Abstract
By generating high quality data without the big time investment and economic cost of real experiments, dynamic greenhouse climate and crop simulation models can support decisions on greenhouse climate control, crop management and greenhouse design. The reliability of simulation-based decisions depends on both the prediction accuracy and interpretability of simulation models. The prediction accuracy of these simulation models can be increased by: 1) improving mechanisms in process-based models; 2) calibrating process-based model parameters; 3) deriving black-box relationships from data. Considering the descending interpretability from (1) to (3), this study presents a knowledge-based data-driven modelling approach where firstly a process-based model is selected and modified based on domain knowledge, then data-driven improvement is applied including two steps: parameter value estimation by particle filter (PF) and further black-box improvement by deep neural networks (DNN). The approach was tested with an example of greenhouse climate-tomato production system modelling. Modules from GreenLight (Katzin et al., 2020) and TOMSIM (Heuvelink, 1995, Heuvelink, 1996) were selected, modified and integrated into a process-based greenhouse climate-tomato model. Validation showed that PF-calibration of five greenhouse parameters decreased the seasonal relative root mean squared error (RRMSE) of indoor air vapor pressure predictions from 40.7% of that before PF-calibration to 16.4%, while it did not decrease the RRMSE of indoor air temperature predictions. Combining the PF-calibrated model with a DNN trained on a season of data decreased the RRMSE of indoor air temperature from 15.0% without DNN to 6.7%, and decreased the RRMSE of indoor air vapor pressure to 12.6%. The knowledge-based data-driven greenhouse climate-tomato model had a relative error of 0.9% for seasonal total fresh yield, and an RRMSE of 6.6% for the cumulative yield throughout the season. If process-based model parameters were not calibrated before combining the model with DNNs, the required amount and diversity of DNN training data increased because more information needed to be learnt from data by the DNNs. Without PF-calibration, combining a DNN trained on 50 days of data with the process-based model resulted in RRMSEs of 44.8% and 31.8% for indoor air temperature and vapor pressure prediction, respectively; with PF-calibration, the RRMSEs were decreased to 13.1% and 17.9%. The proposed three-step knowledge-based data-driven approach can not only improve the model prediction accuracy, but can also help to track and interpret the improvements.
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 8
Claim · model-written summary
PF-calibration decreased the seasonal RRMSE of indoor air vapor pressure predictions to 16.4% from 40.7% before calibration.
- Taxon
- tomato
Solanum lycopersicumt0 - Treatment
- PF-calibration of five greenhouse parameterse0
- Outcome
- seasonal relative root mean squared error (RRMSE) of indoor air vapor pressure predictionse0
- Direction
- decreasede0
- Value
- 16.4%e0
- Comparator
- before PF-calibratione0
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e0“Validation showed that PF-calibration of five greenhouse parameters decreased the seasonal relative root mean squared error (RRMSE) of indoor air vapor pressure predictions from 40.7% of that before PF-calibration to 16.4%”
Claim · model-written summary
PF-calibration did not decrease the RRMSE of indoor air temperature predictions.
- Taxon
- tomatoSolanum lycopersicumt0
- Treatment
- PF-calibration of five greenhouse parameterse0
- Outcome
- RRMSE of indoor air temperature predictionse1
- Direction
- did not decreasee1
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e0“Validation showed that PF-calibration of five greenhouse parameters decreased the seasonal relative root mean squared error (RRMSE) of indoor air vapor pressure predictions from 40.7% of that before PF-calibration to 16.4%”
e1“while it did not decrease the RRMSE of indoor air temperature predictions”
Claim · model-written summary
Combining the PF-calibrated model with a DNN decreased the RRMSE of indoor air temperature to 6.7% from 15.0% without DNN.
- Taxon
- tomatoSolanum lycopersicumt0
- Treatment
- Combining the PF-calibrated model with a DNN trained on a season of datae2
- Outcome
- RRMSE of indoor air temperaturee2
- Direction
- decreasede2
- Value
- 6.7%e2
- Comparator
- without DNNe2
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e2“Combining the PF-calibrated model with a DNN trained on a season of data decreased the RRMSE of indoor air temperature from 15.0% without DNN to 6.7%”
Claim · model-written summary
Combining with a DNN decreased the RRMSE of indoor air vapor pressure to 12.6%.
- Taxon
- tomatoSolanum lycopersicumt0
- Outcome
- RRMSE of indoor air vapor pressuree3
- Direction
- decreasede3
- Value
- 12.6%e3
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e3“and decreased the RRMSE of indoor air vapor pressure to 12.6%”
Claim · model-written summary
The model had a relative error of 0.9% for seasonal total fresh yield.
- Taxon
- tomatoSolanum lycopersicumt0
- Outcome
- seasonal total fresh yielde4
- Value
- 0.9%e4
- Qualifier
- relative errore4
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e4“had a relative error of 0.9% for seasonal total fresh yield”
Claim · model-written summary
The model had an RRMSE of 6.6% for cumulative yield over the season.
- Taxon
- tomatoSolanum lycopersicumt0
- Outcome
- cumulative yielde5
- Value
- 6.6%e5
- Qualifier
- RRMSEe5
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e5“an RRMSE of 6.6% for the cumulative yield throughout the season”
Claim · model-written summary
Without calibration before combining with DNNs, the required amount and diversity of DNN training data increased.
- Taxon
- tomatoSolanum lycopersicumt0
- Treatment
- not calibrated before combining the model with DNNse6
- Outcome
- the required amount and diversity of DNN training datae6
- Direction
- increasede6
- Qualifier
- because more information needed to be learnt from data by the DNNse6
Not reportedComparator
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e6“If process-based model parameters were not calibrated before combining the model with DNNs, the required amount and diversity of DNN training data increased because more information needed to be learnt from data by the DNNs”
Claim · model-written summary
The three-step knowledge-based data-driven approach is recommended to improve model prediction accuracy.
- Taxon
- tomatoSolanum lycopersicumt0
- Outcome
- improve the model prediction accuracye7
Organisms pictured · Representative pictures of the matched species, not figures from the paper.
Verbatim
t0“greenhouse climate-tomato production model”
e7“The proposed three-step knowledge-based data-driven approach can not only improve the model prediction accuracy, but can also help to track and interpret the improvements”
The same questions in other papers
cultivation system · soil: physics · Solanum lycopersicum (tomato)Positive and negative results
2 papers in the same direction as this paper · 5 in the opposite direction
- positive
- positive
- positive
4 papers more in this group
See all 8 papers →Claims are grouped by organism, treatment and what was measured or targeted. A paper counts as positive when any of its claims in the group is positive.