Presentation Details
Predictive Agrivoltaics: A Novel Physics-guided Deconvolution Approach Anticipates Multi-year Indiana Commodity Crop Yield Distribution under Single-Axis Tracking AgPV (yes)

Jabir Bin Jahangir, Geoffrey Sanchez, Muhmmad Ashraful Alam, Peter Bermel.

Elmore Family School of Electrical and Computer Engineering, Purdue University, West Lafayette, IN, USA

Abstract


In recent years, agrivoltaics (AgPV), the simultaneous use of land for agriculture and photovoltaics, has emerged as a solution to the inevitable land-use conflicts arising from solar farm expansions. This has led to the development of numerous AgPV test programs worldwide, providing valuable insights into various crops. However, these efforts have often neglected commodity crops, such as corn and soybeans, which account for over 75% of U.S. cropland, largely because of their relatively higher light, land, and equipment requirements. In addition, the farm-level crop yield distribution is critical for risk assessment, as its tails determine potential losses and indemnity payouts in AgPV systems. Here, we develop a model that combines physics-based modeling with statistical inference to predict both PV performance and crop yield distributions in an AgPV farm. The model is validated using an extensive multiyear dataset of corn and soybean grown under single-axis tracking AgPV at Purdue’s ACRE facility in Indiana, USA. We expect this approach to help inform stakeholders and provide much-needed clarity for evidence-based decision-making toward sustainably meeting the world’s food and energy needs.

No part of this publication may be reproduced, distributed, or transmitted in any form or by any means, including photocopying, recording, or other electronic or mechanical methods, without the prior written permission of the author.