Extending multi-output methods for long-term aboveground biomass time series forecasting using convolutional neural networks

Authors: NOA-YARASCA, EFRAIN - Texas A&M Agrilife; OSORIO LEYTON, JAVIER - Texas A&M Agrilife; Angerer, Jay

Submitted to: Machine Learning and Knowledge Extraction
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/15/2024
Publication Date: 7/17/2024
Citation: Noa-Yarasca, E., Osorio Leyton, J.M., Angerer, J.P. 2024. Extending multi-output methods for long-term aboveground biomass time series forecasting using convolutional neural networks. Machine Learning and Knowledge Extraction. 6(3): 1633-1652. https://doi.org/10.3390/make6030079.
DOI: https://doi.org/10.3390/make6030079

Interpretive Summary: The ability to accurately predict future rangeland forage conditions can be helpful for livestock decision making in the face of drought and other stocking/destocking conditions. Many existing forecasting tools can only forecast a single future period. This study compared two new methods for predicting future forage conditions (from 1 month to 12 months) with current methods that can predict single and multiple dates in the future. Results indicated that methods predicting single dates were more accurate for short term forecasts. For longer term forecasts, both single and multiple date forecasts were useful. As would be expected, all models were less reliable when predicting forage biomass for distant future dates. However, the two new forecasting methods evaluated in this study had increased reliability in forecasting longer term forecasts (6-12 months).

Technical Abstract: Accurate aboveground vegetation biomass forecasting is essential for livestock management, climate impact assessment, and ecosystem health. While AI techniques have advanced time-series forecasting, a research gap persists in predicting aboveground biomass time series beyond single values. This study introduces RECMO and DirRecMO, two multi-output methods for forecasting aboveground vegetation biomass. Using convolutional neural networks, their efficacy is evaluated across short, medium, and long-term horizons on six Kenyan grassland biomass datasets, comparing them with existing single-output methods (Recursive, Direct, DirRec) and multi-output methods (MIMO, DIRMO). Results indicate the superiority of single-output methods in short-term predictions, while both single-output and multi-output methods exhibit comparable effectiveness in long-term forecasts. RECMO and DirRecMO outperform established multi-output methods, showcasing their potential for biomass forecasting. The study underscores the significant impact of multi-output size on forecast accuracy, highlighting the need for optimal size adjustments and showcasing the proposed methods' flexibility in long-term forecasts. Furthermore, short-term predictions show less significant differences among methods, complicating the identification of the best performer. However, clear distinctions emerge in medium and long-term forecasts, underscoring the greater importance of method choice for long-term predictions. Moreover, as the forecast horizon extends, errors escalate across all methods, reflecting the challenges of predicting distant future periods. This study suggests advancing hybrid models (e.g., RECMO, DirRecMO) to improve extended horizon forecasting. Future research should enhance adaptability, investigate multi-output impact, and conduct comparative studies across diverse domains, datasets, and AI algorithms for robust insights.