Big data, small explanatory and predictive power: Lessons from random forest modeling of on-farm yield variability and implications for data-driven agronomy

Silva, João Vasco; Heerwaarden, Joost van; Reidsma, P.; Tesfaye, Kindie; Ittersum, M.K. van

doi:10.1016/j.fcr.2023.109063

Cited by 9 publications

(4 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, these data are generally aggregated at large administrative scales, such counties or countries. Experimental or on-farm data are also a valuable source of information [46], because they are more precisely located than the gridded-cell data and provide information regarding farm management However, these data cover restricted spatial area and time period, and are thus not suitable for modeling yields at the continental or global scale. The reanalysis data used in this study combines the advantages to span large scale in space and time while covering a large diversity of yield and climate conditions, as shown in table 1.…”

Section: Discussionmentioning

confidence: 99%

Comparison of methods to aggregate climate data to predict crop yield: an application to soybean

Chen,

Guilpart,

Makowski

2024

Environ. Res. Lett.

View full text Add to dashboard Cite

High-dimensional climate data collected on a daily, monthly or seasonal time step are now commonly used to predict crop yields worldwide with standard statistical models or machine learning models. Since the use of all available individual climate variables generally leads to calculation problems, over-fitting, and over-parameterization, it is necessary to aggregate the climate data used as predictors. However, there is no consensus on the best way to perform this task, and little is known about the impacts of the type of aggregation method used and of the temporal resolution of weather data on model performances. Based on historical data from 1981 to 2016 of soybean yield and climate on 3,447 sites worldwide, this study compares different temporal resolutions (daily, monthly, or seasonal) and dimension reduction techniques (principal component analysis, partial least square regression, and their functional counterparts) to aggregate climate data used as inputs of machine learning and linear regression models predicting yields. Results showed that random forest models outperformed and were less sensitive to climate aggregation methods than linear regressions when predicting soybean yields. With our models, the use of daily climate data did not improve predictive performance compared to monthly data. Models based on principal component analysis or averages of monthly data showed better predictive performance compared to those relying on more sophisticated dimension reduction techniques. By highlighting the high sensitivity of projected impact of climate on crop yields to the temporal resolution and aggregation of climate input data, this study reveals that model performances can be improved by choosing the most appropriate time resolution and aggregation techniques. Practical recommendations are formulated in this article based on our results.

show abstract

Section: Discussionmentioning

confidence: 99%

Comparison of methods to aggregate climate data to predict crop yield: an application to soybean

Chen,

Guilpart,

Makowski

2024

Environ. Res. Lett.

View full text Add to dashboard Cite

show abstract

“…RF is a supervised ensemble-learning algorithm that improves regression combining multiple decision trees to enhance the accuracy of the model and its generalization. Due to its accuracy and ability in finding non-parametric relationships, RF is used in the fields of remote sensing and agronomy for prediction and modeling ( Belgiu and Drăgu, 2016 ; Nayak et al., 2022 ; Silva et al., 2023 ). In this work, RF model implemented in the “ranger” package in RStudio was used ( Wright and Ziegler, 2017 ); to avoid the overfitting of the model, the 10-fold-cross-validation was applied, by using the trainControl function of the package “caret” ( Kuhn, 2008 ).…”

Section: Methodsmentioning

confidence: 99%

Predicting carob tree physiological parameters under different irrigation systems using Random Forest and Planet satellite images

Garofalo,

Giannico,

Lorente

et al. 2024

Front. Plant Sci.

View full text Add to dashboard Cite

IntroductionIn the context of climate change, monitoring the spatial and temporal variability of plant physiological parameters has become increasingly important. Remote spectral imaging and GIS software have shown effectiveness in mapping field variability. Additionally, the application of machine learning techniques, essential for processing large data volumes, has seen a significant rise in agricultural applications. This research was focused on carob tree, a drought-resistant tree crop spread through the Mediterranean basin. The study aimed to develop robust models to predict the net assimilation and stomatal conductance of carob trees and to use these models to analyze seasonal variability and the impact of different irrigation systems.MethodsPlanet satellite images were acquired on the day of field data measurement. The reflectance values of Planet spectral bands were used as predictors to develop the models. The study employed the Random Forest modeling approach, and its performances were compared with that of traditional multiple linear regression.Results and discussionThe findings reveal that Random Forest, utilizing Planet spectral bands as predictors, achieved high accuracy in predicting net assimilation (R² = 0.81) and stomatal conductance (R² = 0.70), with the yellow and red spectral regions being particularly influential. Furthermore, the research indicates no significant difference in intrinsic water use efficiency between the various irrigation systems and rainfed conditions. This work highlighted the potential of combining satellite remote sensing and machine learning in precision agriculture, with the goal of the efficient monitoring of physiological parameters.

show abstract

“…Temporal variability was another important constituent of yield variability explaining 10 -55% of the yield variation in all datasets, highlighting the importance of weather conditions on yield at all scales (Fig. 2.7; see also Ray et al, 2015;Silva et al, 2023). Our method allowed us to compare temporal yield variability with spatial yield variability revealing that temporal yield variability was particularly important for the two large commercial farms, for which most fields were distributed across a relatively small area.…”

Section: Figure 22 Long-term Median Scaled Yield Per Farm (In T Ha -1...mentioning

confidence: 83%

“…Finally, we compared yield variability across spatial scales and years with a random effects model fitted with year and spatial scale as random effects (see also Silva et al, 2023). Random effects models are a particular type of linear mixed models that consider only random effects.…”

Section: Introductionmentioning

confidence: 99%

Exploring variability in yield, resource use efficiency and environmental impact of ware potato production in the Netherlands

Ravensbergen

View full text Add to dashboard Cite

Potato as a crop has three different production purposes: (1) as a ware crop for consumption of fresh potatoes and processed products, (2) as a starch crop with multiple industrial purposes and (3) as a seed crop to produce seed potatoes that will be planted in the next growing season. Potatoes are cultivated across the country, but it depends on the region which type of potato crop is most dominant (CBS, 2022). Starch potatoes are mostly cultivated in the northeast of the Netherlands in an area called the 'Veenkoloniën'. Seed potatoes are mostly cultivated in the northern coastal regions of the Netherlands. Ware potatoes are cultivated across the country, although its dominance is less in the areas where also seed and starch potatoes are cultivated. This thesis focusses on ware potato cultivation as it is the most important production purpose in terms of land area (Fig. 1 .1A), and because ha), (B) gross profit (in € ha -1 ), (C) average crop yield (in t dry matter ha -1 ), (D) average yield gap (in % of potential yield), (E) coefficient of variation of crop yield (in %), expressing spatial yield variability, (F) average applied active ingredients (in kg active ingredients ha -1 ), (G) active ingredients (in % of total use), (H) median nitrogen use efficiency (in kg N kg N -1 ), (I) average nitrogen surplus (in kg ha -1 ) and (J) average water productivity (kg DM mm -1 ha -1 ). Panel A is based on data from CBS (average of 2011 -2020). Panel B is based on data from (van derVoort, 2022).

show abstract

Big data, small explanatory and predictive power: Lessons from random forest modeling of on-farm yield variability and implications for data-driven agronomy

Cited by 9 publications

References 47 publications

Comparison of methods to aggregate climate data to predict crop yield: an application to soybean

Comparison of methods to aggregate climate data to predict crop yield: an application to soybean

Predicting carob tree physiological parameters under different irrigation systems using Random Forest and Planet satellite images

Exploring variability in yield, resource use efficiency and environmental impact of ware potato production in the Netherlands

Contact Info

Product

Resources

About