Abstract
Spatial crop yield prediction provides valuable insights for supporting sustainable and precise crop management decisions. This study assessed the capabilities of advanced Deep Learning (DL) architectures in predicting within-field soybean yields using spectral bands from Sentinel-2 (RS-Inputs), weather (W-Inputs), and topographic attributes (TA-Inputs). DL architectures included 1-D convolutional neural network (1D-CNN), long short-term memory (LSTM) and transformer-based (TRFM). We used an extensive dataset ( ∼ 700 K) of yield observations, collected with a combine harvester, from 310 fields across three growing seasons (2020, 2021 and 2022) in Uruguay. The DL architectures were assessed under two testing strategies: across-fields and across-years. We compared results from DL architectures against a baseline that uses a process-based method with data assimilation. Our results revealed that DL architectures outperformed the baseline in testing across-fields only (RRMSE 35 % vs 40 %). The DL architectures encountered more challenges with temporal extrapolation (RRMSE 51 % in across-years). There were no substantial differences in performance among the DL architectures. The TA-Inputs enhanced accuracy in 1D-CNN (reduced RRMSE by ∼ 13 %), while W-Inputs led to a small improvement in 1D-CNN and LSTM (reduced RRMSE by ∼ 2 %) when tested across-years. All combinations of DL architectures and input settings encountered challenges in predicting the tails of the yield distribution (mean bias ∼ 1000 kg ha −1 ). We discussed the current limitations of DL architectures in capturing crop yield complexity using openly available spatial data and provided further directions for improving the reliability and interpretability of data-driven models by integrating process-based approaches. Beyond this, the performance of data-driven methods alone is expected to improve with the increasing availability of collected and stored data. Incorporating historical yield maps and in-season crop management data into open-source datasets will facilitate continuous training and enhancement for tailored models. • A process-based approach was superior to DL in temporal extrapolation. • DL is more promising for predicting unseen fields than years (RRMSE 35 % vs 51 %). • Performance was comparable among DL architectures. • Adding weather and topography inputs had a minor impact on accuracy. • Errors increased towards the tails of the distribution across all models and inputs.
Abstract from the paper’s OpenAlex record; open copy licensed cc-by. Read at the source ↗
The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.
Claims 6
Claim · model-written summary
Deep learning architectures outperformed the process-based baseline in across-field testing of soybean yield prediction (RRMSE 35% vs 40%).
- Taxon
- soybean
Glycine maxe0 - Outcome
- RRMSEe1
- Direction
- outperformede1
- Value
- 35 %e1
- Qualifier
- across-fields onlye1
- Comparator
- baselinee1
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“predicting within-field soybean yields”
e1“DL architectures outperformed the baseline in testing across-fields only (RRMSE 35 % vs 40 %)”
Claim · model-written summary
Deep learning architectures encountered more challenges in temporal extrapolation of soybean yield, with RRMSE 51% across years.
- Taxon
- soybeanGlycine maxe0
- Outcome
- RRMSEe2
- Value
- 51 %e2
- Qualifier
- temporal extrapolatione2
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“predicting within-field soybean yields”
e2“The DL architectures encountered more challenges with temporal extrapolation (RRMSE 51 % in across-years)”
Claim · model-written summary
There were no substantial differences in soybean yield prediction performance among the deep learning architectures.
- Taxon
- soybeanGlycine maxe0
- Outcome
- performancee3
- Direction
- no substantial differencese3
- Comparator
- DL architecturese3
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“predicting within-field soybean yields”
e3“There were no substantial differences in performance among the DL architectures”
Claim · model-written summary
Topographic attribute inputs reduced RRMSE of the 1D-CNN by about 13%.
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“predicting within-field soybean yields”
e4“The TA-Inputs enhanced accuracy in 1D-CNN (reduced RRMSE by ∼ 13 %)”
Claim · model-written summary
Weather inputs reduced RRMSE of the 1D-CNN and LSTM by about 2% when tested across years.
- Taxon
- soybeanGlycine maxe0
- Treatment
- W-Inputse5
- Outcome
- RRMSEe5
- Direction
- reducede5
- Value
- ∼ 2 %e5
- Qualifier
- when tested across-yearse5
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“predicting within-field soybean yields”
e5“W-Inputs led to a small improvement in 1D-CNN and LSTM (reduced RRMSE by ∼ 2 %) when tested across-years”
Claim · model-written summary
All architectures and input settings had a mean bias of about 1000 kg ha -1 in predicting the tails of the soybean yield distribution.
- Taxon
- soybeanGlycine maxe0
- Outcome
- mean biase6
- Value
- ∼ 1000 kg ha −1e6
- Qualifier
- tails of the yield distributione6
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
e0“predicting within-field soybean yields”
e6“All combinations of DL architectures and input settings encountered challenges in predicting the tails of the yield distribution (mean bias ∼ 1000 kg ha −1 )”