I am about to use the BFAST Lite (Article BFAST Lite: A Lightweight Break Detection Method for Time Se...
) algorithm to perform a satellite time-series analysis (STSA). One of the parameters I have to specify is the lenght of the stable period (a time period within the time series data that is assumed to be free from significant structural changes).The data I am about to use are monthly satellite nighttime lights (NTL) for the years 2013 to 2023. From a basic visual inspection, I saw that during 2014, there was a brightening in my study area and during the 2019 a dimming (due to COVID?). The issue is that I wanted to to use the years 2013 to 2018 as a stable period, but now I can't.
If I use them, the algorithm doesn't identify a break in 2019 (I haven't tested the behaviour of the method if a lessen the length of the stable period).
My question is: What are the benefits and drawbacks of a smaller stable period in the BFAST method? Is there a recommended time length? I'm thinking to use data for the period 2018-2023.
Note: BFAST Lite requires data for at least 4 years.
Benefits of Stable Period in BFAST Lite Algorithm
Drawbacks of Stable Period in BFAST Lite Algorithm
- Badges
- Science topic
- Similar topics
- Computer Science
- Research Topics