Geospatial Science - Science topic

Waiting for the answer

it depend on your data values range . bourogh in his Phd tried to use several fuzzy raster and make overlay by Gamma function and he recommended that maximum value of your fuzzificated data is your gamma value. for example if your data varied between 0.2 to 0.934 your Gamma value should be 0.934

Greetings, here is a list of 10 free software for creating flowcharts.

1. LucidChart

2. Draw.io

3. HEFLO

4. Coggle

5. Bizagi Modeler

6. Mind Meister

7. Funnelytics

8. Gliffy

9. Visio

10. SmartDraw

More information at the link:

Dear Linh

are you interested in using radar data for soil erosion survey?

Indeed, good start

hello, i downloaded the collection 2 level 2 data (landsat 5 and 8) from USGS. i cant able to open in erdas imagine 2015 . it comes with error ( unwanted compression type) . please help with this!!!!!. thankyou

Left.

A nicely maintained list is here:

List of freely accessible LiDAR data and digital terrain models: https://arheologijaslovenija.blogspot.com/p/blog-page_81.html

These links may be helpful for you.

- https://desktop.arcgis.com/en/arcmap/latest/manage-data/geodatabases/an-overview-of-geodatabase-design.htm

Hello Babak Ranjgar

When talking about IoT, we're talking consequently about interconectivity among devices. In this way, the application of IoT concept combined with GIS relates to the use of several devices conected to a server in order to obtain a lot of geographic information to be user to feed the information system.

For instance: your smartphone, using a GPS software, may provide severam information to a central server about your localization, weather conditions, vehicles traffic, shopping facilities, and a lot of other information that may be provided by other user of the same network you're connected, helping you to find what you need.

This may involve a lot of devices beyond smartphones: personal computers (mainly laptops), smartwhatches, equipmentes dedicated to work with geoprocessing, detection of weather conditions, etc.

A new term also emerged from the combination of IoT with geoprocessing: Geography of Things (GoT).

This text presents some ideias about you question: https://geografiadascoisas.com.br/artigos/geography-of-things/. The video on the end of the text is a TED talk in portuguese, but you may enable the subtitles in english.

Dear Gamal,

welcome,

Any analog function f(t) or f(x) which is a function of time or position in the sense they vary with time or position or both when drawn in a linted time span or position range one depicts the variations of the amplitude of the function with time and position. Such variations may be called as the profile of the function or its waveform. Such look is abstract and complex.

The scientists were convinced that such complex shapes an be constructed from a unitary periodic wave forms which they discovered to be the sinusoidal waves or the complex exponential signals. So, it is so that any waveform in t or x or any variable is composed from its basic sinusoidal components. So, any waveform can be resolved into its constituting cosinusoidal weaves or complex exponential functions. These sinusoidal components or vibrations can be represented in the frequency domain as they are extending from - infinity to plus infinity. Such representation is called the frequency spectrum of the signal. It is normally designated by X(f).

So, one has devised two equivalent descriptions of the signal, the time domain description and the frequency domain description where the real domain in which the signals happens is the time domain while the frequency domain is the domain where the signal is analysed to its fundamental sinusoidal components.

The Fourier devised the Fourier transform FT to convert the signal from the time domain to the frequency domain. This transform is reversible which means that if one has the frequency components of the signal one can get the time domain signal.

There is great benefit to analyse the signal and get its frequency spectrum. One of great benefit is to determine its frequency range and its bandwidth. One can frequency limit this signal and filter it. One can synthesize it from its frequency components.

So assume that one has x(t) transformed it by FT to X(f),

This from is called analog form which the values are continuous with time.

For digital signal processing one has to transform the analog signal into digital form which can xd= x[n], which is a sequence of values of the x called digitized samples. The sampling rate fs must be accomplished according the Nyquist rate fs>= 2fm where fm is the maximum frequency in x(t),

The Fourier transform of this digital signal x[n] is called the Digital Fourier Transform DFT, X[k] where k is the sample index in the frequency domain.

Th DFT can be considered the sampled frequency domain.

Since DFT is time consuming one has invented the fast Fourier transform FTT to save calculations in DFT.

The most important point is that one can resolve any complex waveform or function into its frequency spectrum content for sake of signal analysis , synthesis and filtering.

Best wishes

As an architect it's the biggest question

TEC - the total number of electrons in a rotation cylinder centered around the line of sight receiver–satellite with a cross section of one square meter. The TEC is expressed in so–called TEC Units (TECU). Example: for L1 solutions with an elevation cut–off angle of 15◦ and a prevailing TEC of 10TECU, you may expect a baseline shrinking of about 10·0.10 = 1.0ppm (or 1.0mm/km).

Have you see Jonas Ellehauge's article in GIM International (October 2016)? https://www.gim-international.com/content/blog/blockchain-in-geospatial-applications-2

More on his Web site, including slides: http://blog.awesomemap.tools/blockchainingeospatial/

I particularly like his IoT use cases.

I suggest that you begin with a book, e.g. "Geostatistics", J.-P. Chiles and P. Delfiner, John Wiley and Sons (2nd edition)

For a tutorial paper see

1991, Myers,D.E.Interpolation and Estimation with Spatially Located Data Chemometrics and Intelligent Laboratory Systems 11, 209-228

This paper uses a free software package called GeoEAS (from the US EPA) then as a step up see the open source free R package called gstat. You can download various binaries (Linux, Windows, Mac) and also the source codes if you want. The software comes with tutorial materials and has very good graphics

Many of the other suggestions listed by others are commercial software and are of varying quality. Since they will not give you access to the source code you will have to know much more about the underlying theory to know what they can do and what they can not do. In general R software packages are better than commercial packages. (Do a search in Google for "R package gstat")

If you going to use the variogram for kriging or simulation remember that the purpose of the variogram is purely utilitarian; you are trying to develop a model that captures the spatial covariance of observed values.  The model you specify for kriging or simulation helps control the smoothness and roughness of the resulting surface.  Therefore you want enough variogram points to indicate a range of spatial dependence,  existence and  magnitude of any very small-scale variability ("nugget effect"), and a value for the sill.  Four points might be enough for some datasets, including yours.  For others, forty might not be enough.  You can play around with different lag sizes but remember that as the number of variogram values increases the number of pairs used to calculated each value decreases and the variogram can become more noisy.

If you create land suitability model so you have face pixel mixing problem for accuracy purposes.

Weighted Overall analysis method for land suitability map & sub- pixel is another very good method for pixel mixing of classes.

QGIS (http://www.qgis.org/zh_CN/site/) is the most commonly used free open source software for analysis and visualization of spatial data. It is user friendly, well-documented, equipped with hundreds of tools, and enables implementation of user's algorithms.

SAGA GIS (https://sourceforge.net/projects/saga-gis/) and GRASS GIS (https://grass.osgeo.org/) are also worth to consider. They can be installed as standalone, but they are additionally integrated in QGIS.

I already finished the project. At the time I asked this question  SNAP had not yet been released. Therefore I used Sentinel-1 toolbox for processing and ENVI for image classification.

But I did some research after that and it is possible to extract  Polarimetric variables from the Sentinel-1 images using SNAP and Polsalpro.

I'll leave a link on this matter.

One of the major problems is that items less than about 10-cm are hard to detect and/or track with terrestrial radar. What about space-based radar? There are satellites in orbit that could do this, should they be re-tasked..

Dear Jackson,

I think cluster analysis is not a good way to analyse such kind of data. Besides, why only using range centroids? I think it is a quite misleading to only use the centroids and disregard the taxa's actual distribution ranges because you loose a lot of information (e.g., the area of each taxon's distribution and whether distributions overlap or not).

I guess the centroids are based on point data, which you could use for an ecological network analysis as proposed by Strona and Veech (2015; http://onlinelibrary.wiley.com/doi/10.1111/2041-210X.12395/abstract).

We recently used this analysis to assess varying spatial patterns of species distributions in geological time. The analysis nicely showed increasing and decreasing range overlap, more or less corresponding to lowering or rising provincialism. Our underyling data structure was a presence-absence matrix based on occurrence data fitted to an equal-area grid of the geographic region of interest.

Best regards,

Thomas

hai first remove bad bands and water vapour 1-7 58-76 165-182 221-242 you have 163 bands around use radiometric calibration tool convert DN to radiance. after that use FLAASH or QUAC my friend.scale factor 0.1 for radiance conversion

Greetings Mohammed. Lot of work is done on aerosols and space observations. But the spectral signature characteristics [ optical or IR and thermal or MW regions in general] associated with anthropogenic and natural aerosols is not a significant part of any study. For this exclusive mapping of aerosols in different types of ambient settings would be essential. Also the physical parameters of weather play important role. In most of the developing countries which are occupied with large scale construction and transportation, one could observe the dominance of anthropogenic aerosols   A pristine blue and green environment may give a aerosols data not influenced by anthropogenic substances. I think this would be possible to distinguish using MODSI products with extensive ground truth..  

Did you check if the polygons are multipart-polygons? If so, try to "break" them and do the conversion again.

This paper might also have some good data/information for you:

ASTER

SEVIRI

AVHRR  

VIIRS  

AATSR  

Hi!

I'll work with this subject soon, and share any advise and/or information i find.

I'll check this papers too.

Best regards...

Hey!

If You use ArcGIS the thing is simple.

1. Download the DEM together with the ENVI header (*.hdr)

2. Next in ArcMap use: ArcToolbox - Conversion Tools - From Raster - Raster to Point (this option will extract all DEM's pixels height values to point shapefile - very useful tool)

3. But, if You have Your own set of points and You want to compare height values from two different DEMs exactly in places where these points are located - You can extract height values (for Your set of points) from one DEM and then from another; You have to use in ArMap: ArcToolbox - Spatial Analyst Tools - Extraction - Extract Values to Points.

Good Luck!

Bart

William is right. I would also mention the opportunities to analyze spatial data and get some indicators that could help you to make some important decisions related to location. Some examples might be the Nearest Neighbor index, or the Local Moran's index.   

Hallo Nils, die Insel ist vermutlich heute unter anderem Namen bekannt. Sie liegt nicht bei Rapanui (Osterinsel) sondern auf der anderen Seite des Pazifiks in Melanesien. Sie wird allerdings im Buch von Katherine Routledge "The Mystery of Easter Island" erwähnt. Dort findet sich das Zitat über Nangone auf Seite 133: Bishop of Wellington, "Notes on the Maoris and Melanesians" The Journal of the Ethnographical Society, New Series, Vol. 1 Session 1868-69. Vielleicht findet sich dort eine Karte. In jedem Fall ist es eine Koralleninsel, und das trifft für Rapanui nicht zu.

Aloha!

Sometimes if two peaks are not clearly identified in histogram then it may lead to error. I had used a r package (EBImage)  for otsu on a raster. You may try that too.

You can use total column precipitable water vapor. To gain more information, you can refer to the following publication:

I agree with Paulo about the google search.  Or you may simply refer to the landsat documentation for details (EVI is on page 13).  See below.

Hi Somayeh. I think that Sentinel-1 can really be used for estimating AGB in woodlands.

But there are some constraints that you must be aware of:

I think the most relevant constraint is that Sentinel-1 works in C-band. It means that the signal it uses interacts mostly with canopy top and do not allow to get a direct analysis of the ground beneath the trees. Then, forest/woodland trees' volume cannot be assessed by a direct analysis, wich means that the analysis of AGB using Sentinel-1 data will certainly have an error that you must to assess if it is acceptable or not to your research objectives.

You have a free plug-in for Adobe Reader on

 and

free converter to geotiff

Thanks you all guys. I found a solution within Data Driven Pages tool. Jan Burdziej put me in the right direction. Within the Data Driven Pages tool window you have an option to vary the spatial reference system throughout your map production. There are three options, but the simplest one is to create a field in the attribute table of the index layer to specify what coordinate system should be used to display each feature (in my case countries). Use the conventional number assign to each coordinate system or point to the prj field corresponding to each coord system. Obviously the first option is less computing intensive. In optional fields of the Data Driven Pages tool you have the additional options including "spatial reference". Display the dropdown menu and choose "field" option and point to the field which set the CS code. 

As Jan said, it can also be done using arcpy, but this option is quite more simple.

Thanks again for your contribution.

Howdy Essayas! There is a conversation on this over here as well which might help give you the answer you need.  https://www.researchgate.net/post/Is_it_incorrect_to_calculate_NDVI_using_DN_Digital_Numbers_values_associated_with_a_satellite_image_instead_of_using_the_Surface_Reflectance_values

I think there is an R package for this (open source software)

You might also want to do a search for "Independent Data Analysis"

Finally it is possible to do geostatistical simulation for multiple variables

1989, Myers,D.E.,Vector Conditional Simulation in Geostatistics, M.Armstrong (ed), D. Reidel Pub. (Proceedings of the Third International Geostatistical Congress, Avignon, France 5-9 September 1988), 283-293

Dear Hein,

The forest depletion (= deforestation + degradation) rate for Borneo is approx. -3%yr-1. This includes -2.2%yr-1 of deforestation. So, your Bolivian findings are consistent with Borneo data.

 We don't have illegal logging problem in Poland. Forests almost entirely belong to the state. The data for GFRA report are prepared based on data provided by the forest district officers and are reliable.

 Will you be interested to do a mutual project on forest dynamics? (kbecek"gmail.com)

 Here are my paper related to Borneo/Brunei.

 Best Regards,

I think, u have stereo image, so better u use Leica photogrammetry suite (LPS) tool available in ERDAS..

It would also be good to define what a "village" is in India, if there is an official description somewhere. The nic.in site appears to divide districts into villages, each with a distinct polygon feature.

As Kazimierz said, you need to look to the sampling theorem. Or you may need to look to the scale-space representation, which addresses the level of details in your signal (here in your question 100 x 100 matrix).

Dear Rajib,

i also used TVX  approach to check errors while relating the LST with T air (MET data). while comparing the LST with Tair u can get some errors you should first check the exact location of MET station, also its land use  and record its height usually the  met stations record the air temperature at about 2.6 meters above ground. 

Also note that their will be time difference between the satellite over pass and the time at which the MET had recorded the air temperature. Therefore  you should use the AWS data for better comparison as they record the temperature more frequently. 

and if you have gone through above two articles i think it will help you  further more i will very soon add my publication where i have tried to explain all these errors in LST.   

Hello Dr. Dongjie Fu,

What you are looking for is readily available @ USGS  http://earthexplorer.usgs.gov/

You don't need to know the path and row details for a specific LANDSAT or MODIS scene as long as you know your study area.

To make your task simple I've uploaded a series of pictures which can walk you through the process of retrieving LANDSAT or MODIS data without knowing any of their orbit details. Just follow these steps:

I hope these steps are easy to follow. If not you can always watch this youtube webinar (~ 50 min) by NASA. 

Hi,

corelator to corelate the data in master/slave images using window size. If your z in between  range, so you may increase correlation values.

-Ram

Hello Gamal: You might be interested in having a look at the following paper:

Mazaheri, M., and Habib, A., 2015. Quaternion-Based Solutions for the Single Photo Resection Problem. Photogrammetric Engineering and Remote Sensing Journal, Vol. 81, No. 3, pp. 209 – 2017.

This paper presents a quaternion-based closed form and iterative solutions for solving the SPR problem without the need for approximate values or partial derivatives.

Best regards, Ayman.

QGIS is fast developing, easy to learn, multiplatform (linux), customizable (plugins etc.) and open-source (the biggest benefit). If more complicated analysis are needed, you can run there functions from GRASS, SAGA or R. Also its more  user-friendly in many cases (for me). I am using QGIS for years and never need go for ArcGIS again.

example using ERDAS model maker

Dear Usman

You can easily carryout gap filling process with Model Builder and apply focal mean with a matrix window of your choice, but preferably 3x3. I am sharing the model, you need input just the input layerstack images.

From experience: any unsupervised classification of mid-resolution imagery will included a class corresponding to bare rock, if the latter is present in the area of interest, or more than one class when various types of bare rock (e.g. dolomite versus granite) occur at the surface. Presumably you know your area of interest well enough to locate some bare rocks on the image, a topographic map or along a road to allow for a supervised image classification. Attached is a case; secondarily we lumped the grass and the bare rock classes as neither of them was the target of our research. But the rock class stood out.

Dear Mohamad Ivan Fanany, 

you can find some global land use / land cover produts here

Oh,I'm sorry! I made a mistake ! Thank you all!   I was  not going to retrieve SST but  SSC(Suspended Sediment Concentration) ! So I have a question for Amir: Is it OK to use FLAASH to  do atmospheric correction for quantitative RS, especially for SSC retrieval in case2 waters?   Thank you very much !

Hi,

A detailed explanation of the problem is given here: http://landsat.gsfc.nasa.gov/?p=9308

A good source of information on image destriping is "Computer Processing of Remotely-Sensed Images - An Introduction" (2011), by Paul Mather and Magaly Koch.

Best wishes

Montalvo, AM, McMillan, PA, Allen, EB (2002) The relative importance of seeding method, soil ripping and soil variables on seeding success. Resoration Ecol 10: 52-67

Ditomaso, JM, Brooks, ML, Allen, EB, Minnich, R, Rice, PM, Kyser, GB (2006) Control of Invasive Weeds with Prescribed Burning. Weed Technology 20: 535-548

Gillespie, IG, Allen, EB (2004) Fire and competition in a southern California grassland: Impacts on the rare forb erodium macrophyllum. J Appl Ecol 41: 643-652

Moyes, AB, Witter, MS, Gamon, JA (2005) Restoration of native perennials in a California annual grassland after prescribed spring burning and solarization. Restoration Ecology 13: 659-666

Heneghan, L, Miller, SP, Baer, S, Callaham Jr, MA, Montgomery, J, Pavao-Zuckerman, M, Rhoades, CC, Richardson, S (2008) Integrating soil ecological knowledge into restoration Management. Restoration Ecol 16: 608-617

Hayes, GF, Holl, KD (2011) Manipulating disturbance regimes and seeding to restore mesic Mediterranean grasslands. Appl Veg Sci 304-315

Getting there: you classified with three imagery-derived variables.

More questions: what input imagery was used for your NDVI, Albedo and GSI?

Desert is presumably defined as a pixel, or group of pixels, with low NDVI (low green biomass), high Albedo and/or high GSI (high sand content) ?

Desertification as decrease in NDVI, increase in albedo and/or GSI?

NDVI and vegetation cover may be statistically tested for correlation, GSI and  proportion of sand at the surface as well; as far I can see you have no proxy for albedo. Albedo and NDVI may be tricky in your desert context as spatial variation in surface lithology (e.g. sand versus rock desert) within your Area-of-Interest may overrule any variation in cover or albedo signal from the "desert"; if so you better stratify your area accordingly. For an example of extraordinary low NDVI values due to calcareous bare rock (arctic-alpine desert) please see attached.

Yes Satya, but it does require fieldwork to examine/collect soils, analyze their properties, and then couple this information with the Remote Sensing data in a GIS system. 

Hi Xuegang,

The following URLs are the potential starting points for you. I also refer you to the official tutorial documents for ENVI. By precisely reading these documents (and other similar ones) you will be able to implement your processing easily.

Good Luck!

*ENVI Tutorial (chapter 9 would be your target!):

Dear Rakesh,

use the Focal Statistics Tool from the Spatial Analyst Extension. Use as Neighborhood Type Weight. Here in a text file you can define any size and values for your filter kernel. Define your 3 different filters, apply them on your slope raster and combine the resulting three smoothed rasters using the raster calculator with the con-funcion.

In fact, you were asking a technical issue of image differencing between the images obtained by different sensors on different dates. If the acquisition dates of your images are the same or close to each other in the same month but in different years,  here are some suggestions for your consideration: (1) In my opinion, you have to conduct first atmospheric correction but not only radiometric correction since two images were acquired on different dates by different captors (TM and OLI). TOA means Top of Atmosphere, so that TOA values either in radiance or reflectance are actually at-satelite values but not ground values.  You may apply the FLAASH model (Perkins et al 2005) or the improved DOS model (Chavez 1988 and 1996) for achieving this correction and conversion of each band (in DN) into reflectance taking the path radiance and scattering effects into account. (2) Having so done, you may have to be aware of the fact that OLI images contain the coastal blue band (B1) so the bands to derive NDVI or other indices are different, e.g., for TM, B3 (red) and B4 (NIR) but for OLI they are B4 (Red) and B5 (NIR). In addition, OLI B5 has narrower spectral region (0.851-0.879 um) than that of TM band 4 (0.76-0.90 um). It is normal that NDVI of OLI is slightly lower than that of TM for the same vegetal biome.  (3) After Step 1 and 2, you may choose some pixels or patches at the same places (defined by the same polygons) in two images, where no clear change has occurred, to check their average NDVI. These pixels can be water (no change with time), cropland (wheat or rice land repeatedly cultivated), and cement/concrete built-up in urban squares (no change if the observation time of year is close to each other). If the difference is < 5%, you can consider that there is no change on these checked land cover types between the two images. The two images are usable for differencing. If the difference of these checked biomes is > 10 %, the two images (at least one of them) have problem in quality. Thus differencing techniques may not be applicable. In this case, you can apply "post-classification comparison" or "post-classification differencing" (Wu 2009) for change detection.

I hope these may help you a bit to sort out your problem. Good try.  Here some publications may be of reference for you.

Perkins, T., Adler-Golden, S., Matthew, M., Berk, A., Anderson, G., Gardner, J. and Felde, G., 2005. Retrieval of atmospheric properties from hyper and multispectral imagery with the FLAASH atmospheric correction algorithm. In: Schäfer, K., Comerón, A.T., Slusser, J. R., Picard, R. H., Carleer, M.R. and Sifakis, N. (Eds), Remote Sensing of Clouds and the Atmosphere X, Proceedings of SPIE, Vol. 5979.

Andrew, the suggestions from Viorel and Alessandro are good, but you need to take some more steps.  After generating the maximum raster and three new rasters, substracting each original raster (1,2 and 3) from the maximum, where 0 would be the cells with the max value in each. For each of them, do the following; set all the cells with non 0 values to null, and the cells with 0 to 1 for raster 1, 2 for raster 2 and 3 for raster 3.  Then combine them in a single new raster, where each cell would indicate the origin raster. Then you can dismiss the intermediate difference rasters and have: the maximum raster, and the origin raster.

Dear Stuart,

in the archaeological domain this kind of operational approach is "still" very diffuse, due to enormous amount of archives material items, such as paper maps, radex and similar data recording supports, furthermore still too much employed in current field excavations and analysis.

Within our SITAR Project (see at http://sitar.archeoroma.beniculturali.it/ and some other resources at my RG profile), digitization activities represent the most important aspect for the first phase of our geo-data bank implementation, necessary for extracting above all polygonal archaeological features, altimetric points and related geospatial informations for mapping archaeological sites, monuments and complexes, and their fragmental partitions.

By the way, more generally such a necessary approach to data "distillation" implies some issues with regard to data heterogeneity, their completeness and final use and re-use. In this sense, this native heterogeneity generates, more or less deeply and distinctly, some questions on precision/completeness both at topographical and descriptive level, for old and also new data, motivating the implementation of semi-automated workflows in order to compare and re-align fragmentary different analogic source data.

Hi, you can try our NSPI or GNSPI methods. You can get IDL code from my webpage http://xiaolinzhu.weebly.com/. Please let me know if you meet any problems.

Fmask is a very powerful tool for cloud mask generation. All details and the tools itself can be found here: http://code.google.com/p/fmask/

I used Fmask for several Landsat 7 ETM+ and Landsat 8 images...

Best wishes Olaf

The MOD13 product is corrected for angular variations in observation and sun-target-sensor variations but not due to topography using a DEM (I believe). It is also important to note that there has occurred a significant degradation of the Terra MODIS sensor affecting VI data. See Wang et al. (2012) Remote Sens. Environ., 119.

I don't know if I understand correctly your question.Your idea to construct a weighting matrix distance-based should be correct. What type of variable are you studying? How your samples are distributed in the study area? For example, you could construct a regular point grid and then calculate the distance of your sample points from the nearest point of the grid, and use this distance as weight. Otherwise you can construct an experimental variogram, model it and use the model in the kriging algorithm to estimate your variable at unsampled points and represent the final map in term of  probability. I hope this to be useful. Best regards, Giancarlo

Concerning your second question: as no/insufficient vegetation available for combustion (fuel load) inhibits the occurrence of a fire, I would at least include some quantitative measure of vegetation. Furthermore, humans are the major source of ignition in many ecosystems (and I could imagine in the Mediterraneans as well) - it thus makes sense to somehow account for those (e.g. proximity to roads, accessability,...) when determining fire risk or probability of ignition.  

Concur with the others, there is really no universal optimal window size. Really depends on what your object size is, the imagery you are using, and the image characteristics you are aiming to capture. In the paper noted below, we tried to use variograms to provide for customization of the windows and the information generated (Wulder et al. 1998, RSE). Pretty old, so am sure more and better has been done since. In general, IMHO for characterizing local variance conditions fancier methods have not really outperformed simpler and more transparent approaches (e.g., variance within a fixed window).  If you have your objects via other means (say segmentation), within object texture is also a useful option. 

Best of luck with your research. 

Dear Majid,

Your question is an interesting one. I investigated some literature and I found the following according to Azad Hossain et al. (2006) many factors such as suspended particle size, shape and color can have large influences on spectral properties of suspended particles.

In two different researches NIR is not used in mapping suspended solids and I guess that is why you have low correlation. Instead the visible wavelength especially the green band.

   

You must apply the indicatrice kriging method (topo_probabilist) , but in all times we find the default set of parameters for Facies, the raison the mineralogical heterogeneities especially with clays and sitstone, when we do the continuous logs facies by the methods of prediction and rebuilding the lithological facies like (Neural Network) by the combination with the real observations on cores and logging measurements, we find a slightly default set of parameters for Facies modeling for a real field,

Thank you Fakhri again. Your advice is that we should divide the whole megalopolis into many subregion, then we research the rules in these subregion. Yes, that's a good idea! However, another question maybe we have to face with is that how to divede the megalopolis?  If we could not provide strong evidence about how to divide the megalopolis, our research results maybe questined. Someone could say that we get the results maybe just by accident. Do you have any idea about that?

Thanks so much!

Then it should be fairly easy to produce a RasterStack from your ncdf fields on which you can use raster::calc() to produce a RasterLayer of pixel averages. See help(calc)

Thank you for replies. It was because I had not defined spatial reference frame. Defined it and now it is working.

Though that was not a problem with earlier versions.

Thanks again.

Since a same LULC may have different radiometric values along time (because images may be taken with different sensors, different sun illuminations etc) you should redefine ROIs. However, beyond the problems mentioned in other answers (registration etc), another problem may appear if LULC has changed : indeed, all classification algorithms define  the spectral boundaries between classes based on the statistical distribution of radiometric values at each date, so that the boundaries may move. The consequence is that a pixel where no change has occurred can jump over the boundary and become another class. Another approach to overcome this problem would be to run a single big classicifation, both multi-spectral and multi-temporal, over all images, with multitemporal classes (e.g. forest becoming grass between date t1 and t2).

Try here with ground truth data

Hi James,

Why don't you start with the most cited (?) paper in remote sensing: http://www.tandfonline.com/doi/abs/10.1080/01431168908903939#.VH460THF98E

Followed by http://www.tandfonline.com/doi/abs/10.1080/014311699213659#.VH47ezHF98E 

Also there is an excellent chapter in Jensen's DIP book. All these will give you the basic idea of available change detection techniques. 

Since you will be using multi-source/temporal data, along with geometric integrity, you should consider the atmospheric effects: http://www.sciencedirect.com/science/article/pii/S0034425700001693.

Now you can follow (reproduce) this work: http://www.tandfonline.com/doi/abs/10.1080/01431160801950162#.VH49ETHF98E

After you have some expertise, you can take the problem in sub-pixel or unmixing domain to deal with the different spatial resolutions. I couldn't find any particular work but I will be happy discuss on this!

Enjoy!

-Ani