Evaluating the accuracy of ARMA and multi-index methods for predicting winter wheat maturity date

Area of Science:

• Agricultural Science
• Remote Sensing
• Time Series Analysis

Background:

• Predicting winter wheat maturity is vital for agricultural management and preventing yield/quality loss.
• Existing methods require improvement for timely and accurate regional crop assessments.

Purpose of the Study:

• To propose and evaluate novel methods for predicting regional winter wheat maturity dates.
• To compare the accuracy of vegetation index-based models against traditional methods.

Main Methods:

• Utilized an autoregressive moving-average model to predict vegetation indices on key dates (May 1, 9, 17).
• Compared vegetation index methods with growing degree days and a local empirical method.
• Employed leave-one-out cross-validation for performance assessment on Guanzhong Plain winter wheat (2003-2013).

Main Results:

• Vegetation index and growing degree days methods outperformed the local empirical method in maturity prediction accuracy.
• A two-step filtering method using future meteorological data achieved the highest prediction accuracy on May 1.
• This advanced method also demonstrated the lowest error fluctuation on May 17.

Conclusions:

• The proposed methods offer enhanced accuracy for regional crop maturity prediction.
• These findings support improved agricultural harvesting equipment deployment and mitigation of weather-related yield losses.