USD $49.95 (Money Back Guarantee!)

PANCROMA Frequently Asked Questions

This page lists some questions that PANCROMA users have asked in the past. I listed my answers in order to provide some instant feedback in case you have the same problem. Do not hesitate to contact me directly if you have an issue, whether you are a customer or just trying the software. If you find a bug, try installing the latest version of the software. It this does not help, please report the problem. The fastest way is to email me with a short description of the problem. I will do my best to correct all such issues with the highest priority. Your satisfaction is my number one concern!

Where can I get Landsat data?

Free Landsat data is available from several sources. Look at the links on the 'Data' page on this website or under the 'Help' menu selection in PANCROMA.

Which Landsat database is the best?

Each of them have their advantages. The USGS GLOVIS site seems to have the most comprehensive set of Landsat data with dozens of data sets per scene. GLOVIS also archives ETM+, TM and other legacy Landsat data.

GLCF has less Landsat selection but may be a bit more user friendly. USGS Earth Explorer seems to duplicate the GLOVIS data. Canada's GEOGRATIS has only Canadian scenes. These sites may be worth searching if you cannot find what you are looking for on GLOVIS and GLCF.

The sites are confusing. How do I actually download the data?

GLOVIS and GLCF both have tutorials and help files on their sites. Alternatively, flog around for a while and you will learn eventually.

The GLCF site archives two types of Landsat files: .tif and .L1G. Which should I select?

Landsat data can be archived in a variety of formats. Geotiff (.tif) is the most prevalent. GLCF archives a considerable amount of .L1G format data. PANCROMA can import either. Don't try to mix data however.

User writes: Where can I download free SPOT data?

The USGS Earth Explorer website archives SPOT color composite data that PANCROMA can decompose to the band files. The Canadian GeoBase website offers full SPOT band data for all of Canada. See the 'Free Data' page on this website or the 'Help' menu.

PANCROMA seems like a complicated program. How can I best get started?

It is not as complicated as it looks. PANCROMA is composed of many utility routines, each of which is straightforward to learn. Few users need all of them. Start by importing and a single band and displaying a grayscale image. (Grayscale is all you can produce from a single image.) Next import three bands (bands 1, 2 and 3 corresponding to blue, green and red channels) and display an RGB (color) 30m resolution image. These are both easy to do. Experiment with the image processing utilities to become familiar with how they work.

The next step is to subset a set of Landsat images. Import bands 1, 2, 3 and the panchromatic band (these have …10, …20, …30 and …80 file suffixes. Create and save a matched set of cropped data. This is essential for learning the pan- sharpening process as trying to iterate complete files is too cumbersome when learning. Use this file set to try pan sharpening. Forget about image processing the first time. Just see if you can create a high resolution (15m) color image.

Then progress to using the image processing utilities to get a better pan sharpened result. Be sure to use the 'Maintain RGB Comparison Image' feature to get the best results.

I tried to display a single band image and all I get is a black screen.

Scroll down and to the right. You are parked on the image border.

There are seven Landsat bands. Which ones do I use?

Each band file corresponds to a different sensor on board Landsat. Bands 1, 2 and 3 (blue, green red) are the most useful for producing images that match the true color of the image. However, the other bands are extremely useful. For example, since blue light is easily scattered by the earth's atmosphere, particularly by water vapor in the atmosphere, band 1 usually produces the murkiest image. Very interesting images can be produced by using bands 3, 4 and 5 instead of 1, 2 and 3. These produce colors that resemble true color but the image is much clearer as a result of these bands "seeing" through atmospheric haze better. They also tend to reveal interesting features of the earth's geology not easily visible in a true color image.

PANCROMA will not let me output my processed file in .jpg format. Why not?

Pan sharpened and even panchromatic Landsat files are very large. The JPEG file compression routines in the libraries used by PANCROMA must all duplicate the full image data set into a flat (unsigned char) array before starting the compression. PANCROMA offers work-arounds like disabling the image display for large files (see the Instruction Manual). Saving your images in PNG format is another alternative as the PNG writer is much more memory efficient than the JPEG writer. PNG files compress to nearly the size of JPEGs and have the advantage of being a lossless compression.

I keep getting an 'Out of Memory' error when subsetting Landsat files.

Most computers should be able to subset an entire set of band and panchromatic files if you uncheck the 'Display Subset Images' check box on the subset Information dialog. The process will take more or less time depending on your computer.

How long will it take to pan sharpen a set of Landsat images?

Pan sharpening a full-size Landsat data set requires more memory than the Microsoft operating systems will allocate to a single process. As a result data must be written to hard disk scratch files, which considerably slows down data access. It will take approximately 850 seconds with a fast computer. At the other end of the spectrum, I was able to successfully pan sharpen a full Landsat data set, save the pan sharpened image in .bmp format, and display the image in Paintshop Pro on my 1.1GHz, 256MB RAM Windows XL laptop. The pan sharpened image was 17,108 by 15,156 pixels and was 759MB in size. PANCROMA version was 1.12. It took about 3 hours to process the data on this machine.

User writes: I got an 'Out of system Resources' message when saving my pan sharpened image.

A common problem is memory resident utilities such as virus shields and the like that use a lot of RAM. Try disabling these during PANCROMA runs. You can see how much memory is used by such utilities by calling the Windows Task Manager and examining your program resources.

User writes: The subsetting option should be explained before the pan sharpening instructions so that if the pan sharpening of the original image fails the person will immediately know what option is available. I had to search around your instructions to find this myself.

Good point. Most users may not need the full sized pan sharpened image any way. I added a 'Getting Started' topic up front in the Instruction Manual that highlights this.

The grassy vegetation areas of my pan sharpened image have a blue cast to them. What is wrong?

The HSI transformation and inverse transform used by PANCROMA is nearly perfect. Test images transformed from the RGB band files to their corresponding HSI bands, expanded, interpolated and then transformed back to RGB match the input files almost exactly. Color distortion is a result of the characteristics of the panchromatic band that is substituted for the Intensity band before the reverses transformation is performed during pan sharpening. The panchromatic band represents radiated light in wavelengths of 0.52 to 0.90 microns. This includes the visible green, red and near IR bands. When it is back-substituted in place of the Intensity band, the colors in the resulting pan sharpened image do not exactly match those of the RGB color composite image constructed from the low-resolution input files. This is because the panchromatic band does not exactly match the Intensity band computed from the input RGB bands. The problem cannot be avoided but it can be corrected. In some cases the correction is nearly perfect with little extra effort. In other cases minor distortions remain even after extensive automatic and manual image processing efforts.

The most useful correction to this problem is to correct distortions caused by the panchromatic image spectral sensitivity using the NIR band 4 data and the XIONG or AJISANE algorithms. These techniques can greatly improve the result, often eliminating the problem completely. Histogram matching, performed automatically by PANCROMA (upon command) adjusts the grayscale level of each pixel in the panchromatic band to more closely match the corresponding pixel in the computed intensity band. This can also result in a pan sharpened image with less color distortion. PANCROMA offers two histogram match methods: linear and non-linear. In general the non-linear method is superior. A non-linear histogram match is performed by PANCROMA as the default condition.

I pan sharpened my image using non-linear histogram matching and the forested regions of my image are still blue. What should I do?

You can use the ENHG Enhanced Green color adjustment for stubborn images. This technique uses the NIR band in addition to the red, green and blue bands to identify and correct the areas of your image covered by vegetation. See the Instruction Manual for more information.

I downloaded the extra NIR band and used the ENHG automatic method. The vegetation is now green but it does not look natural.

The XIONG algorithm is actually the best for correcting the spectral distortion described above. To use it, load five bands (Blue, Green, Red, panchromatic and NIR) and then select 'Pan Sharpening' | 'HSI Transform' | 'XIONG Five File (Automatic Method)'.

I went step by step through Video 7 with the same image set and got a different output (very dark areas).

The reason that you did not get the same result is that I added histogram matching since I recorded the video. To get the same result, disable histogram matching. Check the 'Activate Image Processing Routines' check box on the main form before pan sharpening. When the image processing box appears, click on the 'No Histogram Adjust' radio button. Then select 'Process Image' This will cause the image to be pan sharpened without histogram matching and you will get the same result that I did.

The black areas are an artifact of the histogram matching algorithm. Certain pixel values will cascade all the way down to zero during the match of the two cumulative histograms. The matching algorithm still needs some help to overcome this issue which occurs only for some images under some conditions. The 'remove highlights' check box invokes an algorithm to solve a similar problem on the high end of the histogram i.e. unwanted cascading toward pixel value 255. I was also able to get rid of the black spots by unchecking the 'clip tails' checkbox but I was not happy with the color result either however. I got the optimum result by using the linear histogram matching instead of non linear matching. Use the radio buttons mentioned above to select this.

Histogram matching can be a very powerful aid but it can also cause unwanted side effects. Settings that work well for one image may not work for another. The defaults I set up work well for most Landsat pan sharpening tasks. SPOT images may benefit from different settings.

Where can I get more information about pan sharpening and PANCROMA?

I periodically publish articles on new utilities and special techniques at www.TERRAINMAP.com

Why does Digital Globe offer their data sets in both 8-bit and 11-bit format? How can I pan sharpen the 11-bit data?

To answer the question it is important to distinguish between dynamic range and bit depth. The dynamic range refers to the capabilities of the sensor while the bit depth refers to the capabilities of the display system. In order to pan sharpen or display data with a higher dynamic range the 11 bits must be rationalized into 8. PANCROMA offers two ways to do this. See the Instruction Manual for more information.

I have higher resolution grayscale imagery than the 15m Landsat panchromatic band. Can I use it to pan sharpen Landsat band data?

You can pan sharpen Landsat RGB data with any high resolution imagery as long as the following conditions are met:

• The high resolution image must be an even 2X, 3X or 4X multiple of the Landsat band resolution to pan sharpen using PANCROMA.
• The four images must be perfectly co registered

If you discover a useful source of high resolution imagery with another multiple of Landsat band data let me know and I can add the capability to handle it to PANCROMA.

I tried to open my GeoTiff format data files but PANCROMA does not read them. What is the problem?

PANCROMA cannot currently read compressed or tiled TIFF format. You can tell if the file is compressed by looking at the compression tag value reported to the echo screen. If it is other than '1' then the file is compressed. PANCROMA reads the more common stripped TIFF format but cannot currently read tiled format. Native Landsat and SPOT data is always archived in uncompressed stripped format. However, some GIS applications will save in compressed or tiled format. So if you are preprocessing in another application and then importing to PANCROMA, you may encounter this problem.

A possible remedy is to adjust the 'save' settings on your GIS export application if possible. Alternatively, you can save in JPEG, BMP or PNG format. You will of course lose any embedded georeferenceing information if you save in other than GeoTiff format.

How large of a data set can I pan sharpen with PANCROMA?

The program was designed to process Landsat data sets. Landsat multispectral bands are typically around 8000 by 8000 pixels while the panchromatic band is double this, around 16,000 by 16,000 pixels. It is possible to process larger data sets. The maximum size will depend on current program architecture, your hardware configuration, and the limitations imposed by the Windows 32-bit operating system that many will use. A panchromatic band size of 20,000 by 20,000 pixels may represent the practical upper limit at present. Larger data sets may result in 'out of memory' error messages followed by abrupt termination of the program.

PANCROMA has two special utilities designed to handle huge files. The 'Display One File' utility is set up to read band data directly from disk to a display color model, with no internal array. This will allow displaying a single grayscale band file of relatively large size. The 'Subset One Band' is designed the same way, allowing subsetting of single bands of similarly large size. As a result you can typically subset extremely large data sets down to a size within PANCROMA's processing capabilities.

I saved my georeferenced file using ArcGis, but I cannot open it using PANCROMA. What is the problem?

ESRI ArcGis exports Geotiff in tiled format. Tiled format is much less common than the stripped format that Landsat and SPOT data use. PANCROMA cannot read tiled images at the moment. I plan to add this capability as an upgrade soon. Until then you might try the following procedure, which was reported by Luke Pinner on the ESRI Support Center website for switching output from tiled to stripped:

Open C:\Program Files\Common Files\ESRI\Raster\defaults\tiff.pdf (note: file is NOT Adobe PDF format, it is ascii, open in a text editor) and change the following line:

create_tiled_images("Create Tiled Images"): "true" "Make newly-created TIFF images tiled

instead of

stripped" boolean "true" "false"; TO

create_tiled_images("Create Tiled Images"): "false" "Make newly-created TIFF images tiled instead of stripped" boolean "true" "false";

Loooots of other TIFF options in there including compression. Thanks to Matt Wilkie for pointing this out recently in an ESRI-L SUM.

Note:The file is loaded on startup, so exit and restart all ArcGIS programs before testing... And don't forget to make a BACKUP!!!

I saved a GeoTiff file using Global Mapper but PANCROMA would not open it. What is the problem?

Global Mapper saves GeoTiff files in compressed format, which PANCROMA is not able to handle at the moment. To fix this, change the default 'packbits' compression to 'no compression' in the Global Mapper GeoTiff export options dialog box before saving the file.

I tried to use the three-file cloud mask utility on my Landsat scene but I keep getting an error message stating 'Unexpected Row and Column Sizes'

The cloud mask utility, like most PANCROMA multi-file utilities requires exactly matched row and column input files. In the case of cloud masking, the Landsat TIR bands are approximately half the row and column count of the band files, but not always exactly. Sometimes they are off by one row or column. Before cloud masking, you must resize the files so that they have the required 2X row and column ratios before masking. You can resize downward use the PANCROMA 'Preprocess' | 'Resize Images' utility. See the Instruction Manual Section 31 for more information.

After loading the image in PANCROMA, some of the menus are still disabled, why?

Certain menu selections will initially be disabled. They will become enabled as you open input files and process them. For example if you open a single band file, then all the menu selections appropriate to processing a single file will become enabled, and all those that require two or more input bands will be disabled. When you open a second band file, the menu selections appropriate for processing a single file will become disabled and those that require two input files will become enabled, etc. When you are done processing a data set, select 'Close Graphics Windows and Reset' to start over again processing a new data set.

I'm trying to do a standard 3-band combination using Bands 3, 2, and 1 of a Landsat scene. I'm opening them up in PANCROMA in the order I want to combine them (in B,G, R order) and everything appears fine until I try to do the band combination. Then, I get two warnings 1) "The utility wants bands input in this order: blue, green. red" and then crashes with 2) "Unexpected row and column sizes. check your input sizes" I can display each band individually with no problems.

It appears that the problem is that all three of your input files are not exactly the same size, i.e. they do not all three contain the same number of bytes. PANCROMA has three levels of file sentinels. The first level checks to see if all three input files are the same size as mentioned above. The second level checks that the file characteristics are correct, i.e. that the file is uncompressed, the bits per sample is 8, the planar configuration is either 1 or 2, etc. The third level checks to make sure if the row and column counts all match up. The reason I do all this checking is that if the files are not consistent, it can cause an untrapped error. These are usually caught by the operating system but at best they would cause PANCROMA to crash by over writing an array or some similar misbehavior. PANCROMA apparently feels that the byte count in all three files do not exactly match and is declining to process them as a precaution. I have never seen such a mismatch with Landsat files downloaded directly from any archive source. However, if one band file is processed by another application, it is possible that the application is writing to the GeoTiff ASCII fields and causing one of the bands to be larger or smaller by a few bytes.

I uploaded PANCROMA version 4.95 tonight that will report the three file sizes if the sentinel is tripped. I also corrected the dialog box error message that reported a row and column size error when the problem is really the file size. One usually causes the other but not always.

If you would not mind downloading the new version and letting me know if the dialog screen reads something like:

File Checksum Error. The input files are not consistent .tif Landsat files. band1FileSize is: 169926
band2FileSize is: 169926
band3FileSize is: 169928 <--- Note different file size

Update: PANCROMA version 4.99 has a switch to disable the file size verification sentinal.

I am trying to subset two Landsat scenes for gap filling. I subsetted the first set OK. When I loaded the second set and entered the coordinate data I got an error message. What went wrong?

When you process the second set, just check the 'Select by Coordinates' radio button. Do not enter any values into the text boxes, as the correct values from the previous scene have been automatically entered into the text boxes. After you have checked the radio button, select 'Enter' and the second set will be subsetted to the same corner coordinates as the first set.