Ruixin Yang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Ruixin Yang has a total of 73 co-authors. Here are the top co-authors by number of shared publications:
- John J. Qu
Department of Geography and Geoinformation Science,George Mason University,Fairfax,United States, USA
4 shared publications - Chaowei Yang
NSF Spatiotemporal Innovation Center, George Mason University
3 shared publications Donglian Sun
3 shared publications- Liping Di
Center for Spatial Information Science and Systems, George Mason University, Fairfax, VA, USA
3 shared publications - Daniel Tong
Department of Atmospheric Oceanic and Earth Sciences George Mason University Fairfax VA USA
2 shared publications - Donglian Sun
Department of Geography and Geoinformation Science, George Mason University, Fairfax, VA 22030, USA
2 shared publications
Latest Publications
- Open Access
- 18 November 2026
- Published by MDPI AG in Agriengineering
- Vol. 8 (11) , 496
Integrating Remote Sensing, Management Information, and Satellite Embeddings for Machine-Learning Prediction of Field-Level Maize Yield in Kenyan Smallholder Systems
Accurate field-level maize-yield estimation in smallholder systems is challenging because plots are small and management is heterogeneous. We analyzed 6016 Kenyan field-years (2017–2020) using conventional Sentinel-2 predictors, climate, soil, terrain, location, survey-based management, and 64-dimensional satellite embeddings. Ten algorithms were compared using random, leave-one-year-out (LOYO), and spatial-block cross-validation, followed by predictor-group comparisons and SHAP analysis. The all-predictor XGBoost model achieved R2 = 0.395, RMSE = 1321 kg ha−1, and MAE = 1010 kg ha−1 under random cross-validation; R2 declined to 0.172 under LOYO and 0.292 under spatial-block validation. Adding management information to conventional remote sensing, climate, soil, terrain, and location raised random-validation R2 from 0.270 to 0.394, with positive gains across all three validation designs. Replacing 252 conventional remote-sensing predictors with 64 embedding dimensions yielded similar accuracy (random-validation R2 = 0.394 and 0.405, respectively) and substantially reduced input dimensionality. These conclusions were broadly consistent with Extra Trees. Because predictors included full-year observations and measurements recorded at harvest, the models provide retrospective, end-of-season yield estimates rather than in-season forecasts. Results highlight management information and compact satellite representations, alongside limitations in transfer to unseen years and locations.
- 18 November 2026
- Published by MDPI AG in Agriengineering
- Vol. 8 (11) , 496
Open AccessIntegrating Remote Sensing, Management Information, and Satellite Embeddings for Machine-Learning Prediction of Field-Level Maize Yield in Kenyan Smallholder Systems
Accurate field-level maize-yield estimation in smallholder systems is challenging because plots are small and management is heterogeneous. We analyzed 6016 Kenyan field-years (2017–2020) using conventional Sentinel-2 predictors, climate, soil, terrain, location, survey-based management, and 64-dimensional satellite embeddings. Ten algorithms were compared using random, leave-one-year-out (LOYO), and spatial-block cross-validation, followed by predictor-group comparisons and SHAP analysis. The all-predictor XGBoost model achieved R2 = 0.395, RMSE = 1321 kg ha−1, and MAE = 1010 kg ha−1 under random cross-validation; R2 declined to 0.172 under LOYO and 0.292 under spatial-block validation. Adding management information to conventional remote sensing, climate, soil, terrain, and location raised random-validation R2 from 0.270 to 0.394, with positive gains across all three validation designs. Replacing 252 conventional remote-sensing predictors with 64 embedding dimensions yielded similar accuracy (random-validation R2 = 0.394 and 0.405, respectively) and substantially reduced input dimensionality. These conclusions were broadly consistent with Extra Trees. Because predictors included full-year observations and measurements recorded at harvest, the models provide retrospective, end-of-season yield estimates rather than in-season forecasts. Results highlight management information and compact satellite representations, alongside limitations in transfer to unseen years and locations.
- Open Access
- 3 November 2026
- Published by MDPI AG in Preprints.org
Integrating Remote Sensing, Management Information, and Satellite Embeddings for Machine-Learning Prediction of Field-Level Maize Yield in Kenyan Smallholder Systems
Field-level crop yield prediction is an important application of precision agriculture, but it remains challenging in African smallholder systems because plots are small, management is heterogeneous, and satellite pixels can mix multiple land covers. This study evaluated machine-learning approaches for predicting maize yield from 6,016 field-year observations in Kenya during 2017–2020 using Sentinel-2 remote-sensing variables, climate, soil, terrain and location attributes, survey-based management information, and 64-dimensional satellite embeddings. Eleven candidate algorithms were compared using cross-validation, followed by XGBoost predictor-group and SHAP analyses. XGBoost achieved the strongest pooled performance (R² = 0.383; RMSE = 1,334 kg ha⁻¹; MAE = 1,013 kg ha⁻¹). Conventional remote sensing alone achieved R² = 0.196; adding climate, soil, terrain, and location increased R² to 0.250, while adding management information increased it to 0.386. Management accounted for 32.4% of grouped SHAP importance, with plant density as the most influential individual predictor. Satellite embeddings matched the predictive performance of conventional remote-sensing predictors while reducing satellite predictor dimensionality by approximately 77%. The results show that remote sensing and machine learning can support field-level yield monitoring in heterogeneous smallholder systems, but plot-level management information remains critical for improving prediction.
- 3 November 2026
- Published by MDPI AG in Preprints.org
Open AccessIntegrating Remote Sensing, Management Information, and Satellite Embeddings for Machine-Learning Prediction of Field-Level Maize Yield in Kenyan Smallholder Systems
Field-level crop yield prediction is an important application of precision agriculture, but it remains challenging in African smallholder systems because plots are small, management is heterogeneous, and satellite pixels can mix multiple land covers. This study evaluated machine-learning approaches for predicting maize yield from 6,016 field-year observations in Kenya during 2017–2020 using Sentinel-2 remote-sensing variables, climate, soil, terrain and location attributes, survey-based management information, and 64-dimensional satellite embeddings. Eleven candidate algorithms were compared using cross-validation, followed by XGBoost predictor-group and SHAP analyses. XGBoost achieved the strongest pooled performance (R² = 0.383; RMSE = 1,334 kg ha⁻¹; MAE = 1,013 kg ha⁻¹). Conventional remote sensing alone achieved R² = 0.196; adding climate, soil, terrain, and location increased R² to 0.250, while adding management information increased it to 0.386. Management accounted for 32.4% of grouped SHAP importance, with plant density as the most influential individual predictor. Satellite embeddings matched the predictive performance of conventional remote-sensing predictors while reducing satellite predictor dimensionality by approximately 77%. The results show that remote sensing and machine learning can support field-level yield monitoring in heterogeneous smallholder systems, but plot-level management information remains critical for improving prediction.
- Open Access
- 30 October 2026
- Published by MDPI AG in Preprints.org
A Machine-Learning Assessment of Climate Impacts on Global Agricultural Productivity
Climate change affects agricultural productivity through nonlinear and seasonally structured responses to temperature and precipitation. Global agricultural total factor productivity (TFP) attribution is commonly represented with a quadratic response to growing-season mean temperature and precipitation. We compare that benchmark with an expanded quadratic model and a Random Forest model using annual TFP growth for 168 countries, agricultural-area-weighted CRU TS observations, and eleven paired CMIP6 simulations with and without anthropogenic forcing. Under five-fold country-grouped cross-validation, the Random Forest shows higher out-of-sample predictive performance than either quadratic specification. Placebo reshuffling produces null R² distributions far below the observed value, while a year-by-region block bootstrap gives an average warming sensitivity near -0.9 percentage points of annual TFP growth per 1 °C. Held-out SHAP shows that precipitation, maximum temperature, and three-month climate windows contribute strongly to prediction, while accumulated local effects reveal nonlinear and direction-sensitive responses to year-to-year climate changes. Historical global attribution remains similar under the two response forms: the baseline quadratic gives -22.1% and the Random Forest -20.9% over 1962-2020. This agreement strengthens the finding that anthropogenic climate change has substantially reduced global agricultural productivity, while showing that a more flexible machine-learning response can reveal richer seasonal and nonlinear climate-TFP relationships. At the country level, attributed impacts are more dynamic and vary across rainfall-based moisture quintiles, with the Random Forest tending to attribute smaller losses than the quadratic in drier countries and larger losses in wetter countries.
- 30 October 2026
- Published by MDPI AG in Preprints.org
Open AccessA Machine-Learning Assessment of Climate Impacts on Global Agricultural Productivity
Climate change affects agricultural productivity through nonlinear and seasonally structured responses to temperature and precipitation. Global agricultural total factor productivity (TFP) attribution is commonly represented with a quadratic response to growing-season mean temperature and precipitation. We compare that benchmark with an expanded quadratic model and a Random Forest model using annual TFP growth for 168 countries, agricultural-area-weighted CRU TS observations, and eleven paired CMIP6 simulations with and without anthropogenic forcing. Under five-fold country-grouped cross-validation, the Random Forest shows higher out-of-sample predictive performance than either quadratic specification. Placebo reshuffling produces null R² distributions far below the observed value, while a year-by-region block bootstrap gives an average warming sensitivity near -0.9 percentage points of annual TFP growth per 1 °C. Held-out SHAP shows that precipitation, maximum temperature, and three-month climate windows contribute strongly to prediction, while accumulated local effects reveal nonlinear and direction-sensitive responses to year-to-year climate changes. Historical global attribution remains similar under the two response forms: the baseline quadratic gives -22.1% and the Random Forest -20.9% over 1962-2020. This agreement strengthens the finding that anthropogenic climate change has substantially reduced global agricultural productivity, while showing that a more flexible machine-learning response can reveal richer seasonal and nonlinear climate-TFP relationships. At the country level, attributed impacts are more dynamic and vary across rainfall-based moisture quintiles, with the Random Forest tending to attribute smaller losses than the quadratic in drier countries and larger losses in wetter countries.
- Open Access
- 15 May 2026
- Published by Elsevier in Smart Agricultural Technology
- Vol. 14
A comprehensive comparison of transformer and RNN models for corn yield prediction: A case study of the U.S. corn belt
- 15 May 2026
- Published by Elsevier in Smart Agricultural Technology
- Vol. 14
Open AccessA comprehensive comparison of transformer and RNN models for corn yield prediction: A case study of the U.S. corn belt
- Open Access
- 23 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19 (19391404) , 5814-5833
In-Season Sugarcane Mapping in the U.S. and Brazil Using Time-Invariant Phenological Features
As the world's largest sugar crop by production volume, sugarcane is a major economic resource used for both food and industrial materials worldwide. Timely sugarcane cultivation mapping will provide spatial distribution information on sugarcane fields and support further decision-making. Traditional time-series mapping using machine learning and remote sensing faces challenges due to limited training data across distant producers and heterogeneous growth seasons. To address these, an automated phenology-based transfer learning workflow was developed, improving upon our previous study, to produce in-season sugarcane maps for the U.S. and Brazil only using historical U.S. training data. Time-series Sentinel-2 multispectral data was transformed to time-invariant phenological features via Linear Cosine Regression (LCR), integrating high-confidence labels from the U.S. Cropland Data Layer (CDL) to develop random forest classifiers. Classifications were further optimized using the Product of Absolute Difference (PAD) and majority filters. The workflow produced four monthly sugarcane maps (June–September 2023) for Lafourche County, Louisiana, U.S., achieving 0.938 overall accuracy and 0.934 F1 score, surpassing CDL2023's 0.894, and reducing non sugarcane regions' average misclassification rate to 0.18%, validated by ground truth data. Similarly, five-monthly sugarcane maps (March–July 2022) for São Paulo, Brazil, attained 0.881 overall accuracy and 0.911 F1 score in May 2022, validated by the European Space Agency (ESA) Global Crop Type Validation Data Set. This study demonstrates the potential of using time-invariant phenological features for in-season sugarcane mapping across geographically distant nations and different time periods.
- 23 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19 (19391404) , 5814-5833
Open AccessIn-Season Sugarcane Mapping in the U.S. and Brazil Using Time-Invariant Phenological Features
As the world's largest sugar crop by production volume, sugarcane is a major economic resource used for both food and industrial materials worldwide. Timely sugarcane cultivation mapping will provide spatial distribution information on sugarcane fields and support further decision-making. Traditional time-series mapping using machine learning and remote sensing faces challenges due to limited training data across distant producers and heterogeneous growth seasons. To address these, an automated phenology-based transfer learning workflow was developed, improving upon our previous study, to produce in-season sugarcane maps for the U.S. and Brazil only using historical U.S. training data. Time-series Sentinel-2 multispectral data was transformed to time-invariant phenological features via Linear Cosine Regression (LCR), integrating high-confidence labels from the U.S. Cropland Data Layer (CDL) to develop random forest classifiers. Classifications were further optimized using the Product of Absolute Difference (PAD) and majority filters. The workflow produced four monthly sugarcane maps (June–September 2023) for Lafourche County, Louisiana, U.S., achieving 0.938 overall accuracy and 0.934 F1 score, surpassing CDL2023's 0.894, and reducing non sugarcane regions' average misclassification rate to 0.18%, validated by ground truth data. Similarly, five-monthly sugarcane maps (March–July 2022) for São Paulo, Brazil, attained 0.881 overall accuracy and 0.911 F1 score in May 2022, validated by the European Space Agency (ESA) Global Crop Type Validation Data Set. This study demonstrates the potential of using time-invariant phenological features for in-season sugarcane mapping across geographically distant nations and different time periods.
- Open Access
- 19 December 2025
- Published by Elsevier in International Journal of Applied Earth Observation and Geoinformation
- Vol. 146
Climatic effects on U.S. agricultural productivity: Evidence and prediction from total factor productivity and crop yields
- 19 December 2025
- Published by Elsevier in International Journal of Applied Earth Observation and Geoinformation
- Vol. 146
Open AccessClimatic effects on U.S. agricultural productivity: Evidence and prediction from total factor productivity and crop yields
- Open Access
- 21 October 2025
- Published by Copernicus GmbH in Proceedings of the ICA
- Vol. 7, 1-8
Identifying Geographic Disparities in Suicide Determinants Across United States Counties with Spatial Modeling
The United States is experiencing an unprecedented rise in suicide rates amid a growing mental health crisis. While many public health studies examine contextual factors associated with suicide using individual or group-level data, these approaches often overlook spatial dynamics that could reveal geographic and scale-dependent variations. County-level research on suicide determinants in the U.S. remains limited, leaving significant gaps in understanding geographic disparities. Therefore, this study leverages spatial analysis and modeling to identify clusters of high and low suicide rates across the U.S., and explores the geographic variations in the associations between social determinants and suicide to better understand these patterns. Using Local Moran’s I for cluster and outlier analysis, alongside comparisons between Ordinary Least Squares (OLS) and Multiscale Geographically Weighted Regression (MGWR) models, the analysis evaluates the relationships between age-adjusted suicide rates and 17 independent variables across 2,410 U.S. counties. Results uncover geographic heterogeneity, with higher suicide rates concentrated in the Western U.S. and several determinants demonstrating spatially varying effects. For instance, the percent of veteran population is a statistically significant predictor across all counties, with the strongest effects in the Northeast. Other variables show more localized patterns: living alone is significant in 14% of counties, mainly in the Southwest, and American Indian population is significant in 17% of counties, primarily in the West. Our findings highlight the non-stationarity of suicide determinants across geographic space and scales, providing a data-driven framework to identify geospatial disparities and inform the development of targeted intervention strategies to reduce suicides in specific locations.
- 21 October 2025
- Published by Copernicus GmbH in Proceedings of the ICA
- Vol. 7, 1-8
Open AccessIdentifying Geographic Disparities in Suicide Determinants Across United States Counties with Spatial Modeling
The United States is experiencing an unprecedented rise in suicide rates amid a growing mental health crisis. While many public health studies examine contextual factors associated with suicide using individual or group-level data, these approaches often overlook spatial dynamics that could reveal geographic and scale-dependent variations. County-level research on suicide determinants in the U.S. remains limited, leaving significant gaps in understanding geographic disparities. Therefore, this study leverages spatial analysis and modeling to identify clusters of high and low suicide rates across the U.S., and explores the geographic variations in the associations between social determinants and suicide to better understand these patterns. Using Local Moran’s I for cluster and outlier analysis, alongside comparisons between Ordinary Least Squares (OLS) and Multiscale Geographically Weighted Regression (MGWR) models, the analysis evaluates the relationships between age-adjusted suicide rates and 17 independent variables across 2,410 U.S. counties. Results uncover geographic heterogeneity, with higher suicide rates concentrated in the Western U.S. and several determinants demonstrating spatially varying effects. For instance, the percent of veteran population is a statistically significant predictor across all counties, with the strongest effects in the Northeast. Other variables show more localized patterns: living alone is significant in 14% of counties, mainly in the Southwest, and American Indian population is significant in 17% of counties, primarily in the West. Our findings highlight the non-stationarity of suicide determinants across geographic space and scales, providing a data-driven framework to identify geospatial disparities and inform the development of targeted intervention strategies to reduce suicides in specific locations.
- Open Access
- 8 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 8271-8290
A Decision Rule and Machine Learning-Based Hybrid Approach for Automated Land-Cover Type Local Climate Zones (LCZs) Mapping Using Multi-Source Remote Sensing Data
This article presents a streamlined, automated classification method to map land-cover-type local climate zones (LCZs). Using a two-phase hybrid approach, we first generated training samples through universal decision rules and subsequently, a machine learning (ML) algorithm was trained on the generated samples to classify LCZs. The proposed model harnesses plant height data, combined with spectral bands and remote sensing indices, to accurately classify various land-cover types, such as dense forest, scattered trees, bush/scrub, low plant/agricultural land, bare rocks/paved surface and bare soil/sands. Targeting global applicability, we tested our method across six diverse locations spanning four continents: Fresno (California), Central Michigan, Western Phoenix (Arizona), Khulna (Bangladesh), Lagos (Nigeria), and Western Sydney. In each location, after generating training samples with the decision rules, a random forest algorithm was employed for LCZ classification. Results showcase that data from sentinels 1 and 2, night-time light, and global ecosystem dynamics investigation relative height are effective in characterizing land-cover-type LCZs and decision-rules can be established. The decision-rules consistently auto-generate training samples, undeterred by varying geographical and climatic conditions. This automated system has achieved promising accuracy across all tested sites, suggesting its potential to map land-cover-type LCZs and vegetation globally with higher accuracy.
- 8 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 8271-8290
Open AccessA Decision Rule and Machine Learning-Based Hybrid Approach for Automated Land-Cover Type Local Climate Zones (LCZs) Mapping Using Multi-Source Remote Sensing Data
This article presents a streamlined, automated classification method to map land-cover-type local climate zones (LCZs). Using a two-phase hybrid approach, we first generated training samples through universal decision rules and subsequently, a machine learning (ML) algorithm was trained on the generated samples to classify LCZs. The proposed model harnesses plant height data, combined with spectral bands and remote sensing indices, to accurately classify various land-cover types, such as dense forest, scattered trees, bush/scrub, low plant/agricultural land, bare rocks/paved surface and bare soil/sands. Targeting global applicability, we tested our method across six diverse locations spanning four continents: Fresno (California), Central Michigan, Western Phoenix (Arizona), Khulna (Bangladesh), Lagos (Nigeria), and Western Sydney. In each location, after generating training samples with the decision rules, a random forest algorithm was employed for LCZ classification. Results showcase that data from sentinels 1 and 2, night-time light, and global ecosystem dynamics investigation relative height are effective in characterizing land-cover-type LCZs and decision-rules can be established. The decision-rules consistently auto-generate training samples, undeterred by varying geographical and climatic conditions. This automated system has achieved promising accuracy across all tested sites, suggesting its potential to map land-cover-type LCZs and vegetation globally with higher accuracy.
- 27 February 2024
- Published by Elsevier in ISPRS Journal of Photogrammetry and Remote Sensing
- Vol. 209, 415-431
Hazard or Non-Hazard Flood: Post Analysis for Paddy Rice, Wetland, and Other Potential Non-Hazard Flood Extraction from the VIIRS Flood Products
- 27 February 2024
- Published by Elsevier in ISPRS Journal of Photogrammetry and Remote Sensing
- Vol. 209, 415-431
Hazard or Non-Hazard Flood: Post Analysis for Paddy Rice, Wetland, and Other Potential Non-Hazard Flood Extraction from the VIIRS Flood Products
- Open Access
- 17 January 2023
- Published by MDPI AG in Atmosphere
- Vol. 14 (2) , 195
Investigating Tropical Cyclone Rapid Intensification with an Advanced Artificial Intelligence System and Gridded Reanalysis Data
Rapid Intensification (RI) in Tropical Cyclone (TC) development is one of the most difficult and still challenging tasks in weather forecasting. In addition to the dynamical numerical simulations, commonly used techniques for RI (as well as TC intensity changes) analysis and prediction are the composite analysis and statistical models based on features derived from the composite analysis. Quite a large number of such selected and pre-determined features related to TC intensity change and RI have been accumulated by the domain scientists, such as those in the widely used SHIPS (Statistical Hurricane Intensity Prediction Scheme) database. Moreover, new features are still being added with new algorithms and/or newly available datasets. However, there are very few unified frameworks for systematically distilling features from a comprehensive data source. One such unified Artificial Intelligence (AI) system was developed for deriving features from TC centers, and here, we expand that system to large-scale environmental condition. In this study, we implemented a deep learning algorithm, the Convolutional Neural Network (CNN), to the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis data and identified and refined potentially new features relevant to RI such as specific humidity in east or northeast, vorticity and horizontal wind in north and south relative to the TC centers, as well as ozone at high altitudes that could help the prediction and understanding of the occurrence of RI based on the deep learning network (named TCNET in this study). By combining the newly derived features and the features from the SHIPS database, the RI prediction performance can be improved by 43%, 23%, and 30% in terms of Kappa, probability of detection (POD), and false alarm rate (FAR) against the same modern classification model but with the SHIPS inputs only.
- 17 January 2023
- Published by MDPI AG in Atmosphere
- Vol. 14 (2) , 195
Open AccessInvestigating Tropical Cyclone Rapid Intensification with an Advanced Artificial Intelligence System and Gridded Reanalysis Data
Rapid Intensification (RI) in Tropical Cyclone (TC) development is one of the most difficult and still challenging tasks in weather forecasting. In addition to the dynamical numerical simulations, commonly used techniques for RI (as well as TC intensity changes) analysis and prediction are the composite analysis and statistical models based on features derived from the composite analysis. Quite a large number of such selected and pre-determined features related to TC intensity change and RI have been accumulated by the domain scientists, such as those in the widely used SHIPS (Statistical Hurricane Intensity Prediction Scheme) database. Moreover, new features are still being added with new algorithms and/or newly available datasets. However, there are very few unified frameworks for systematically distilling features from a comprehensive data source. One such unified Artificial Intelligence (AI) system was developed for deriving features from TC centers, and here, we expand that system to large-scale environmental condition. In this study, we implemented a deep learning algorithm, the Convolutional Neural Network (CNN), to the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim reanalysis data and identified and refined potentially new features relevant to RI such as specific humidity in east or northeast, vorticity and horizontal wind in north and south relative to the TC centers, as well as ozone at high altitudes that could help the prediction and understanding of the occurrence of RI based on the deep learning network (named TCNET in this study). By combining the newly derived features and the features from the SHIPS database, the RI prediction performance can be improved by 43%, 23%, and 30% in terms of Kappa, probability of detection (POD), and false alarm rate (FAR) against the same modern classification model but with the SHIPS inputs only.