Spatio-temporal analysis of vegetation dynamics of New Delhi (India) using satellite data
Loading...
Date
2022-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
G.B. Pant University of Agriculture and Technology, Pantnagar, District Udham Singh Nagar, Uttarakhand. PIN - 263145
Abstract
Understanding and analysing vegetation cover changes is crucial for a number of reasons, especially when it comes to taking the necessary conservation measures. This study asses the vegetation changes in the New Delhi (India) over the years from 2000 to march 2022 based on NDVI (Normalized Difference Vegetation Index). The NDVI values have been collected from MODIS terra satellite imagery. Using this NDVI data the study finds that the vegetation greenness of Delhi has increased by 18.63% from year 2001 to 2021. A dataset of 509 NDVI values have been used for making the time series. An attempt has been carried out to predict the vegetation change using this MODIS NDVI time series data and LSTM (Long Short Term Memory) network. The prediction has been carried out on two different LSTM models side by side on the same data and comparative study has been done. The LSTM networks has been trained with 80% of the data and rest 20% are used for testing the model’s accuracy. The results show that both the LSTM model are capable of predicting the future NDVI values with appreciable accuracy but model-1 predicts with better accuracy and lesser errors. Model-1 predicts the future NDVI values with RMSE less than 0.034 and R2 of more than 0.77. Model-2 is not far behind, it predicts with RMSE of around 0.036 and R2 of around 0.74. So, this study concludes that using LSTM networks it is possible to accurately predict vegetation changes well in advance and take appropriate proactive measures to protect and enhance the vegetation in any area.