Atmospheric Error and Correction

 

• Atmospheric Correction Models and Softwares:

Anyway so for atmospheric correct models and software what are available today. Because everyone will not be developing their own models for atmospheric correction and neither coding things so what are currently what is available. So for most atmospheric correction algorithms is it necessary to solve the radiative transfer equation and establish as just we have seen. Establish look-up tables also we have seen one slide back, for rapid atmospheric corrections resulting in a series of atmospheric correction models. And which is based on the theory of atmospheric radiative transfer and these models like MODTRAN moderate-resolution atmospheric transmittance and radiance code. Now one more issue or intimation is coming for moderate resolution. 

There was resolution high moderate coarse these are all relative terms, so what was moderate yesterday is coarse today and what was high yesterday is a moderate today.m So none the less not for very high-resolution satellite images but moderate resolution satellite images there is a model which is called MODTRAN. But when we go for very high-resolution images the area of the swath becomes very small. And therefore getting input data only for that part of the image becomes further difficult. But when we go for coarse or moderate resolution satellite images then somewhere at least within that image I might be getting those atmospheric data. So that is why you know for fine resolutions it is another problem. Then there is a 6S and other atmospheric radiation approximation calculation models are also available. Now the usage of these models and software is complex and that is why I said that note everyone would be developing their own. And therefore there are interphase-based atmospheric correction software packages are available. And which are like FLAASSH, ACTOR or ACORN or maybe few more one can find. But which one will provide the best results is very difficult to say unless one applies it to his own data set.

So based on the MODTRAN one example these models are constructed from a large number of look-up tables for convenient and rapid atmospheric corrections for a variety of sensors. Very briefly we will also see what is this MODTRAN basically it is used to calculate atmospheric transmittance and the radiative transfer with moderate spectral resolution data. And MODTRAN can be used to calculate atmospheric transference, atmospheric background radiation the radiance of single solar or lunar scattering, direct solar radiance, and likewise. And MODTRAN was based on LOWTRAN with a spectral resolution of 1 per centimeter. So the earlier version was LOWTRAN basically it was for low resolution or coarse resolution. Then moderate maybe in the future we may have HIGHTRAN, so the input parameters for the operation of MODTRAN can be divided into 5 types. A lot of input parameters are required, first is controlling and operating parameters, atmospheric parameters. That is atmospheric parameters and surface parameters which means somebody has to be present with the all kinds of recordings of atmospheric data of that part of the earth where the image will be acquired by a satellite. So then the current atmospheric parameters and that time corresponding surface parameters will be required.

Then observation geometry, how these have been observed, sensor parameters, of course, that is these are the fixed part observatory and these are the fixed part. But the other 3 are required corresponding or that time if required. So the main output, the result would be from MODTRAN is simulated apparent radiance and it will give the simulated apparent radiance. That can be used further by that software which just I have mentioned and then atmospheric corrections can be done. 

One example is here before correction on the left side and after correction on the right side and this is an example for Landsat TM images. And we are seeing in colors, so 3 bands 1, 2, 3 been used in RGB scheme. And this is what you see that mainly in this central part top central part you are seeing effects of atmosphere maybe because of thin clouds and others. And that effect is distortions have been minimized in the right side image. So atmospheric correction has shown significant improvement in the accuracy or visual interpretation of that satellite image. Now there is one more point here as long as one is using satellite images for qualitative analysis in assessment not many problems are there. But once we go for quantitative analysis of satellite data then it is a problem.

 For example, if I am an earth scientist or civil engineer and I am using satellite images to identify certain objects. Create a land use map may be an ecological map or for some other studies where I do not require quantities, I just require the clean image. So that my interpretation becomes much easier, for those purposes, not much atmospheric correction or no atmospheric correction would be required. But if I go and would like to find out how much concentration of certain pollutants is in the water or what is the real, what is the chlorophyll content in the vegetation. Then definitely I am moving towards the quantitative analysis or remote sensing data and that means I have to perform atmospheric correction tool without which it is not possible to estimate that thing. So this is what is as long as the remote sensing data is being used for qualitative analysis then it is rather easy and straightforward as I have just mentioned and given example. But when we imply quantitative analysis then all kinds of complications will arise and it can be seen in atmospheric correction. And therefore if remote sensing data is intended only for qualitative analysis. Then the brute force method or linear contrast stretch method can be employed. And it will give you a better quality image which is sufficient for the purpose which we are using the satellite images. So in that way, it becomes much easier, so this brings us to the end of the discussion. But in summary, I would like to tell that atmospheric correction performing because it is a complex atmosphere itself is a complex and dynamic phenomenon. A lot of parameters are required to correct or remove these atmospheric distortions and especially those parameters are required at that time when satellite image was been acquired. 

So corresponding data is required and it becomes very difficult to go into the field and collect all the data for the full extent of the satellite image. Therefore some model-based atmospheric corrections have become popular or brute force atmospheric corrections. If the quantitative analysis is not intended then brute force or simple linear contrast stretch can also improve your image quality though there are some none linear contrast stretch also there.

One can also perform and visualize whether the quality of an image has improved for better interpretation are not. And that can also be acceptable or can be accepted for such kinds of applications. So this brings us to the end of this discussion, thank you very much.