Abstract:<p>Climate sensitivity of vegetation has long been explored using statistical or process‐based models. However, great uncertainties still remain due to the methodologies' deficiency in capturing the complex interactions between climate and vegetation. Here, we developed global gridded climate–vegetation models based on long short‐term memory (LSTM) network, which is a powerful deep‐learning algorithm for long‐time series modeling, to achieve accurate vegetation monitoring and investigate the complex relationship between climate and vegetation. We selected the normalized difference vegetation index (NDVI) that represents vegetation greenness as model outputs. The climate data (monthly temperature and precipitation) were used as inputs. We trained the networks with data from 1982 to 2003, and the data from 2004 to 2015 were used to validate the models. Error analysis and sensitivity analysis were performed to assess the model errors and investigate the sensitivity of global vegetation to climate change. Results show that models based on deep learning are very effective in simulating and predicting the vegetation greenness dynamics. For models training, the root mean square error (RMSE) is <0.01. Model validation also assure the accuracy of our models. Furthermore, sensitivity analysis of models revealed a spatial pattern of global vegetation to climate, which provides us a new way to investigate the climate sensitivity of vegetation. Our study suggests that it is a good way to integrate deep‐learning method to monitor the vegetation change under global change. In the future, we can explore more complex climatic and ecological systems with deep learning and coupling with certain physical process to better understand the nature.</p>

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to model vegetation greenness and its sensitivity to climate through deep - learning techniques, especially the Long - Short - Term Memory (LSTM) network. Specifically, the research aims to: 1. **Improve the accuracy of vegetation dynamic monitoring**: Traditional statistical models and ecological - process - based models have deficiencies in capturing the complex interactions between climate change and vegetation, resulting in low prediction accuracy. This study uses deep - learning techniques, especially the LSTM network, to improve the simulation and prediction of vegetation dynamics. 2. **Explore the sensitivity of vegetation to climate change**: By establishing a global gridded climate - vegetation model, researchers hope to reveal the response patterns of vegetation in different regions to temperature and precipitation changes, so as to better understand the impact of climate change on ecosystems. 3. **Develop new methods to analyze the vegetation - climate relationship**: Traditional methods have limitations when dealing with complex social - ecological systems. This paper proposes to introduce deep - learning methods into vegetation dynamic modeling to provide a new and more effective tool to study the relationship between vegetation and climate. ### Research Background In the context of global change, terrestrial ecosystems are facing severe challenges. More and more studies have shown that climate change has affected the greenness and distribution patterns of vegetation, and the changes in vegetation will in turn produce feedback on the climate system. Therefore, revealing the complex relationship between climate and vegetation is crucial for better understanding climate feedback, ecosystem health, and sustainable development. However, existing models have limitations in performance and uncertainty, especially when dealing with the long - term dynamics of complex climate - ecosystems. ### Solutions To overcome these limitations, the author adopts a new method, that is, using the LSTM network in deep - learning to model vegetation dynamics. The specific steps include: 1. **Data collection and pre - processing**: Use the GIMMS NDVI 3g data set to represent vegetation greenness, and combine the monthly temperature and precipitation in the CRU TS 4.01 climate data set as input variables. 2. **Model construction and training**: Construct a deep - learning model with two LSTM layers and one fully - connected layer. The input is the time - series data of temperature and precipitation, and the output is the NDVI value. The model is trained with data from 1982 to 2003, and the data from 2004 to 2015 are used for verification. 3. **Sensitivity analysis**: Through Permutation Importance (PI) analysis, evaluate the influence of temperature and precipitation on vegetation greenness, and reveal the sensitivity patterns of global vegetation to climate factors. 4. **Error analysis**: Analyze the relationship between model errors and factors such as inter - annual variation of vegetation, average temperature, and precipitation, and explore the reasons for poor model performance. ### Main Findings 1. **Good model performance**: The RMSE of about 92% of the global regions is less than 0.01, and the CV of 75% of the models is less than 15%, indicating that the deep - learning model has high precision in simulating vegetation dynamics. 2. **Significant spatial differences**: The model performance in the Southern Hemisphere is better than that in the Northern Hemisphere. The model performance in the tropical rainforests and grasslands near the equator is the best, while the model performance in the northern frigid zone is poor. 3. **Sensitivity of vegetation to climate**: 47.96% of terrestrial ecosystems are more sensitive to temperature, especially in the high - latitude regions of the Northern Hemisphere; 18.48% of the regions are more sensitive to precipitation, mainly distributed in savannas and grasslands. Through these findings, the research not only shows the potential of deep - learning in vegetation dynamic modeling, but also provides new ideas for the future monitoring and management of ecosystems.

Modeling vegetation greenness and its climate sensitivity with deep‐learning technology

The sensitivity of vegetation cover to climate change in multiple climatic zones using machine learning algorithms

Spatial–Temporal Prediction of Vegetation Index With Deep Recurrent Neural Networks

Multi-modal learning for geospatial vegetation forecasting

A Novel Approach for Modelling Vegetation Distributions and Analysing Vegetation Sensitivity Through Trait-Climate Relationships in China

Enhanced prediction of vegetation responses to extreme drought using deep learning and Earth observation data

Attribution of Vegetation Dynamics in the Yellow River Water Conservation Area Based on the Deep ConvLSTM Model

Assessing the Sensitivity of Vegetation Cover to Climate Change in the Yarlung Zangbo River Basin Using Machine Learning Algorithms

Deep learning projects future warming-induced vegetation growth changes under SSP scenarios

Learning to forecast vegetation greenness at fine resolution over Africa with ConvLSTMs

Forecasting vegetation dynamics in an open ecosystem by integrating deep learning and environmental variables

Modeling with Hysteresis Better Captures Grassland Growth in Asian Drylands

Projected Climate Change Effects on Global Vegetation Growth: A Machine Learning Approach

Upscaling of Surface Soil Moisture Using a Deep Learning Model with VIIRS RDR

Focal-TSMP: deep learning for vegetation health prediction and agricultural drought assessment from a regional climate simulation

Time-lagged response of vegetation dynamics to climatic and teleconnection factors

Vegetation classification in a subtropical region with Sentinel-2 time series data and deep learning

Geographically Weighted Convolutional Long Short-Term Memory Neural Networks: A Geospatial Deep Learning Model for Monthly NDVI Prediction

Root-zone soil moisture estimation based on remote sensing data and deep learning

A comprehensive study of deep learning for soil moisture prediction

A long short‐term memory‐based model for greenhouse climate prediction