrl":"10.1007/s11600-026-02057-2","url":null,"abstract":"<div><p>Slope stability is critical to ensuring safe mining operations; yet, traditional monitoring techniques often suffer from insufficient spatial resolution, limited coverage, and inadequate predictive accuracy. To address these limitations, this study proposes a collaborative monitoring and prediction framework that integrates temporal interferometric synthetic aperture radar (InSAR) with deep learning, using an open-pit mine in Guizhou as a case study. Small baseline subset InSAR (SBAS-InSAR) results reveal maximum subsidence rates of − 40 mm·yr⁻¹ in the phosphate quarry and − 70 mm·yr⁻¹ in the waste dump, with cumulative settlements of 106 and 236 mm, respectively. Comparison with ground-based leveling measurements from 22 monitoring benchmarks across two field campaigns yielded a correlation coefficient <i>R</i> of 0.92 between the SBAS-InSAR and leveling datasets. Based on the SBAS-InSAR-derived deformation time series, four predictive models—Long Short-Term Memory (LSTM), Transformer, LSTM–Kolmogorov–Arnold Neural Network (KAN), and Transformer–KAN—were constructed and evaluated. Among them, the Transformer–KAN model exhibited the best performance. On average across the six evaluated feature points, the Transformer–KAN model achieved a mean absolute error (MAE) of 6.12 (± 0.42) mm and a root mean square error (RMSE) of 7.98 (± 0.37) mm. At the best-performing point (Point F), the MAE and RMSE were 5.26 (± 0.27) mm and 6.81 (± 0.33) mm, respectively, representing a 42% improvement over the LSTM model at that location. The KAN module was used as a predictive component in the present study; its interpretability was not quantitatively evaluated. Further validation across additional sites is needed.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 6","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176718","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Oil-bearing potential identification of low-resistivity layers during drilling: a small-sample smart method based on gas-logging and LWD gamma ray","authors":"Jialing Zou,&nbsp;Jiali Wang,&nbsp;He Zhang,&nbsp;Yi Yang,&nbsp;Yitao Hu","doi":"10.1007/s11600-026-02041-w","DOIUrl":"10.1007/s11600-026-02041-w","url":null,"abstract":"<div><p>Low-resistivity oil layers are difficult to identify during drilling because their resistivity responses may overlap substantially with those of water-bearing intervals, while reliable calibration samples obtained from post-drilling field verification are often limited. To address these challenges, this study proposes a small-sample intelligent method for evaluating the oil-bearing potential of low-resistivity intervals using gas-logging parameters and LWD natural gamma-ray (GR) data. The model performs binary classification between oil-bearing and water-bearing intervals, with normal oil-layer and low-resistivity oil-layer samples combined into the oil-bearing class; the original low-resistivity oil-layer labels are retained for subgroup-specific evaluation. Within the training dataset, SMOTE-based augmentation was applied to the original geological subclasses to improve sample representation, and Kernel Principal Component Analysis (KPCA) was employed for nonlinear feature extraction and redundancy reduction. A KPCA-PSO-CatBoost model was subsequently established, with particle swarm optimization (PSO) used for hyperparameter optimization. On an independent test set comprising 25 intervals from two previously unused wells, the optimized model achieved an accuracy of 92.0% and a Macro-F1 score of 0.890. Ablation experiments indicated that KPCA provided the more substantial performance improvement, while PSO contributed an additional complementary gain through hyperparameter optimization. The proposed method shows potential as a data-driven auxiliary workflow for rapid evaluation of oil-bearing potential during drilling, particularly for low-resistivity intervals under conditions of limited calibration samples and ambiguous electrical responses.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 6","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176645","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Global assessment of topside effective scale heights in the low- and mid-latitude ionosphere using COSMIC-2 observations and IRI-2020 topside COR2 model","authors":"S. Iswariya,&nbsp;Alessio Pignalberi,&nbsp;Sampad Kumar Panda","doi":"10.1007/s11600-026-02058-1","DOIUrl":"10.1007/s11600-026-02058-1","url":null,"abstract":"<div><p>The effective topside scale height (<i>H</i>_eff) governs the vertical distribution of electron density in the topside ionosphere, and its accurate representation is crucial for ionospheric modeling and space weather applications. This study examines the global distribution of <i>H</i>_eff spanning the low- and mid-latitude regions and its diurnal, seasonal, longitudinal variability, as well as statistical characteristics, using Constellation Observing System for Meteorology, Ionosphere, and Climate 2 (COSMIC-2) radio occultation observations during the solar cycle 25 (2019–2024). The results are compared with the corresponding predictions from the International Reference Ionosphere (IRI)—2020 topside sub-model IRI-COR2, and assimilated IRI-COR2 obtained by ingesting COSMIC-2-derived <i>fo</i>F2 and <i>hm</i>F2 parameters into IRI-COR2. It is observed that <i>H</i>_eff values during high solar activity (HSA) consistently exceeded the corresponding values during low solar activity (LSA), across all datasets, seasons, and longitude sectors, confirming the strong solar cycle modulation of <i>H</i>_eff driven by enhanced solar EUV flux. An equatorial pre-sunrise enhancement of <i>H</i>_eff, centered near the magnetic equator between 3 and 6 local time, was identified across all datasets under both solar activity conditions, which could be associated with reduced recombination rates and plasma redistribution along magnetic field lines. The NDJ (November, December, January) consistently yielded the highest equatorial <i>H</i>_eff, whereas the FMA (February, March, April) showed the most suppressed values, consistent with inter-hemispheric thermospheric neutral wind asymmetry during the December solstice. IRI-COR2 systematically overestimates <i>H</i>_eff relative to COSMIC-2 with RMSEs of 31.81 km and 45.34 km during LSA and HSA, respectively, which is most pronounced near the magnetic equator. Assimilation of COSMIC-2 <i>fo</i>F2 and <i>hm</i>F2 marginally reduced the RMSE to 31.35 km and 44.26 km, respectively, although systematic biases persisted, highlighting the need for full-profile assimilation to further improve topside ionospheric representation.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 6","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Residual moveout correction based on the improved dynamic time warping method for wide-azimuth seismic exploration","authors":"Liyan Zhang,&nbsp;Shouying Du,&nbsp;Xinyu Hu,&nbsp;Ang Li","doi":"10.1007/s11600-026-02032-x","DOIUrl":"10.1007/s11600-026-02032-x","url":null,"abstract":"<div><p>Conventional processing (e.g., velocity analysis, normal moveout (NMO) correction, static correction) often fails to fully flatten the seismic events due to subsurface anisotropy and heterogeneity. Consequently, residual moveout persists in imaging gathers, which is particularly pronounced in wide-azimuth data. Residual moveout compromises both seismic imaging accuracy and prestack inversion requirements. To address this, we propose a gather correction method using cross-correlations dynamic time warping (CDTW), which constructs an error matrix from cross-correlation in time and offset. This significantly improves trace matching accuracy and stability. The method overcomes traditional dynamic time warping, such as low precision and instability, while effectively preserving gather waveform characteristics. Theoretical model tests and field confirm the method’s robustness and practical value. After correction, seismic events show improved continuity and resolution. The processed data exhibit a 4-Hz bandwidth broadening over the original. These results provide high-quality input for subsequent seismic interpretation and prestack inversion.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 6","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149176367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Process-based assessment of conditioning factors, triggers, and flood dynamics of the October 2023 South Lhonak glacial lake outburst flood (GLOF)","authors":"Vishal Singh,&nbsp;Atar Singh,&nbsp;Akshaya Verma,&nbsp;Ashutosh Sharma,&nbsp;Bhupendra Joshi,&nbsp;Kuldeep Sharma,&nbsp;Surjeet Singh","doi":"10.1007/s11600-026-02042-9","DOIUrl":"10.1007/s11600-026-02042-9","url":null,"abstract":"<div><p>On October 4, 2023, South Lhonak Lake (SLL) in the Sikkim Himalaya catastrophically breached, generating flash floods in the Teesta Valley that caused ~70 fatalities and severe infrastructure damage, including destruction of the 1200 MW Teesta-III hydropower project. While previous studies emphasize breach hydraulics or downstream impacts, process-based understanding of preconditioning and compound triggers remains limited. This study develops an integrated GLOF framework to identify multiple GLOF triggers by combining geodetic mass balance of the parent glacier, glacier-melt sensitivity analysis, identifying meteorological changes, and performing 2D hydrodynamic GLOF simulation. The analysis shows a 127% expansion in the SLL area from 2000 to 2023 and volume with significant mass loss of the parent glacier. Additionally, the high-resolution meteorological data indicate that intense precipitation (during October 1–3, 2023) coincided with elevated temperature anomalies, amplifying meltwater input to SLL and destabilized the loose moraine and glacier body. Due to this, destabilization of the moraine (estimated at ~3.2 × 10⁷ m³) together with ice-calving-induced impulse waves likely contributed to a ~91 m breach, resulting in a peak discharge of ~18,000 m³/s within ~30 min at just near to lake outlet. Simulations reconstruct a maximum flood depth of 23.95 m (cross section at Chungthang) and a maximum velocity of 26.51 m/s (cross section near the lake outlet) along the ~68-km simulated profile, with markedly different values at these two specific cross sections (13.79 m depth and 26.51 m/s at the lake outlet and 23.95 m depth and 4.53 m/s at Chungthang), emphasizing evolving glacier lake feedbacks as critical GLOF precursors.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Process-based assessment of conditioning factors, triggers, and flood dynamics of the October 2023 South Lhonak glacial lake outburst flood (GLOF)","authors":"Vishal Singh,&nbsp;Atar Singh,&nbsp;Akshaya Verma,&nbsp;Ashutosh Sharma,&nbsp;Bhupendra Joshi,&nbsp;Kuldeep Sharma,&nbsp;Surjeet Singh","doi":"10.1007/s11600-026-02042-9","DOIUrl":"10.1007/s11600-026-02042-9","url":null,"abstract":"<div><p>On October 4, 2023, South Lhonak Lake (SLL) in the Sikkim Himalaya catastrophically breached, generating flash floods in the Teesta Valley that caused ~70 fatalities and severe infrastructure damage, including destruction of the 1200 MW Teesta-III hydropower project. While previous studies emphasize breach hydraulics or downstream impacts, process-based understanding of preconditioning and compound triggers remains limited. This study develops an integrated GLOF framework to identify multiple GLOF triggers by combining geodetic mass balance of the parent glacier, glacier-melt sensitivity analysis, identifying meteorological changes, and performing 2D hydrodynamic GLOF simulation. The analysis shows a 127% expansion in the SLL area from 2000 to 2023 and volume with significant mass loss of the parent glacier. Additionally, the high-resolution meteorological data indicate that intense precipitation (during October 1–3, 2023) coincided with elevated temperature anomalies, amplifying meltwater input to SLL and destabilized the loose moraine and glacier body. Due to this, destabilization of the moraine (estimated at ~3.2 × 10⁷ m³) together with ice-calving-induced impulse waves likely contributed to a ~91 m breach, resulting in a peak discharge of ~18,000 m³/s within ~30 min at just near to lake outlet. Simulations reconstruct a maximum flood depth of 23.95 m (cross section at Chungthang) and a maximum velocity of 26.51 m/s (cross section near the lake outlet) along the ~68-km simulated profile, with markedly different values at these two specific cross sections (13.79 m depth and 26.51 m/s at the lake outlet and 23.95 m depth and 4.53 m/s at Chungthang), emphasizing evolving glacier lake feedbacks as critical GLOF precursors.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Performance of artificial intelligence algorithms for reference evapotranspiration forecasting across different climatic zones","authors":"Fatemeh Barzegari Banadkooki,&nbsp;Mahboube Ebrahimi,&nbsp;Ali Torabi Haghighi,&nbsp;Yong Xiao","doi":"10.1007/s11600-026-02028-7","DOIUrl":"10.1007/s11600-026-02028-7","url":null,"abstract":"<div><p>Accurate forecasting of reference evapotranspiration (ET₀) is important for irrigation planning and water resources management, particularly in regions characterized by contrasting climatic conditions. Although various artificial intelligence models have been applied to ET₀ prediction, systematic comparisons of advanced hybrid architectures across distinct climatic settings remain limited. This study evaluated one-month-ahead forecasting of monthly ET₀ at three synoptic stations representing extremely arid, semi-arid, and arid coastal conditions in Iran. Seven forecasting approaches were compared: Autoregressive Integrated Moving Average-Long Short-Term Memory (ARIMA-LSTM), Support Vector Machine implemented for regression (SVM), Support Vector Machine optimized using a Genetic Algorithm (SVM-GA), Support Vector Machine-Random Forest (SVM-RF), Transformer, Temporal Convolutional Network (TCN), and a hybrid TCN-Transformer model. Model performance was evaluated using the coefficient of determination (R²) and root-mean-square error (RMSE). The TCN-Transformer model consistently produced the most accurate forecasts across the three investigated stations, with R² values ranging from 0.95 to 0.99 and dimensionless normalized RMSE values ranging from 0.06 to 0.11. The conventional SVM-based models generally produced lower forecasting accuracy, particularly at the arid coastal station. These results indicate that the ET₀ series at the arid coastal station was more difficult for the evaluated models to forecast. Overall, the findings demonstrate the potential of combining convolution-based feature extraction with attention-based modeling for accurate one-month-ahead monthly ET₀ forecasting under the climatic conditions investigated in this study.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Performance of artificial intelligence algorithms for reference evapotranspiration forecasting across different climatic zones","authors":"Fatemeh Barzegari Banadkooki,&nbsp;Mahboube Ebrahimi,&nbsp;Ali Torabi Haghighi,&nbsp;Yong Xiao","doi":"10.1007/s11600-026-02028-7","DOIUrl":"10.1007/s11600-026-02028-7","url":null,"abstract":"<div><p>Accurate forecasting of reference evapotranspiration (ET₀) is important for irrigation planning and water resources management, particularly in regions characterized by contrasting climatic conditions. Although various artificial intelligence models have been applied to ET₀ prediction, systematic comparisons of advanced hybrid architectures across distinct climatic settings remain limited. This study evaluated one-month-ahead forecasting of monthly ET₀ at three synoptic stations representing extremely arid, semi-arid, and arid coastal conditions in Iran. Seven forecasting approaches were compared: Autoregressive Integrated Moving Average-Long Short-Term Memory (ARIMA-LSTM), Support Vector Machine implemented for regression (SVM), Support Vector Machine optimized using a Genetic Algorithm (SVM-GA), Support Vector Machine-Random Forest (SVM-RF), Transformer, Temporal Convolutional Network (TCN), and a hybrid TCN-Transformer model. Model performance was evaluated using the coefficient of determination (R²) and root-mean-square error (RMSE). The TCN-Transformer model consistently produced the most accurate forecasts across the three investigated stations, with R² values ranging from 0.95 to 0.99 and dimensionless normalized RMSE values ranging from 0.06 to 0.11. The conventional SVM-based models generally produced lower forecasting accuracy, particularly at the arid coastal station. These results indicate that the ET₀ series at the arid coastal station was more difficult for the evaluated models to forecast. Overall, the findings demonstrate the potential of combining convolution-based feature extraction with attention-based modeling for accurate one-month-ahead monthly ET₀ forecasting under the climatic conditions investigated in this study.</p></div>","PeriodicalId":6988,"journal":{"name":"Acta Geophysica","volume":"74 5","pages":""},"PeriodicalIF":2.1,"publicationDate":"2026-10-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149132989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}