Why did the Normalized Difference Built-Up Index (NDBI) values for a tropical urban area decrease from -0.03 in 1988 to -0.16 in 2020 when using Landsat 5 and Landsat 8 data? Given the expected urban growth, why do the NDBI values also differ (-0.82 to 0.62 in 1988 and -0.45 to 0.35 in 2020)? What factors could explain these changes, and how should we consider these differences when comparing data from different satellites and time periods?
Anitha,
Perhaps it is a calibration issue between the two sensors and/or a radiometric correction issue (atmospheric corrections). Even though the data are corrected for these (C2L2 data set that is), there might be enough error in the corrections that it is not accurate enough for this application for your type of landscape. You should consider including Landsat 5 data a few years later and Landsat 8 a few years sooner in your analyses. See if when you use only Landsat 5 or only Landsat 8 the trend is more what you would were expecting.
Also, perhaps you are using data when the climate conditions were very different (drought/dry vs wet / very wet years / periods). When the NIR band is used for any type of temporal change analyses for landscapes with a fair amount of vegetation it becomes a more difficult problem.
Pat
Pat S. Chavez, Jr.
Dear Pat,
Thank you for your insightful response to my question on ResearchGate regarding the observed changes in Normalized Difference Built-Up Index (NDBI) values over time.
Considering the potential issues you highlighted, particularly those related to calibration, radiometric correction, and the influence of climate conditions on NDBI values, I am seeking clarification on the appropriateness of using the provided data for my research paper. My specific focus involves utilizing NDBI for correlation work, and I have already conducted Land Use Land Cover (LULC) analysis using Maximum Likelihood Classification (MLC), along with the calculation of built-up areas.
Given the mentioned NDBI value issues, could you please advise if it is acceptable to use this dataset for my research paper, especially in the context of correlation analysis? If there are any precautions or additional considerations I should be mindful of, I would greatly appreciate your guidance.
Thank you for your time and assistance.
Anitha
You can always try other indices. Here is a paper that might help you started, Evaluation of spectral built-up indices for impervious surface extraction using Sentinel-2A MSI imageries: A case of Addis Ababa city, Ethiopia. The authors claim that indices using the SWIR bands are better in discriminating urban areas from other land covers. Specifically, the indices NBAI and NBI.
I know you don't want to perform classification but they might help you to show the increase in the urban area.
A couple of things to add:
1. Yes, other indices can/should be examined, however, they may have similar issues related to radiometric calibration between sensors if you want to go back to the 1980s and/or atmospheric corrections problems within the same sensor but at different times.
2. If a NIR band is used in the index to be analyzed it will have the same concerns related to moderate to high vegetation cover landscapes and climate conditions.
3. I am not sure that a non-tropical vegetated urban area can be compared with a tropical vegetated area because of the NIR issue; that is, an index that works for a non-tropical area may not for a tropical area due to the sensitive of the NIR band.
4. Since the spatial / textural patterns of urban areas vs non-urban areas is typically very different it might be worth investigating change in the spatial characteristics (high pass filter at local to moderate scales about a pixel) about a pixel rather than change in it spectral signature.
That's correct. The paper I posted shows other alternatives to built-up indices. Many of them contain the NIR band but some of them don't.
As you said, its worth investigating the indices that do not contain the NIR band (have a look at those who are who are using the SWIR)
Nikolaos Tziokas
Dear Nikolaos Tziokas,
I wanted to express my sincere gratitude for your valuable insights and recommendations. Your suggestion to explore the paper on the evaluation of spectral built-up indices for impervious surface extraction using Sentinel-2A MSI imageries in Addis Ababa has been immensely helpful. I have downloaded the paper and am in the process of thoroughly reviewing it.
I am particularly intrigued by the mention of the SWIR band and the indices NBAI and BRBA, which do not rely on the NIR band. Your guidance aligns perfectly with my objective to investigate indices without the NIR band for NDBI analysis. I appreciate your willingness to share knowledge and point me in the right direction.
Thank you once again for your assistance. I am confident that the information from the paper will significantly contribute to my research and analysis.
Anitha
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers