Impact of Spatial Soil and Climate Input Data Aggregation on Regional Yield Simulations

Area of Science:

• Agricultural science
• Environmental modeling
• Geographic information systems

Background:

• Large-scale crop simulations often use low-resolution, aggregated climate and soil data.
• Data aggregation methods like averaging and area majority sampling can introduce spatial bias.
• This bias can significantly affect the accuracy of simulated crop yields at regional to continental scales.

Purpose of the Study:

• To quantify the error in water-limited crop yields caused by spatially aggregated soil and climate data.
• To evaluate this aggregation error across 14 different crop models.
• To assess the impact of aggregating soil data versus both soil and climate data.

Main Methods:

• Simulated water-limited yields for winter wheat and silage maize using 14 crop models.
• Calculated aggregation error by comparing yields simulated at high resolution (1 km) to those at lower resolutions (up to 100 km).
• Conducted simulations for North Rhine-Westphalia, Germany, focusing on spatial aggregation effects.

Main Results:

• Most models exhibited yield biases of less than 15% when only soil data was aggregated.
• The relative mean absolute error (rMAE) due to aggregated soil data was comparable to or exceeded inter-annual and inter-model yield variability.
• Aggregating both climate and soil data further increased the simulation error, with distinct patterns suggesting predictability from soil variables.

Conclusions:

• Spatial aggregation of input data, especially both soil and climate, introduces significant errors in crop yield simulations.
• The magnitude of this error varies across crop models, highlighting the need for model-specific assessments.
• This study provides a foundational assessment for understanding and potentially mitigating aggregation errors in large-scale agricultural modeling.