For a change detection analysis covering 30 years on a land area of about 140km (rural and agricultural landscape) , would the use of MODIS 250 m be ideal in the normalization of poor quality Landsat images available in historical years of interest in a bid to create a time series?
The temporal and spatial scale of Landsat makes it very suitable for the analysis but the inconsistency of data over that is worrying nad again MODIS delivers in 250 m upwards.
Or would SOPT images be prefereble?
Many thanks
Kansy,
It would mostly depend on the minimum area on the ground that you want to analyse for change detection. Similarly, it depends if you are fine with 250 m of MODIS if that is the minimum area you want to analyse. However, for normalizing poor quality Landsat images I would prefer SPOT images compared to 250 m MODIS products as it is closer to Landsat spatial resolutions. What spectral resolutions are you considering? You can look at ASTER and other comparable sensors too.
https://www.itc.nl/Pub/sensordb/searchsen.aspx
Dear Aroma and James ,
Many thanks for your replies.
Given that my temporal resolution spans about 30 years starting from the 80s, I chose Landsat due to its spatial resolution, 30m.
Unfortunately the landsat data over that period is very limited between 1984 till 1999 and worse also with images from 2000 when the SLC off event occured. So far I have been able to see 3 good quality cloud-free images between 1989 and 1999. There is a hugge gap in data in these years (
1989
1990
1992
1993
1994
1995
1996
1997
1998
2003
2004
2006
2008
2009
2011
2012
2013
I have had back and forth correspondence with USGS but there is little they can do to assist because during this period, there was no international ground station and no recorder on board LS 4 - 5.
Even though, it is attrctive to use MODIS because there are a few models /algorithms that can harmonize products from hese two sensors, tehre is still a huge challenge because MODIS provides data from year 2000 implying that I have to find data for the years I have listed.
So I think "SPOT sounds like it.
Amy ideas about any model /algorithm that can normakize SPOT products to produce LS -like images ?
Thanks in advance for your useful bits.
Kansy, thanks for clarifying the question. No easy answers it seems. I assume you have had enough brainstorming on why SPOT might be the best alternative while you had correspondence with USGS. Nevertheless, I thought to pose another trivial question: Is it not important then to also consider spectral in addition to spatial and temporal resolutions while you look for alternative sensors to patch up the gap in your poor quality Landsat data? As you stated your aim is to do a change detection analysis in a region with rural and agricultural landscape and I guess you will be using at least the NIR and Red wavelength bands for this purpose. I suppose you have already considered this aspect and SPOT might be the closest match or the best trade off to Landsat considering the temporal, spatial and spectral resolutions. I highly doubt if you will find a better alternative?
I found some similar links like the ones attached by Jenice. You can share your question with some of the authors of the articles such as Dr. Michael Wulder from NRC among others who have extensive experience in similar topics.
http://www.harrisgeospatial.com/portals/0/pdfs/envi/TM_SPOT_Fusion.pdf
Article Generation of dense time series synthetic Landsat data throu...
Article Data fusion of SPOT and LANDSAT images using additive multir...
Landsat 8 makes it very suitable for the analysis ? i work research and no problem
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing