Monthly Reconstructed Texas AOD (95% threshold)

This is the polished aggregated AOD quarterly composites for the state of Texas, USA from Jan 2010 to Aug 2022, using spatio-temporal LSTM convolutional autoencoder model. Given that AOD data can be influenced by short-term events like heavy rain or thunderstorm, we generated “smoothly aggregated” data composites to mitigate the impact of temporary disturbances or anomalous measurements, allowing for a more accurate characterization of the average AOD levels. Specifically, within a month, the daily data tails beyond the 2nd standard deviation ranges were excluded. This means that data points falling outside the 95% distribution were not considered in the monthly averaging process.