Benchmarking the State of AI for Crop Yield Forecasting: A Global Assessment Across Modeling Paradigms
Authors: Michiel Kallenberg, Philip Janz, Christoph Jörges, Vageesh Saxena, Pratishtha Poudel, Mohammed Musthafa Rafi, Ioannis N. Athanasiadis
Organizations: Artificial Intelligence Group, Wageningen University and Research, Wageningen, the Netherlands · Earth Observation for Agroecosystems, Agroscope, Zurich, Switzerland · Potsdam Institute for Climate Impact Research (PIK), Germany · Department of Geography, Ludwig-Maximilians-Universität Munich, Munich, Germany · Department of Advanced Computing Sciences, Maastricht University, Maastricht, the Netherlands · Department of Agronomy, Purdue University, West Lafayette, Indiana, USA · Department of Computer Science, Iowa State University, Ames, Iowa, USA
Accurate crop yield forecasting is essential for food security, agricultural policy, and commodity markets. Although statistical, process-based, and machine learning approaches have all been proposed, their relative strengths remain poorly understood due to the lack of large-scale, standardized comparisons. We address this gap through a comprehensive benchmark of five modeling paradigms -- statistical baselines, process-based models, conventional feature-engineered machine learning, deep sequence models, and tabular foundation models -- using the CY-Bench benchmark spanning 63 country-crop datasets (>12,000 administrative regions) across six continents. Beyond conventional pooled accuracy metrics, we introduce a multi-dimensional evaluation framework that separately quantifies overall, spatial, temporal, and anomaly prediction skill. Across multiple crops and countries, feature-engineered machine learning and tabular foundation models achieve the strongest overall predictive performance. Data-driven models reduce NRMSE in 72% of countries growing maize and 78% of countries growing wheat, with median reductions of 8.8% and 11.9%, respectively. Across all modeling paradigms, spatial prediction is substantially easier than temporal and anomaly prediction. While data-driven models markedly improve temporal and anomaly prediction over statistical baselines, accurately forecasting year-to-year yield variability remains challenging. These findings suggest that future progress in large-scale crop yield forecasting will depend on the improved characterization of the environmental and management factors driving inter-annual yield variability, founded on effective representations for crop yield forecasting systems that support agricultural decision-making, early-warning applications, and food security planning
Figures & tables
Figure 1. Overview of the benchmark. CY-Bench provides harmonized sub-national crop yield observations and environmental predictors for benchmarking statistical, process-based, tabular, and deep sequence models under a common experimental protocol. Performance is evaluated using four complementary dimensions: overall, spatial, temporal, and anomaly prediction. Overview of workflow.
Overall
Spatial
Temporal
Anomaly
Family
Model
NRMSE ↓
R2↑
r↑
r↑
r↑
Maize (36 countries; 8,211 regions)
Baseline
Trend
25.4 [19.7, 33.7]
0.18 [-0.33, 0.54]
0.71 [0.58, 0.82]
-0.27 [-0.53, 0.07]
-0.10 [-0.17, 0.02]
Process-based
LPJmL
34.1 † [23.5, 50.5]
-0.31 † [-1.05, -0.05]
0.49 † [0.37, 0.65]
0.22 ∗ [0.04, 0.49]
0.08 ∗ [-0.01, 0.28]
Feature ML
RF
24.8 [17.9, 35.3]
0.25 [-0.24, 0.52]
0.72 [0.46, 0.82]
0.27 ∗ [-0.09, 0.50]
0.16 ∗ [-0.02, 0.25]
Foundation
TabICL
24.2 ∗ [17.0, 33.8]
0.25 ∗ [-0.13, 0.52]
0.73 [0.56, 0.83]
0.33 ∗ [0.07, 0.61]
0.14 ∗ [0.04, 0.33]
Table 1. Performance of representative models for each model family. Values are reported as median [IQR] across countries.
Figure 2. Relative reduction in NRMSE of the best data-driven paradigm compared with the best traditional approach. Positive values indicate improved predictive performance through data-driven modeling. Relative reduction in NRMSE of the best data-driven paradigm (feature-engineered, foundation, or sequence) compared with the best traditional approach (statistical or process-based). Positive values indicate improved predictive performance through data-driven modeling. Results are shown separately for (a) maize and (b) wheat.
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Overall
Spatial
Temporal
Anomaly
Family
Model
Δ NRMSE
ΔR2
Δr
Δr
Δr
Maize (36 countries; 8,211 regions)
Baseline
Trend
0.0 [0.0, 0.0]
0.00 [0.00, 0.00]
0.00 [0.00, 0.00]
0.00 [0.00, 0.00]
0.00 [0.00, 0.00]
Process-based
LPJmL
N/A
N/A
N/A
N/A
N/A
Feature ML
RF
0.3 [-0.7, 1.3]
0.01 [-0.03, 0.09]
0.00 [-0.04, 0.02]
0.06 [-0.10, 0.16]
0.01 [-0.05, 0.07]
Foundation
TabICL
0.4 ∗ [-1.4, 2.4]
0.03 ∗ [-0.07, 0.10]
0.03 [0.00, 0.07]
0.12 ∗ [-0.04, 0.26]
0.03 [-0.03, 0.15]
Appendix
Table 2. Difference between mid-season ( ∼ 50% of the growing season observed) and end-of-season ( ∼ 100%) performance for the representative model of each family. Values denote the median Δ [IQR] across countries. Positive Δ indicates improved end-of-season performance.
Reliable agricultural yield statistics are typically reported at coarse administrative scales, whereas modern geospatial machine learning methods require spatially explicit, pixel level supervision. This mismatch has limited the development of large-scale benchmarks for crop yield learning using multimodal Earth observation data. A reproducible benchmark, AgroBench, is presented for transforming publicly available U.S. county level crop yield statistics into weakly supervised pixel-level crop time series. Each crop pixel time series is paired with a county-level yield value as a weak supervisory signal rather than a directly measured pixel-level yield label. Our geospatial data generation pipeline integrates USDA crop yield statistics with crop-specific land cover masks, Sentinel 2 multispectral imagery, Sentinel-1 synthetic aperture radar observations, climatic variables, and terrain information to produce temporally aligned multimodal sequences describing individual crop pixels throughout the growing season. The resulting benchmark contains over 13 million observations from 788,654 unique crop pixels spanning 5,107 county year combinations across eight growing seasons (2017 to 2024) for five major U.S. crops. To facilitate standardized evaluation, we establish a crop yield prediction benchmark using a Leave-One-Year-Out evaluation protocol and provide baseline results using representative machine learning models. By releasing the complete data generation pipeline, benchmark dataset, and evaluation protocol, AgroBench provides a reproducible foundation for future research in weakly supervised learning, multimodal remote sensing, spatiotemporal modeling, and geospatial foundation models for agriculture.
This paper proposes a new method for crop yield prediction, which is essential for developing management strategies, informing insurance assessments, and ensuring long-term food security. Although existing data-driven approaches have shown promise in this domain, their performance often degrades when applied across large geographic regions and long time periods. This limitation arises from two key challenges: (1) difficulty in jointly capturing short-term and long-term temporal patterns, and (2) inability to effectively accommodate spatial data variability in agricultural systems. Ignoring these issues often leads to unreliable predictions for specific regions or years, which ultimately affects policy decisions and resource allocation. In this paper, we propose a new predictive framework to address these challenges. First, we introduce a new backbone model architecture that captures both short-term daily-scale crop growth dynamics and long-term dependencies across years. To further improve generalization across diverse spatial regions, we augment this model with a retrieval-based adaptation strategy. Recognizing the substantial yield variation across years, we design a novel retrieval-and-refinement pipeline that adjusts retrieved samples by removing cross-year bias not explained by input features. Our experiments on real-world county-level corn yield data over 630 counties in the US demonstrate that our method consistently outperforms different types of baselines. The results also verify the effectiveness of the retrieval-based augmentation method in improving model robustness under spatial heterogeneity.
Yiming Sun, Qi Cheng, Licheng Liu +3
University of Pittsburgh, Pittsburgh, PA, USA · University of Wisconsin-Madison, Madison, WI, USA · University of Alabama, Tuscaloosa, AL, USA +1
The impact of climate variability on food production has led to the creation of various forecasting models that uses machine learning (ML), numerical weather predictors (NWP) or a hybrid of ML-NWP models to identify structural and physical relationships between meteorological drivers and crop growth, in order to predict crop yield. Droughts, for example the 2012 Midwestern US (Corn Belt) drought, are extreme events that affect crop production and test the limits of these forecasting models. Using 16 meteorological drivers as predictors, we compare ML (non-deep learning) and deep learning forecasting models to predict the county-level corn yield for the extreme drought year, 2012. This forecasting problem is characterized by a dissimilarity between the feature distributions of the training and test data, where the meteorological conditions of the extreme drought year fall outside the range of historically observed values. Additionally, the dataset consists of spatial and temporal irregularities where counties with missing yields introduce spatial sparsity and the use of only a subset of daily values per year introduce temporal sparsity. To overcome this, we use sample weighting and feature selection as modifications to improve our forecasting models. These modifications lead to an improvement for ML models; however, the deep learning model VITA shows little to no improvement. While VITA outperforms the ML models with or without modifications, our current study sheds light on the effect of dissimilarity between train and test feature distributions on forecasting models, compares deep learning versus non-deep learning models, and introduces modifications that are effective for non-deep learning models.