Monthly vegetation drought forecasting using copula functions, numerical weather prediction and artificial intelligence models

Drought inhibits vegetation growth, triggers wildfires, reduces agricultural production and has a significant impact on the health of terrestrial ecosystems. Continuously monitoring and forecasting the effects of drought on vegetation health can provide effective information for ecosystem management. The purpose of this study is to forecast the effect of meteorological drought on vegetation, that is, the ecological drought of vegetation. Because vegetation drought is a complex phenomenon, it should be approached based on the probabilistic relationship between meteorological drought and vegetation. Accordingly, a probabilistic approach was constructed to model the bivariate joint probability distribution between meteorological drought and vegetation using the copula function. In order to predict ecological drought based on the joint probability distribution, predictive information on meteorological drought and vegetation health is required. To this end, a meteorological drought was predicted using numerical weather prediction, and a short-term vegetation prediction model considering the meteorological drought prediction results was developed. The vegetation prediction model combining Convolutional Long Short-Term Memory and Random Forest was able to improve the prediction performance of vegetation by considering spatial and temporal patterns. The vegetation drought was forecast by linking the prediction information of vegetation and meteorological drought with the joint probability distribution. The approach of this study will be able to provide useful information to respond to the drought risk in terms of ecology.

 

Acknowledgement

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. NRF-2022R1A2B5B01001750).