Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data

Accurate pre-harvest prediction of yield and quality (tenderometric reading, TR) is crucial for the processing pea industry due to a narrow optimal harvest window. Machine learning (ML) models offer potential, but their real-world utility depends on their performance stability across different years...

Full description

Saved in:
Bibliographic Details
Main Authors: Michele Croci, Manuele Ragazzi, Alessandro Grassi, Giorgio Impollonia, Stefano Amaducci
Format: Article
Language:English
Published: Elsevier 2025-12-01
Series:Smart Agricultural Technology
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2772375525004381
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849408439627808768
author Michele Croci
Manuele Ragazzi
Alessandro Grassi
Giorgio Impollonia
Stefano Amaducci
author_facet Michele Croci
Manuele Ragazzi
Alessandro Grassi
Giorgio Impollonia
Stefano Amaducci
author_sort Michele Croci
collection DOAJ
description Accurate pre-harvest prediction of yield and quality (tenderometric reading, TR) is crucial for the processing pea industry due to a narrow optimal harvest window. Machine learning (ML) models offer potential, but their real-world utility depends on their performance stability across different years. This ability, known as temporal transferability, is often overestimated by standard validation methods, creating a gap between theoretical performance and operational reliability. This study aims to rigorously quantify this temporal transferability gap for both pea yield and TR prediction. Four ML algorithms (RF, XGBoost, GPR, SVMr) were evaluated using Sentinel-2 and ERA5-Land data from 2018 to 2024 in northern Italy. A comparison was made between a standard group-based cross-validation (LOGOCV) and a temporally rigorous Leave-One-Year-Out Cross-Validation (LOYOCV). For yield prediction, ML models outperformed a baseline (NullModel) under LOGOCV (SVMr nRMSE = 18.4 %), but performance degraded significantly under LOYOCV, revealing a clear transferability gap. TR prediction was more challenging while RF showed promising results in LOGOCV (nRMSE = 22.1 %), all ML models were outperformed by the NullModel in the more realistic LOYOCV scenario. The findings highlight a critical temporal transferability gap, especially for the TR parameter, limiting the current operational readiness of standard ML models. It is recommended that future work focus on more robust approaches, such as models designed for temporal data (e.g., RNNs, Transformers) and higher-resolution data, to bridge the gap towards reliable real-world application.
format Article
id doaj-art-7f70bcccce6647369692e47bdb4273bd
institution Kabale University
issn 2772-3755
language English
publishDate 2025-12-01
publisher Elsevier
record_format Article
series Smart Agricultural Technology
spelling doaj-art-7f70bcccce6647369692e47bdb4273bd2025-08-20T03:35:47ZengElsevierSmart Agricultural Technology2772-37552025-12-011210120710.1016/j.atech.2025.101207Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land dataMichele Croci0Manuele Ragazzi1Alessandro Grassi2Giorgio Impollonia3Stefano Amaducci4Department of Sustainable Crop Production, Università Cattolica del Sacro Cuore di Piacenza, via Emilia Parmense, 84, 29122 Piacenza, PC, Italy; Corresponding author.Department of Sustainable Crop Production, Università Cattolica del Sacro Cuore di Piacenza, via Emilia Parmense, 84, 29122 Piacenza, PC, ItalyConsorzio Casalasco Del Pomodoro, Strada provinciale, 32, 26036 Rivarolo del Re, CR, ItalyDepartment of Sustainable Crop Production, Università Cattolica del Sacro Cuore di Piacenza, via Emilia Parmense, 84, 29122 Piacenza, PC, ItalyDepartment of Sustainable Crop Production, Università Cattolica del Sacro Cuore di Piacenza, via Emilia Parmense, 84, 29122 Piacenza, PC, ItalyAccurate pre-harvest prediction of yield and quality (tenderometric reading, TR) is crucial for the processing pea industry due to a narrow optimal harvest window. Machine learning (ML) models offer potential, but their real-world utility depends on their performance stability across different years. This ability, known as temporal transferability, is often overestimated by standard validation methods, creating a gap between theoretical performance and operational reliability. This study aims to rigorously quantify this temporal transferability gap for both pea yield and TR prediction. Four ML algorithms (RF, XGBoost, GPR, SVMr) were evaluated using Sentinel-2 and ERA5-Land data from 2018 to 2024 in northern Italy. A comparison was made between a standard group-based cross-validation (LOGOCV) and a temporally rigorous Leave-One-Year-Out Cross-Validation (LOYOCV). For yield prediction, ML models outperformed a baseline (NullModel) under LOGOCV (SVMr nRMSE = 18.4 %), but performance degraded significantly under LOYOCV, revealing a clear transferability gap. TR prediction was more challenging while RF showed promising results in LOGOCV (nRMSE = 22.1 %), all ML models were outperformed by the NullModel in the more realistic LOYOCV scenario. The findings highlight a critical temporal transferability gap, especially for the TR parameter, limiting the current operational readiness of standard ML models. It is recommended that future work focus on more robust approaches, such as models designed for temporal data (e.g., RNNs, Transformers) and higher-resolution data, to bridge the gap towards reliable real-world application.http://www.sciencedirect.com/science/article/pii/S2772375525004381Yield predictionQuality predictionSentinel-2Machine learningTemporal transferability
spellingShingle Michele Croci
Manuele Ragazzi
Alessandro Grassi
Giorgio Impollonia
Stefano Amaducci
Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data
Smart Agricultural Technology
Yield prediction
Quality prediction
Sentinel-2
Machine learning
Temporal transferability
title Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data
title_full Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data
title_fullStr Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data
title_full_unstemmed Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data
title_short Assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using Sentinel-2 and ERA5-land data
title_sort assessing the temporal transferability of machine learning models for predicting processing pea yield and quality using sentinel 2 and era5 land data
topic Yield prediction
Quality prediction
Sentinel-2
Machine learning
Temporal transferability
url http://www.sciencedirect.com/science/article/pii/S2772375525004381
work_keys_str_mv AT michelecroci assessingthetemporaltransferabilityofmachinelearningmodelsforpredictingprocessingpeayieldandqualityusingsentinel2andera5landdata
AT manueleragazzi assessingthetemporaltransferabilityofmachinelearningmodelsforpredictingprocessingpeayieldandqualityusingsentinel2andera5landdata
AT alessandrograssi assessingthetemporaltransferabilityofmachinelearningmodelsforpredictingprocessingpeayieldandqualityusingsentinel2andera5landdata
AT giorgioimpollonia assessingthetemporaltransferabilityofmachinelearningmodelsforpredictingprocessingpeayieldandqualityusingsentinel2andera5landdata
AT stefanoamaducci assessingthetemporaltransferabilityofmachinelearningmodelsforpredictingprocessingpeayieldandqualityusingsentinel2andera5landdata