Landsat NDVI time series reconstruction

Local Knowledge Graph (Top 24 entities)

Loading graph...

Knowledge graph centered on Landsat NDVI time series reconstruction with 24 nodes and 59 connections. Top connected: Engelmann Spruce, Pekania pennanti, Pinus ponderosa, Ponderosa pine, A deep learning hybrid predictive modeling (HPM) a.

Method synopsis

Method to reconstruct continuous daily NDVI time series from sparse Landsat satellite observations using LSTM neural networks with cloud masking and gap-filling.

Synthesized from method descriptions across 1 paper using this protocol.

Procedure from a recent peer-reviewed implementation

Steps below were extracted from the most recent peer-reviewed implementation of this protocol in the corpus — A deep learning hybrid predictive modeling (HPM) approach for estimating evapotranspiration and ecosystem respiration (2021), Hydrology and Earth Systems Science. Implementations in other papers (listed below) may differ.

NDVI time series reconstruction
Calculated daily NDVI values from Landsat 5, 7, and 8 images using red and near-infrared bands. Applied cloud-scoring algorithm in Google Earth Engine to mask clouds with threshold score below 20. Used LSTM neural network to reconstruct continuous daily NDVI time series from sparse satellite observations, achieving training and validation adjusted R² of 0.65 for both ER-BT and ER-SP sites.
Quantities: Daily NDVI reconstruction over 16-day Landsat return period, cloud score threshold of 20, adjusted R² of 0.65Duration: 1995-2017 for SNOTEL sites, 1999-2018 for East River Watershed analysisConditions: Cloud masking applied to remove atmospheric interference, temporal gaps filled using LSTM interpolation
Equipment: Landsat 5 satellite, Landsat 7 satellite, Landsat 8 satellite, Google Earth Engine cloud computing platform