Dear Hasan,

My understanding of your question is that you would like to detect the felling of individual trees in a forest environment from space, possibly using Landsat. This is a rather more specific question than assessing forest disturbance, so it is critical to decide whether you need to know about the gaps created by the felling of individual trees or about the appearance of large clearings within the forest.

In either case, the first issue you need to address is to assess the horizontal size of your typical gaps when they are observed from above. This is usually called the 'mean canopy diameter (MCD)' for individual trees, or the linear size of the clearing area.

The next issue is to confirm the average height of those trees, and to compare that to the typical MCD: if the trees are rather thin and densely packed but quite elongated, removing a single individual from a full canopy may not result in a large enough difference in top of canopy reflectance to be measurable from space.

Then, you should document what is actually left once the gap has been created: will all branches and leaves remain on the ground? Is there an understory? Will the bare ground be visible? In other words, what sort of spectral and structural change are you expecting form such an operation?

A critical issue to address early on is to define what accuracy you will require: Do you absolutely need to identify each and every individual tree removed, or is it sufficient to detect that some trees were felled? If you are interested in clearings, do you need to know about the area cleared or the amount of wood removed? What will be the consequences of omission and commission errors?

If remote sensing from space is required (and even this should be justified in comparison with alternative approaches, in terms of feasibility, cost, accuracy, operations, etc.), and assuming you are considering detectors working in the visible and near-infrared spectral range (since you mentioned Landsat), you should select a space-borne instrument that has a spatial resolution rather finer than the size of the gaps you would like to detect: The finer the spatial resolution of the instrument the more likely you will 'see' those gaps on the image (or detect them with a computer program). Beware that the higher the spatial resolution, the higher the cost of data acquisition, and the smaller the total area that can be observed at once. The Ground Sampling Distance (GSD) of the instrument should definitely be smaller than half of the gap diameter. On the other hand, if your trees have large canopies (say 10 m or more), or if you care mostly about sizable clearings, you may not need sub-meter resolution.

And of course to detect changes (the appearance of gaps), you will need to have remote sensing images before and after the felling. This implies that you also need to be clear about the length of the period over which you need to monitor the forest, as well as the frequency of observations and the timeliness of the information retrieval: For instance, if you just need to report on forest disappearance on a yearly basis, you only need to make an assessment from time to time during the year (to account for cloud obscurations), but if you need to provide such information to law enforcement agencies in order to catch illegal loggers, you will need to arrange a system that can detect such changes in near-real time...

A rational approach to address these and related questions would be to build a model of the spectral and directional reflectance of your forest canopy, as it is observed from space. This would naturally include the perturbing effects of absorbing and scattering gases, clouds and aerosols in the atmosphere. Such a tool would allow you to estimate the likely effect of removing one tree (or some trees, or many trees) from the canopy on the measured signal at satellite level, and therefore to run sensitivity analyses to determine whether the approach is workable, and if so under what conditions.

In summary, document the average horizontal size of the typical disturbance you are interested in, the spectral and directional characteristics and properties of the gaps, and then select the instrument that will provide a GSD 2 to 5 times smaller than the size of those gaps and measurements that will be clearly sensitive to the type of changes expected, while remaining affordable. And pay lots of attention on the issues of accuracy and reliability, as well as data processing if you want to go beyond 'looking' at a few pictures.

Good luck in your investigation. Michel.

for measuring forest disturbance index first you have to understand what are the different parameter which is mostly active creating forest disturbance.

anthropogenic disturbance such as village population in vicinity of forest and roads are majour source of disturbance.

environmental factor such as slope, aspect and drainage etc can be incorporated.

higher the slope leads more soil erosion and restrict natural regeneration.

nothern aspect reatain more soil moisture with respect to other aspect...gives suitable environment for natural regeneration.

drainage is important because most of the rural village people remove forest ( in less than 5% slope ) and practice agriculture for their livelihood because these area are rich  in soil fertility and availablity of soil moisture.

by integrating all parameter  in gis ....you will get a map can be categorized into high medium, low disturbance index.

https://earthenginepartners.appspot.com/science-2013-global-forest

and attached

have fun

Article What Drives Conversion of Tropical Forest in Carrasco Provin...

Dear Hasan,

My understanding of your question is that you would like to detect the felling of individual trees in a forest environment from space, possibly using Landsat. This is a rather more specific question than assessing forest disturbance, so it is critical to decide whether you need to know about the gaps created by the felling of individual trees or about the appearance of large clearings within the forest.

In either case, the first issue you need to address is to assess the horizontal size of your typical gaps when they are observed from above. This is usually called the 'mean canopy diameter (MCD)' for individual trees, or the linear size of the clearing area.

The next issue is to confirm the average height of those trees, and to compare that to the typical MCD: if the trees are rather thin and densely packed but quite elongated, removing a single individual from a full canopy may not result in a large enough difference in top of canopy reflectance to be measurable from space.

Then, you should document what is actually left once the gap has been created: will all branches and leaves remain on the ground? Is there an understory? Will the bare ground be visible? In other words, what sort of spectral and structural change are you expecting form such an operation?

A critical issue to address early on is to define what accuracy you will require: Do you absolutely need to identify each and every individual tree removed, or is it sufficient to detect that some trees were felled? If you are interested in clearings, do you need to know about the area cleared or the amount of wood removed? What will be the consequences of omission and commission errors?

If remote sensing from space is required (and even this should be justified in comparison with alternative approaches, in terms of feasibility, cost, accuracy, operations, etc.), and assuming you are considering detectors working in the visible and near-infrared spectral range (since you mentioned Landsat), you should select a space-borne instrument that has a spatial resolution rather finer than the size of the gaps you would like to detect: The finer the spatial resolution of the instrument the more likely you will 'see' those gaps on the image (or detect them with a computer program). Beware that the higher the spatial resolution, the higher the cost of data acquisition, and the smaller the total area that can be observed at once. The Ground Sampling Distance (GSD) of the instrument should definitely be smaller than half of the gap diameter. On the other hand, if your trees have large canopies (say 10 m or more), or if you care mostly about sizable clearings, you may not need sub-meter resolution.

And of course to detect changes (the appearance of gaps), you will need to have remote sensing images before and after the felling. This implies that you also need to be clear about the length of the period over which you need to monitor the forest, as well as the frequency of observations and the timeliness of the information retrieval: For instance, if you just need to report on forest disappearance on a yearly basis, you only need to make an assessment from time to time during the year (to account for cloud obscurations), but if you need to provide such information to law enforcement agencies in order to catch illegal loggers, you will need to arrange a system that can detect such changes in near-real time...

A rational approach to address these and related questions would be to build a model of the spectral and directional reflectance of your forest canopy, as it is observed from space. This would naturally include the perturbing effects of absorbing and scattering gases, clouds and aerosols in the atmosphere. Such a tool would allow you to estimate the likely effect of removing one tree (or some trees, or many trees) from the canopy on the measured signal at satellite level, and therefore to run sensitivity analyses to determine whether the approach is workable, and if so under what conditions.

In summary, document the average horizontal size of the typical disturbance you are interested in, the spectral and directional characteristics and properties of the gaps, and then select the instrument that will provide a GSD 2 to 5 times smaller than the size of those gaps and measurements that will be clearly sensitive to the type of changes expected, while remaining affordable. And pay lots of attention on the issues of accuracy and reliability, as well as data processing if you want to go beyond 'looking' at a few pictures.

Good luck in your investigation. Michel.

dear Hasan ahmed

there some link of article related to disturbance model used for forest cover mapping and analysis 

https://www.researchgate.net/publication/285574324_Assessing_spatio-temporal_health_of_forest_cover_using_forest_canopy_density_model_and_forest_fragmentation_approach_in_Sundarban_reserve_forest_India

https://www.researchgate.net/publication/309589918_Driving_force_for_forest_fragmentation_explored_by_land_use_change_in_Song_watershed_India

https://www.researchgate.net/publication/308916544_Assessing_and_monitoring_forest_health_using_a_forest_fragmentation_approach_in_Sariska_Tiger_Reserve_India

Article Assessing spatio-temporal health of forest cover using fores...

Article Assessing and monitoring forest health using a forest fragme...

Article Driving force for forest fragmentation explored by land use ...

It is so hard to detect  tree gap trough Landsat data , especially if your forest is the coniferous one.These gaps usually have an area less then spatial resolution of Landsat images and besides the vegetation cover of other trees around the gap can cause some error in your  final estimation .Of course in this occasions some methods like subpixel classification is offered and you can find them in Image processing soft wares like Erdas Imagine. 

Read this

Urban change detection of Lahore (Pakistan) using the Thematic Mapper Images of Landsat since 1992-2010

Dear Hassan

The change detection tool is the best option to begin with if its for a small project,but make sure  you can take images from different periods with anniversary or nearly anniversary dates to avoid any phenological difference which may give false results if you use images from different periods.

GO THROUGH THIS ARTICLES

https://www.researchgate.net/publication/301609558_Geospatial_analysis_of_forest_fragmentation_in_Uttara_Kannada_District_India

https://www.researchgate.net/publication/260908980_Spatio-temporal_dynamics_along_the_terrain_gradient_of_diverse_landscape

Article Spatio-temporal dynamics along the terrain gradient of diver...

Article Geospatial analysis of forest fragmentation in Uttara Kannad...

My suggestion would be Change Detection analysis in ENVI software, using 2 separate images from the same area in 2 different time. then by loading NIR band instead of Red band you can detect any changing which may has occurred in your area. just mind that in this process you should create some ROIs as well.

The dificults for these study become from geometry of sun illumination. May be these paper illuminate these question

Article Multi-temporal analysis of radiometric changes in satellite ...