I am using Google Earth Engine to compute NDWI time series, in particular by plotting "Landsat 7 Collection 1 Tier 1 8-Day NDWI Composite" data values
The product description says: These Landsat 7 Collection 1 Tier 1 composites are made from Tier 1 orthorectified scenes, using the computed top-of-atmosphere (TOA) reflectance. It is derived from the Near-IR band and a second IR band, ≈1.24μm when available and the nearest available IR band otherwise.
This seems strange to me, since Landsat 7 has no 1.24μm band, but only a SWIR band at 1.55-1.75 μm https://landsat.usgs.gov/what-are-band-designations-landsat-satellites
How does the Google algorithm calculates it?
I think it used the algorithm that proposed by Gao (1996):
NDWI=(Xnir-Xswir)/(Xnir+Xswir)
Hi Muhammad Ramdhan .
I find it quite difficult, since since Landsat 7 has no 1.24μm band ....
L7 does not has this wavelength Band (1,24) as you said, thus, I think that It Will be used the nearest available, the band 5 - Shortwave Infrared (SWIR) 1. So, i think the GEE is calculating right anyway.
This confirms my guess. But it is not Gao Version. Gao used 0.86-μm and the 1.24-μm wavelenght, but considering ETM+ the computation occurs with 0.85 and 1.65. It should be labeled as NDII - Normalized Difference Infrared Index. See here for more info:https://www.hydrol-earth-syst-sci.net/20/3361/2016/
See the Jensen (2009) in chapter eleven. There, it shows how to calculate the NDWI with landsat 5 utilizing the band 4 ( 0.77 - 0.90 ) and band 5 (1.55 - 1.75 ). See the figure attached!
Dear Roberto, thanks a lot.
I considered as follows:
The original paper on NDWI is from GAO, where NDWI is calculated as you can find here
Article NDWI?A Normalized Difference Water Index for Remote Sensing ...
Here NDWI is defined as (r_0.85 - r_1.24)/(r_0.85 + r_1.24)
If I considers the formula you attached I obtain:
(r_0.85 - r_1.65)/(r_0.85 + r_1.65)
This is the formula of another index: NDII, as reported on this reference on HESS:
https://www.hydrol-earth-syst-sci.net/20/3361/2016/
So I presume that with Landsat 7 ETM+ sensor, it is possible to compute NDII, even if in some references this is called NDWI.
Giulio Castelli, Thank you for putting this question in front of the research community.
I found one interesting fact while searching the NDWI functionalities on Google Earth Engine and got an answer on this link- https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LT05_C01_T1_8DAY_NDWI
Where, it is written that "It is derived from the Near-IR band and a second IR band, ≈1.24μm when available and the nearest available IR band otherwise."
Hence, the calculation is obviously happening by using two NIR bands only for L5/7, which are of course not following the exact bands/frequency as proposed by Gao (1986) but near to that frequency only.
The confusion here derives from the fact that there are two definition of NDWI, and both from 1996!
One is by Gao (1996) where NDWI is defined as the normalized difference between NIR and SWIR. This is used to monitor changes in water content of leaves in plants.
Another one is by McFeeters (1996) and is defined as the normalized difference between GREEN and NIR. This is used to monitor changes in the water bodies 0n the landscape.
The latter can be used with L4 to L8, the former not.
References:
Gao B. 1996. NDWI—A normalized difference water index for remote sensing of vegetation liquid water from space.R.S. of the Env. 58(3): 257-266
Article NDWI?A Normalized Difference Water Index for Remote Sensing ...
McFeeters S.K., 1996. The use of Normalized Difference Water Index (NDWI) in the delineation of open water features.
Int. J. of Remote Sensing, 17(7):1425-1432
Article The Use of Normalized Difference Water Index (NDWI) in the D...
You can use this code
// calculate ndwi from landsat 5 & 7
function NDWI(img) {
var ndwi = img.normalizedDifference(['B2', 'B4']).rename('NDWI');
return img.addBands(ndwi);
}
Can anyone tell me the exact value to use to differentiate waterbodies from other land features in NDWI obtained from GEE?
var dataset = ee.ImageCollection('LANDSAT/LT05/C01/T1_8DAY_NDWI')
.filterDate('2011-01-01', '2011-12-31');
var colorized = dataset.select('NDWI');
var colorizedVis = {
min: 0.0,
max: 1.0,
palette: ['0000ff', '00ffff', 'ffff00', 'ff0000', 'ffffff'],
};
Map.setCenter(6.746, 46.529, 6);
Map.addLayer(colorized, colorizedVis, 'Colorized');
Nigomananda Biswas I think it would be helpful...
You may change the code according to your specifications...
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing