I use Landsat 8 images. In my study area I have a high presence of clouds, but in some scenes the density of the clouds is low (~20%) and allows me to visualize my plots. Is it correct to use images that completely cover the study area, but its density is low?
Franz Pucha-Cofrep, Presence of clouds (whether optically thick and dense clouds or thin and wispy cirrus clouds) are tricky and do affect the NDVI values in all optical imagery. Look at the attached link. We are fortunate to have a Cirrus band (Band 9) in Landsat 8 to detect thin, high altitude clouds which are sometimes not even seen in the visible bands. Previous Landsat sensors did not have this band.
In my opinion, whether you mask out clouds from Landsat 8 while performing NDVI analysis really depends on your research application: whether you are only doing a qualitative analysis and need it just for visualization purpose or whether you want to do a detailed quantitative research. If the aim is latter, I would definitely mask out the clouds before moving ahead. USGS has been providing a Pixel QA band with each Landsat 8 scene, which you can use to identify pixels that are cloud 'contaminated' (among others) and subsequently mask them out. I would suggest you to begin with the pixel_qa band first and then decide.
https://www.mdpi.com/2072-4292/9/8/834/htm
https://landsat.usgs.gov/landsat-surface-reflectance-quality-assessment
Hello @Franz,
Yes, it affects the values of NDVI, and it may not give the desired results sometimes. Clouds and snow tend to be rather bright in the red (as well as other visible wavelengths) and quite dark in the near-infrared which implies based on mathematical equation of NDVI applied on dense forest. It can be seen from its mathematical definition that the NDVI of an area containing a dense vegetation canopy will tend to positive values (0.3 to 0.8) while clouds and snow fields will be characterized by negative values of this index.
Satellite studies taken over multiple days can be utilized to eliminate the effects of cloud coverage on an area, as long as there is not cloud coverage for the entire observation period.
The clouds effect on NDVI is then eliminated by selecting for each pixel the greatest value on any day during the 14-day period (the highest NDVI value is assumed to represent the maximum vegetation "greenness" during the period). This process eliminates clouds from the composite (except in areas that are cloudy for all 14 days).
I hope this helps,
Kind regards
I agree with Dr. James Varghese and Dr. Adil Salhi.
Definitely, if you want to compare an image without cloud cover and another with cloud cover, the values are going to be different. The easiest way is to find images without cloud cover. If you can't try what is recommended above.
I hope this will help you
I would simply look for images without cloud cover, however, if you could not find one, you should mask out the cloud. Fmask add-on on QGIS could be of help.
Yes it affects the calculation of NDVI. As stated in previous replies, try and get an image with no cloud clover over your area of study or mask out the cloud.
Dear Franz,
base on answer Salhi and james not clear atmospher decrease NDVI and GNDVI and EVI and other vegetation indices
You can eliminate affected pixel via quality band in Landsat8 and Landsat7 L1TP products.
The only thing you have to do checking the corresponding pixel value in Landsat BQA file(it is the auxilary file in the unzipped Landsat collection1 L1TP format that,s name teminated with BQA).
for clear Sky (not cloud not snow not sirus not shade) the BQA value will be
2720,2724,2728,2732 for Landsat8
672,676,680,684 for Landsat7
for more detail read this link:
https://landsat.usgs.gov/collectionqualityband
for example 2 landsat8 images atthached
one is the visual RGB image affected via not clear sky
and other image is the NDVI eliminated image
thanks regard,
it is indisputable, NDVI values is affected by clouds presence over studied region. it is better to get another scene image for the region without them.
Sure, clouds influence any index values, as they disperse, reflect different wavelengths of solar radiation various ways. The characher of these changes depends on cloud type and density. Moreover, clouds produce heterogeneity in the surrounding atmosphere. Thus, the impact of cloud cover on index values (simply on certain pixel values in each band) is mostly unpredictable. Thus, the developed automatic cloud masking algorithms always exclude from the account the territories near clouds along with their shadows.
The imagery with the transparent clouds may be sometimes applicable just for detecting the boundaries (not a quantitative analysis) of some square objects like, say, water bodies. Although even in that case the application of cloudy images is very restricted, and special locally adjusted algorithms should be used.
Yes it does, always use imagery less than 10% cloud cover for all your analyses.
And to be honest with you... you don't solve the problem flying under the clouds :) its better, but... depending on the cloud pattern dealing with quality date for quantitative purposes it's still a challenge...
I agree that clouds and snow should be masked out from NDVI calculations but what about shadows? Do those affect the NDVI values significantly? Or can we neglect the effects of shadows altogether?
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing