[Paper Review] Accuracy Requirements for Early Estimation of Crop Production in Senegal

Damien Jacques, Pierre Defourny|arXiv (Cornell University)|Jun 9, 2019

Precipitation Measurement and Analysis41 references4 citations

TL;DR

This study develops a methodological framework to define accuracy requirements for early crop production estimators in Senegal, using 20 years of historical agricultural data. It shows that crop yield variability is the main constraint on pre-harvest forecasting accuracy, while cropland area mapping significantly improves predictions for key crops like groundnuts, millet, and rice.

ABSTRACT

Early warning systems for food security rely on timely and accurate estimations of crop production. Several approaches have been developed to get early estimations of area and yield, the two components of crop production. The most common methods, based on Earth observation data, are image classification for crop area and correlation with vegetation index for crop yield. Regardless of the approach used, early estimators of cropland area, crop area or crop yield should have an accuracy providing lower production error than existing historical crop statistics. The objective of this study is to develop a methodological framework to define the accuracy requirements for early estimators of cropland area, crop area and crop yield in Senegal. These requirements are made according to (i) the inter-annual variability and the trend of historical data, (ii) the calendar of official statistics data collection, and (iii) the time at which early estimations of cropland area, crop area and crop yield can theoretically be available. This framework is applied to the seven main crops in Senegal using 20 years of crop production data. Results show that the inter-annual variability of crop yield is the main factor limiting the accuracy of pre-harvest production forecast. Estimators of cropland area can be used to improve production prediction of groundnuts, millet and rice, the three main crops in Senegal stressing the value of cropland mapping for food security. While applied to Senegal, this study could easily be reproduced in any country where reliable agricultural statistics are available.

Motivation & Objective

• To establish accuracy requirements for early estimators of cropland area, crop area, and crop yield in Senegal.
• To account for inter-annual variability and trends in historical crop production data.
• To align estimator accuracy with the timing of official statistical data collection in Senegal.
• To evaluate whether early Earth observation-based estimators can outperform simple historical averages in pre-harvest forecasting.
• To assess the value of cropland mapping for improving food security early warning systems.

Proposed method

• Analyzes 20 years of official crop production statistics for seven main crops in Senegal.
• Uses inter-annual variability and trend analysis of historical yield and area data to define minimum accuracy thresholds.
• Integrates the calendar of official data collection (crop area by end of September, yield by end of November) as a temporal constraint.
• Applies a framework to determine the required accuracy of early estimators such that their error is lower than that of historical averages.
• Evaluates the potential of Earth observation-based estimators for cropland and crop area, and vegetation index-based yield estimates.
• Assesses the theoretical earliest possible time for accurate pre-harvest estimations based on seasonal phenology and data availability.

Experimental results

Research questions

• RQ1 What accuracy is required of early estimators of cropland area, crop area, and crop yield to outperform historical averages in Senegal?
• RQ2 How does inter-annual variability in crop yield affect the feasibility of accurate pre-harvest production forecasts?
• RQ3 At what point in the growing season can reliable early estimates of crop area and yield be achieved?
• RQ4 To what extent can cropland area mapping improve early production forecasts for major crops in Senegal?
• RQ5 How do the timing of official statistics collection and the seasonal dynamics of vegetation affect early warning accuracy?

Key findings

• Inter-annual variability in crop yield is the primary factor limiting the accuracy of pre-harvest production forecasts in Senegal.
• Even with perfect crop area estimation, the lowest achievable coefficient of variation of the root mean square error (CV(RMSE))) for pre-harvest yield forecasts ranged from 18% to 40% for major crops.
• Cropland area estimators significantly improve production forecasts for groundnuts, millet, and rice, highlighting the value of cropland mapping in food security early warning.
• For rice, the yield trend alone provides a relatively stable baseline forecast, but for other crops, high variability makes early quantitative estimation challenging.
• Early estimation of crop yield using Earth observation data is unlikely to be accurate before mid-September, due to the timing of peak vegetation growth.
• While quantitative early forecasts may be limited by high variability, Earth observation data remain valuable for qualitative anomaly detection and geospatial monitoring.