SUMMARY:
UCLA researchers in the Department of Geography have developed a robust geospatial trend detection method that works with any combination of spatial aggregations, dimensions and variables, allowing researchers to compare geospatial trends across a variety of resolution levels.
BACKGROUND:
Since the adaptation of satellites that were capable of recording image data of our planet, we have generated a massive quantity of geospatial data. This data has proven useful in everything from the building of accurate road maps to the efficient distribution of fertilizer on agricultural land and much more. With even more data available, it is common for time series data to be collected on an area. Using this data, algorithms can detect trends like deforestation, coast line shifts and rapid urban development. In addition, such data can allow researchers to monitor crop health. One issue is that these trends can be very sensitive to the resolution of the data that is provided. A trend visible at 10m resolution may not follow the trend seen at 20m or 40m resolution, an example of Simpson’s paradox. Thus, there is a clear need for algorithms that are robust to sensitivity caused by only analyzing data at a single resolution.
INNOVATION:
Researchers at UCLA have developed a method that is capable of calculating, diagnosing and visualizing the accuracy of trends in geospatial data which is robust to artifacts caused by analysis only at a single resolution. This method allows users to make informed decisions under circumstances when geospatial data conforms to Simpson’s paradox. Thus, researchers studying various geospatial trends will have a robust tool to compare data across various resolutions that may follow contradictory trends.
POTENTIAL APPLICATIONS:
- Time Series Geospatial data
- Surface water hydrology
- Climate change
- Earth Observation Satellite (EOS)
- Geographic Information System (GIS Software)
ADVANTAGES
- Robust to Resolution Artifacts
- Robust to Simpson’s paradox
DEVELOPMENT-TO-DATE:
A successful demonstration of the method has been completed.