ISRO - Time-Series Geospatial Data Modeling
Time-series analysis and modeling of geospatial sensor data from satellite observations.
Project Snapshot
Role
Data Engineer Research Intern
Organization
ISRO - National Remote Sensing Centre (NRSC), Hyderabad
Timeline
February – June 2025
Status
Research / POC
Deployment
POC / Research
Tech Stack
Impact & Results
- Time-series analysis on geospatial satellite sensor data
A research-focused proof-of-concept at ISRO NRSC applying time-series modeling techniques to geospatial data streams from satellite observations.
The Problem
Detecting trends, identifying anomalies, and analyzing temporal patterns in satellite sensor data requires robust time-series analysis tools capable of handling large-scale geospatial sequences.
The Solution
A data modeling pipeline that analyzes sequences of satellite observations to uncover temporal patterns and anomalies.
Architecture & Implementation
Modeling: Applied time-series modeling techniques to sequences of geospatial sensor data.
Analysis: Focused on trend detection and anomaly identification across temporal observation sequences.
Data Used: Geospatial sensor data streams from satellite observations.
Tech Stack: Python, Pandas, NumPy, PyTorch.
Challenges & Solutions
Irregular Data: Dealing with irregular sampling intervals due to varying satellite orbit paths and cloud cover.
Noise: Filtering out noisy sensor data and atmospheric interference.
Sequence Length: Managing long temporal sequences for trend detection without vanishing gradients.
Key Takeaways
Time-series sequence modeling applied to geospatial data.
Utilizing recurrent neural networks (RNNs/LSTMs) for pattern recognition.
Robust anomaly detection algorithms.