2024
1.
Ozmen Erkin Kokten; Raviv Raich; James Holmes; Alan Fern
Learning Extended Forecasts of Soil Water Content via Physically-Inspired Autoregressive Models Proceedings Article
In: 2024 International Conference on Machine Learning and Applications (ICMLA), pp. 400–407, 2024, (ISSN: 1946-0759).
Abstract | Links | BibTeX | Tags: autoregressive training, non-linear state-space models, Pipelines, Predictive models, Soil measurements, Soil Water Content, State-space methods, Stress, teacher-forcing, time-series, Training, Training data, Weather forecasting
@inproceedings{kokten_learning_2024,
title = {Learning Extended Forecasts of Soil Water Content via Physically-Inspired Autoregressive Models},
author = {Ozmen Erkin Kokten and Raviv Raich and James Holmes and Alan Fern},
url = {https://ieeexplore.ieee.org/abstract/document/10903312},
doi = {10.1109/ICMLA61862.2024.00060},
year = {2024},
date = {2024-12-01},
urldate = {2024-12-01},
booktitle = {2024 International Conference on Machine Learning and Applications (ICMLA)},
pages = {400\textendash407},
abstract = {Vine stress resulting from soil water content (SWC) restrictions allows growers to improve grape and subsequent wine quality. In this work, we consider learning models that can forecast SWC to assist growers' irrigation decisions. In particular, we investigate training auto-regressive recurrent neural networks to make multi-day hourly forecasts of SWC based on historical data from soil-moisture sensors, irrigation sched-ules, and evapotranspiration estimates. Our work addresses two practical challenges in training such models. First, trained auto-regressive models are prone to error propagation, which quickly degrades longer-term forecasts. Second, it is difficult to learn the underlying causal relationship between irrigation and soil moisture due to the training data having limited coverage of the primary control input, irrigation. We propose a training strategy that combines one-step teacher forcing loss with a loss over multi-step autoregressive predictions and novel regularization terms to ensure SWC forecasts align with scientific models, effectively addressing the key challenges. We present results from five irrigation blocks with two cultivars, using datasets ranging from 2947 to 4784 hourly measurements of SWC, irrigation, and weather. Our methodology achieves precise SWC predictions and generates realistic forecasts for untrained irrigation scenarios.},
note = {ISSN: 1946-0759},
keywords = {autoregressive training, non-linear state-space models, Pipelines, Predictive models, Soil measurements, Soil Water Content, State-space methods, Stress, teacher-forcing, time-series, Training, Training data, Weather forecasting},
pubstate = {published},
tppubtype = {inproceedings}
}
Vine stress resulting from soil water content (SWC) restrictions allows growers to improve grape and subsequent wine quality. In this work, we consider learning models that can forecast SWC to assist growers' irrigation decisions. In particular, we investigate training auto-regressive recurrent neural networks to make multi-day hourly forecasts of SWC based on historical data from soil-moisture sensors, irrigation sched-ules, and evapotranspiration estimates. Our work addresses two practical challenges in training such models. First, trained auto-regressive models are prone to error propagation, which quickly degrades longer-term forecasts. Second, it is difficult to learn the underlying causal relationship between irrigation and soil moisture due to the training data having limited coverage of the primary control input, irrigation. We propose a training strategy that combines one-step teacher forcing loss with a loss over multi-step autoregressive predictions and novel regularization terms to ensure SWC forecasts align with scientific models, effectively addressing the key challenges. We present results from five irrigation blocks with two cultivars, using datasets ranging from 2947 to 4784 hourly measurements of SWC, irrigation, and weather. Our methodology achieves precise SWC predictions and generates realistic forecasts for untrained irrigation scenarios.
2.
Srikanth Gorthi; Dattatray G. Bhalekar; Lav R. Khot; Markus Keller
Modeling Grape Berry Temperature for Effective Heat Stress Management in Vineyards Proceedings Article
In: 2024 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor), pp. 237–241, 2024.
Abstract | Links | BibTeX | Tags: Atmospheric modeling, Berry Temperature, Lasso Regression, Pipelines, Random Forest, Random forests, Real-time systems, Ridge Regression, Soil, Soil measurements, Solar radiation, Stress, Temperature measurement, Water heating
@inproceedings{gorthi_modeling_2024,
title = {Modeling Grape Berry Temperature for Effective Heat Stress Management in Vineyards},
author = {Srikanth Gorthi and Dattatray G. Bhalekar and Lav R. Khot and Markus Keller},
url = {https://ieeexplore.ieee.org/document/10948825},
doi = {10.1109/MetroAgriFor63043.2024.10948825},
year = {2024},
date = {2024-10-01},
urldate = {2024-10-01},
booktitle = {2024 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor)},
pages = {237\textendash241},
abstract = {This study developed a machine learning model to predict berry temperature using localized weather and soil attributes measured during the summer of 2023. Berry temperature was observed to be higher compared to canopy and air temperature during extreme heat events. A tree-based pipeline optimization tool was used to find an optimum machine learning algorithm. Amongst the tested models, Lasso Regression exhibited reasonable accuracy (R2 = 0.98) and root mean squared error of 0.93 °C on the test dataset.},
keywords = {Atmospheric modeling, Berry Temperature, Lasso Regression, Pipelines, Random Forest, Random forests, Real-time systems, Ridge Regression, Soil, Soil measurements, Solar radiation, Stress, Temperature measurement, Water heating},
pubstate = {published},
tppubtype = {inproceedings}
}
This study developed a machine learning model to predict berry temperature using localized weather and soil attributes measured during the summer of 2023. Berry temperature was observed to be higher compared to canopy and air temperature during extreme heat events. A tree-based pipeline optimization tool was used to find an optimum machine learning algorithm. Amongst the tested models, Lasso Regression exhibited reasonable accuracy (R2 = 0.98) and root mean squared error of 0.93 °C on the test dataset.
3.
Basavaraj R. Amogi; Lav R. Khot; Bernardita V. Sallato
Localized Crop Physiology Sensing System Driven Apple Fruit Color Progression Monitoring Proceedings Article
In: 2024 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor), pp. 232–236, 2024.
Abstract | Links | BibTeX | Tags: Biomedical monitoring, Crops, cyber physical system, fruit color, hue, Image color analysis, Monitoring, Prevention and mitigation, Real-time systems, Sensors, Solar heating, Solar radiation, Stress
@inproceedings{amogi_localized_2024,
title = {Localized Crop Physiology Sensing System Driven Apple Fruit Color Progression Monitoring},
author = {Basavaraj R. Amogi and Lav R. Khot and Bernardita V. Sallato},
url = {https://ieeexplore.ieee.org/document/10948846},
doi = {10.1109/MetroAgriFor63043.2024.10948846},
year = {2024},
date = {2024-10-01},
urldate = {2024-10-01},
booktitle = {2024 IEEE International Workshop on Metrology for Agriculture and Forestry (MetroAgriFor)},
pages = {232\textendash236},
abstract = {Fruit color is a critical quality attribute that significantly affects the commercial value of apples. Elevated air temperatures and solar radiation during heat waves can substantially impact fruit coloration, while mitigation techniques such as netting can further compromise fruit color development by trapping heat. Continuous monitoring of fruit color throughout the growing season, particularly under heat stress conditions, is thus essential for informed grower decision-making. This study leverages a previously developed localized crop physiology sensing system (CPSS) to enable real time monitoring of apple fruit color progression. Using visible imagery captured by the CPSS, apple fruit color was quantified by extracting the hue angle (ˆtextbackslashcirctextbackslashmathbfh). The results showed that, the influence of sunlight on measured color accuracy (ΔE) is lower and more stable during mid-day hours (1000 h − 1300 h), whereas notable variations were observed during early morning and late afternoon periods. The ˆtextbackslashcirctextbackslashmathbfh hence calculated for RGB images captured around 1200 h was utilized to track fruit color progression. Its monitoring over the summer 2022 growing season showed variations as an effect of environmental stressors, especially in response to heat wave. The developed approach offers a tool for growers to adjust different heat mitigation techniques during heat wave/events to mitigate any negative implication on fruit coloration.},
keywords = {Biomedical monitoring, Crops, cyber physical system, fruit color, hue, Image color analysis, Monitoring, Prevention and mitigation, Real-time systems, Sensors, Solar heating, Solar radiation, Stress},
pubstate = {published},
tppubtype = {inproceedings}
}
Fruit color is a critical quality attribute that significantly affects the commercial value of apples. Elevated air temperatures and solar radiation during heat waves can substantially impact fruit coloration, while mitigation techniques such as netting can further compromise fruit color development by trapping heat. Continuous monitoring of fruit color throughout the growing season, particularly under heat stress conditions, is thus essential for informed grower decision-making. This study leverages a previously developed localized crop physiology sensing system (CPSS) to enable real time monitoring of apple fruit color progression. Using visible imagery captured by the CPSS, apple fruit color was quantified by extracting the hue angle (ˆtextbackslashcirctextbackslashmathbfh). The results showed that, the influence of sunlight on measured color accuracy (ΔE) is lower and more stable during mid-day hours (1000 h − 1300 h), whereas notable variations were observed during early morning and late afternoon periods. The ˆtextbackslashcirctextbackslashmathbfh hence calculated for RGB images captured around 1200 h was utilized to track fruit color progression. Its monitoring over the summer 2022 growing season showed variations as an effect of environmental stressors, especially in response to heat wave. The developed approach offers a tool for growers to adjust different heat mitigation techniques during heat wave/events to mitigate any negative implication on fruit coloration.
