Bayesian Multi-modeling of Deep Neural Nets for Probabilistic Crop Yield Prediction

Abbaszadeh, Peyman; Gavahi, Keyhan; Alipour, Atieh; Deb, Proloy; Moradkhani, Hamid

doi:10.1016/j.agrformet.2021.108773

Cited by 58 publications

(19 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We observed that the 1‐ to 3‐month lead SSFI predictions mostly fell within the 25% predictive interval, indicating that our proposed model can produce a higher accuracy to the predicted hydrological drought. Besides, the uncertainty interval associated with predictions can help stakeholders effectively decide and plan taking into account the risk of crop production failure and water stress (Abbaszadeh et al., 2022).…”

Section: Resultsmentioning

confidence: 99%

“…Bayesian model averaging (BMA), a widely utilized multiple model combination technique, can assign different weights to each ensemble member which depends on the explanatory power of the member itself for the specified objective (Duan et al., 2019; Liu et al., 2021; Long et al., 2017; Raftery et al., 2005; Zarekarizi et al., 2021). This technique has gained popularity in geosciences and hydrology, such as multi‐source data processing and ensemble prediction (Abbaszadeh et al., 2022; Ajami et al., 2007; Duan et al., 2007; Madadgar & Moradkhani, 2014a; Miao et al., 2020; Sehgal et al., 2017). For instance, Miao et al.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Predicting Hydrological Drought With Bayesian Model Averaging Ensemble Vine Copula (BMAViC) Model

Singh

et al. 2022

Water Resources Research

View full text Add to dashboard Cite

Streamflow deficit (hydrological drought) poses a large risk to water resources management, agricultural production, water supply, hydropower generation, and ecosystem services. Reliable and robust hydrological drought predictions are critical for water and food security and ecosystem health under anthropogenic warming. However, the prevalent statistical prediction methods, for example, the meta‐Gaussian (MG) model, usually do not lead to accurate drought predictions. We therefore developed a new drought prediction model utilizing the Bayesian Model Averaging coupled with Vine Copula, called Bayesian Model Averaging Ensemble Vine Copula (BMAViC) model, in which previous meteorological drought, antecedent evaporative drought, and preceding hydrological drought were selected as three predictors. The BMAViC model was applied to the Upper Yellow River basin and showed robust skills during calibration and validation periods for 1‐ to 3‐month lead hydrological drought predictions. In comparison with the MG model (reference model), the skills of the proposed model were relatively stable and superior under diverse lead times. Good performances under the 1‐ to 3‐month lead times strongly implied that the BMAViC model yielded robust and accurate hydrological drought predictions. The study results enhance our confidence in seasonal drought prediction and help us understand drought dynamics in future months.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Predicting Hydrological Drought With Bayesian Model Averaging Ensemble Vine Copula (BMAViC) Model

Singh

et al. 2022

Water Resources Research

View full text Add to dashboard Cite

show abstract

“…The incorporation of DL into environmental remote sensing has allowed for its use in a wide variety of applications, such as land cover mapping, environmental parameter retrieval, data fusion and downscaling, information production and prediction, and so on, all thanks to DL's superior ability in feature representation [78].…”

Section: How ML Helps Agrarian Factors and Remote Sensing In Crop Yie...mentioning

confidence: 99%

Predicting Crop Yield Using Deep Learning and Remote Sensing

Bharadiya

Tzenios²,

Reddy

2023

JERR

View full text Add to dashboard Cite

The art of predicting crop production is done before the crop is harvested. Crop output forecasts will help people make timely judgments concerning food policy, prices in markets, import and export laws, and acceptable warehousing. It is possible to reduce the socioeconomic effects of crop loss brought on by a natural disaster, such as a flood or a drought, and to organize humanitarian food assistance. It has been suggested that deep learning, which lets the model to automatically extricate features and learn from the datasets, could be useful for predicting agricultural yields. This review helps to understand that how vegetation indices and environmental variables affect agricultural output by revealing gaps in our understanding of deep learning methodologies and remote sensing data in a specific area. Literature review of 2011-2022 has been collected from different databases and sites and analyzed to meet the aims of this review. The study mainly focused on the benefits of machine learning, agrarian factors and remote sensing for forecasting crop yield. The most often employed form of remote sensing is satellite technology, namely the usage of the Moderate-Resolution Imaging Spectro radiometer. Vegetation indices referred to as the most often employed attribute for forecasting crop yield, according to the results. This review compares all these techniques and pros and cons related to them.

show abstract

“…An average coefficient of determination (R 2 ) of 0.77 was reported for the testing year from 2010 to 2019 for the U.S. Corn Belt. Abbaszadeh et al (2022) proposed a statistical framework to perform probabilistic prediction of crop yield coupling the Bayesian model with deep learning models. It should be noted that despite the improved prediction accuracy, the underlying black-box characteristics of the neural network-based crop yield prediction models make it challenging to interpret the model and subsequent decisionmaking (Rudin, 2019).…”

Section: Introductionmentioning

confidence: 99%

“…An average coefficient of determination (R 2 ) of 0.77 was reported for the testing year from 2010 to 2019 for the U.S. Corn Belt. Abbaszadeh et al. (2022) proposed a statistical framework to perform probabilistic prediction of crop yield coupling the Bayesian model with deep learning models.…”

Section: Introductionmentioning

confidence: 99%

County-scale crop yield prediction by integrating crop simulation with machine learning models

et al. 2022

View full text Add to dashboard Cite

Crop yield prediction is of great importance for decision making, yet it remains an ongoing scientific challenge. Interactions among different genetic, environmental, and management factors and uncertainty in input values are making crop yield prediction complex. Building upon a previous work in which we coupled crop modeling with machine learning (ML) models to predict maize yields for three US Corn Belt states, here, we expand the concept to the entire US Corn Belt (12 states). More specifically, we built five new ML models and their ensemble models, considering the scenarios with and without crop modeling variables. Additional input values in our models are soil, weather, management, and historical yield data. A unique aspect of our work is the spatial analysis to investigate causes for low or high model prediction errors. Our results indicated that the prediction accuracy increases by coupling crop modeling with machine learning. The ensemble model overperformed the individual ML models, having a relative root mean square error (RRMSE) of about 9% for the test years (2018, 2019, and 2020), which is comparable to previous studies. In addition, analysis of the sources of error revealed that counties and crop reporting districts with low cropland ratios have high RRMSE. Furthermore, we found that soil input data and extreme weather events were responsible for high errors in some regions. The proposed models can be deployed for large-scale prediction at the county level and, contingent upon data availability, can be utilized for field level prediction.

show abstract

Bayesian Multi-modeling of Deep Neural Nets for Probabilistic Crop Yield Prediction

Cited by 58 publications

References 49 publications

Predicting Hydrological Drought With Bayesian Model Averaging Ensemble Vine Copula (BMAViC) Model

Predicting Hydrological Drought With Bayesian Model Averaging Ensemble Vine Copula (BMAViC) Model

Predicting Crop Yield Using Deep Learning and Remote Sensing

County-scale crop yield prediction by integrating crop simulation with machine learning models

Contact Info

Product

Resources

About