# Predicting time series of vegetation leaf area index across North America based on climate variables for land surface modeling using attention-enhanced LSTM

# The code enclosed in this repository includes the proposed model stracture associated with manuscript, you can regenerate the training model according to different input data or researches.

# The model file includes the training models associated with manuscript, you can directly use them to predict the time series of vegetation leaf area index with multiyear.

# The Inputs folder holds the csv files for the model inputs. We converted the time series of each pixel into sequential csv files, and then input them into the AELSTM model to obtain the predicted LAI time series. There are a total of 596151 valid pixels in our study area.

# The Results folder holds the AELSTM predictions, including the predicted LAI, the GPP after coupling CoLM 2014 and the predicted LAI under future climate scenarios.


(The input data for the entire study area is too large to be uploaded. Instead, we have uploaded input data for five segments for your reference. Please refer to the manuscript for details on data preprocessing, and feel free to contact us for further information.)