Machine learning for optimizing complex site-specific management

by Yuji Saikai, Vivak Patel, and Paul D. Mitchell

• APSIM 7.10 is assumed to be installed on Windows (see oracle_m.py or oracle_h.py for more information)
  • Python specifies a simulation by writing .apsim file, calls APSIM with it, and receives a result (.out file)
• APSIM files are constructed in advanced for each site and stored in oracle_m and oracle_h directories
• Run the following on Windows:
  • medium.py for simulations with medium complexity
  • high.py for simulations with high complexity respectively

 

Abstract

Despite the promise of precision agriculture for increasing the productivity by implementing site-specific management, farmers remain skeptical and its utilization rate is lower than expected. A major cause is a lack of concrete approaches to higher profitability. When involving many variables in both controlled management and monitored environment, optimal site-specific management for such high-dimensional cropping systems is considerably more complex than the traditional low-dimensional cases widely studied in the existing literature, calling for a paradigm shift in optimization of site-specific management. We develop a machine learning algorithm that enables farmers to efficiently learn their own site-specific management through on-farm experiments. We test its performance in two simulated scenarios---one of medium complexity with 150 management variables and one of high complexity with 864 management variables. Results show that, relative to uniform management, site-specific management learned from 5-year experiments generates $43/ha higher profits with 25 kg/ha less nitrogen fertilizer in the first scenario and $40/ha higher profits with 55 kg/ha less nitrogen fertilizer in the second scenario. Thus, complex site-specific management can be learned efficiently and be more profitable and environmentally sustainable than uniform management.

[preprint], [published]

 

Learned site-specific management