Spectral-Temporal–Spatial Feature Optimization for <i>Dioscorea Polystachya</i> Turczaninow Classification Using Time Series Sentinel-2 Data

<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dioscorea Polystachya</i> Turczaninow is one of the most famous traditional Chinese Materia Medica. However, there is lack of large-scale classification method which is crucial for its growth status monitoring and yield estimation. This study proposed a reliable <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dioscorea Polystachya</i> Turczaninow classification model based on spectral-temporal-spatial feature optimization using time-series Sentinel-2 data. Firstly, 16 mono-temporal classification models were developed using five vegetation indices (VIs) and random forest algorithm. Then, temporal feature optimization was conducted by identifying the most effective time phase combinations based on Sentinel-2 time-series normalized difference vegetation index (NDVI) data, gaussian mixture modeling algorithm and F1 score of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dioscorea Polystachya</i> Turczaninow in each mono-temporal model. Next, spectral features were optimized by replacing NDVI with the optimal VI corresponding to each time phase, thus constructing a multi-VIs-based time-series dataset. Finally, the spatial feature optimization was conducted using the three-dimensional convolutional neural network (3-D CNN) algorithm and the multi-VIs-based time series Sentinel-2 data. Following the comprehensive feature optimization, the final <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dioscorea Polystachya</i> Turczaninow classification model was determined. The results found that Sentinel-2 data acquired during the rhizome enlargement stage played a crucial role in classifying the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dioscorea Polystachya</i> Turczaninow. By using the optimized features, the classification model achieved the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dioscorea Polystachya</i> Turczaninow F1 score of 95.00%, which improved by 11.49% compared to only using the time series NDVI data. This spectral, temporal and spatial feature optimization method also has the potential to the development of large-scale, dynamic, and accurate mapping for other crops.