CONTACT MI
  Resellers
  Consultants
  MicroImages
  About MI
  Visiting
  Prices
  Send Email
  Reseller Resources

SHOWROOM
  Gallery
  Technical Guides
  New Features
  Testimonials
  Reviews
  World Languages

   View PDF Version (941 Kb)

Georeferencing.

As noted earlier, the tentative Part 2 of the JPEG2000 standard covers georeferencing.  Most of what is covered in Part 2 concerns other kinds of images and will not be pertinent to still images and your use of them.  But without a georeference standard we must make do with work arounds for georeferencing images until Part 2 is approved.  Needless to say, several different makeshift workarounds will appear from various sources and become part of the geospatial format quagmire, and we will all have to deal with these JP2 extensions for some years to come.  The most logical of these temporary approaches is to adopt the same patched up method used for other formats that do not directly incorporate georeferencing.  This is accomplished by creating and carrying along the georeferencing in an ArcWorld file of the same name as the JPEG2000 (*.jp2) file but with the extension of *.j2w.  This follows the ESRI convention of a TIFF’s *.tfw, MrSID’s *.sdw, TAB’s *.tbw, JPEG’s *.jpw, and so on.

V6.70 of the TNT products export the raster contents to a JP2 image file and create in the same directory a J2W georeference file using the ArcWorld format for the georeference. Only other programs that can read both these files will be able to use the TNT created JP2 file as georeferenced.  If you select a JP2 file in a TNT product, its corresponding J2W file of the same name will be sought and used to georeference this raster.  This J2W file may have been created earlier by the TNT product as part of an export or may be created by some other software that yields a JP2 file with a properly structured J2W file. 

TNTmips’ export of a JP2 file also creates an RVC Link File (*.rlk) in the same folder as the JP2 file.  Just as with other directly used, linkable file formats, this RLK file contains information needed to treat an external raster file as if it is actually a raster object in the Project File.  For example, the RLK file contains the pyramid layers for the external raster if it has none of its own. The RLK file created and associated with a linked JP2 contains the georeferencing for that JP2 file.  Thus, a georeferenced raster object or linked external raster exported to JP2 can be immediately linked and is georeferenced.   

Technical Characteristics.

Performance.

As noted earlier, a small RLK file is automatically created in the same folder as the exported JP2 file.  This file contains the georeferencing since it can not yet be stored in the JP2 file.  Since the RLK file is automatically available after exporting, it is automatically available for linking to the JP2 file.  No pyramid objects have to be created and maintained as JP2 files automatically have Multiple Resolution Levels (see below).  Thus, first time direct use of a JP2 file is almost as fast and efficient as using an uncompressed raster object of the same image.  A JP2 compressed image and its internal levels can be highly compressed so that reading this smaller amount of data offsets the time needed to decompress it.

Unlimited File Size. 

JP2 library used in the TNT products supports the same unlimited file size as raster objects in all the TNT products.  It is subject to the limits on single file size imposed by your operating system, which is 4 Tb for WNT and W2000 using NTFS (NT File System), 4 Gb for Mac OS 10.1.5, and 4 Tb for the most recent versions of Linux (for example, RedHat 7.30) and UNIX but 2 Gb for older versions of both.  You may also be limited to an input raster object of 137 Gb by your 28-bit hard drive controller.  However, 48-bit LBA (Logical Block Addressing) drive controllers (up to 144 Pb files) are beginning to appear on good quality motherboards and add-in controller boards.  Temporarily your maximum output file size is limited (after compression) to about 2 times your real memory above which the process will go virtual and, thus, slowly.  This real memory requirement is likely to be resolved by improved memory buffering in the next release of the Kakadu library.

Data Precision.  

Each image raster color component compressed can range from 1- to 38-bit signed or unsigned integers.  Floating point and complex rasters (dual floating point) must be converted to integer data values before JPEG2000 export can be applied.

Color Composites.

RGB color composites can be exported directly to a JP2.  During the process the color components will be separated, compressed individually and then placed in a single JP2 file.  Scanners, other image sources, and software are becoming available that produce and/or work with 48-bit color composites.  Since these are broken down into 16-bit rasters for compression, they are well within the 38-bit precision of JPEG2000 compression process.

Other Standard Features.

Mixed Level of Detail.

Multiple levels of detail can be created during export for JP2 files destined for use in other software.  These are not needed if their intended use is in the TNT products. 

Progression Orders.  

Progression order (lowest resolution first proceeding to highest resolution) is created in the JP2 file during export for use in other software. This order is not needed if their intended use is in the TNT products.

Multispectral and Hyperspectral Images.

Exports of multispectral and hyperspectral images are treated just like color composites as outlined above.

Multiple Resolution Levels.

The multiple resolution levels defined in the JPEG2000 still image specifications function like the pyramid structures in a TNT raster object.  They are automatically created during export.  As a result, TNT products use these directly from the JP2 file and no pyramid layers ever need to be created.  This is similar to the way TNT products make direct use of ECW and MrSID files. 

Not Supported.

Streaming.

This refers to the construction of JP2 files for controlling the order in which the image content is sent from a server.  It does not necessarily refer to starting at low resolution and increasing the detail.  It might mean to stream out spatially providing the face before the background or the advertisement before the content.  It has no application in TNTmips.

Region of Interest.

This Region of Interest is not the same ROI concept used in the TNT products.  It means varying the level of detail from 1 portion of the image to another.  For example, the center of the image might have more detail or some other area where zooming in and more detail is expected.

Other.

The security features of JPEG2000 such as watermarking, labeling, stamping, and encryption are not implemented.

Landscape Builder.

Additional Background Materials.

As usual, this MEMO introduces the features that are new in V6.70.  However, if you are using the Landscape Builder for the first time, please review this same section in the MicroImages’ MEMO shipped with your V6.60 or posted at www.microimages.com/relnotes/v66/ for additional introductory materials not duplicated here. The Tutorial booklet entitled Building 3D Landscapes provided for the first time in V6.70 is current with this version of the Landscape Builder.

Multiple Textures.

The TNTsim3D / Multiple Textures section of this MEMO describes the various uses of several texture layers in a simulation.  The Landscape Builder now provides the additional features needed to add multiple texture layers into a single Landscape File.  The procedure operates similarly to that in V6.60, but after you have added a texture, it will prompt to determine if you want to add another texture layer.  If you already have a Landscape File prepared in V6.60 or V6.70 you can now also select and open it in this process and add additional textures.  This operation is illustrated in the attached color plate entitled Preparing Multiple Textures for TNTsim3D.

Handling Null and No-Data Areas.

The raster objects used in any TNT process can be any kind of irregular shape (for example, islands and coastlines) with any kind of interior holes (for example, lakes and masked out areas).  Rasters can be used in processes even though their extents only partially overlap.  V6.60 of the Landscape Builder dutifully transferred these null areas into terrain and texture layers in a Landscape File and handled the areas of mismatched extents.  However, the results of the Landscape Builder and TNTsim3D did not work properly together to render these no-data and null areas.

The Landscape Builder has now been modified to fill terrain no-data areas with the minimum real value from the terrain.  The resulting simulation may show a “step” down to this minimum area at the original edge of the terrain but the texture overlay will now render over the entire area dropping into any null terrain areas.  This change also would allow you to create a simulation from an image of an island with the surrounding sea and an elevation model covering only the island area.  In TNTsim3D the island would appear to rise from the flat ocean surface (the ocean areas of the image rendered over the flat, minimum value areas of the output terrain raster).

If the raster object selected for the texture layer does not cover the full area of the simulation (which means, has any extent less than the terrain), its areas of no coverage were assigned in V6.60 to be equal to the null value.  But, the cell value designated as null in that raster object and, thus, in the texture layer could be the minimum data value for that raster data type (which means, R=0, G=0, and B=0 for 16-bit and 24-bit texture rasters).  This meant that if such a raster object also had areas of this total black they would end up with the same texture cell value as designated for the nulls.  Thus, these valid black areas would become transparent in TNTsim3D.  To avoid this special condition, the Landscape Builder now makes a slight adjustment to the color of any total black texture areas so they are assigned a texture cell value that is indistinguishable from total black  (which means, not 0,0,0) and yet different than the null value.

Stalked and Bill Boarded Point Symbols (a post V6.70 prototype feature).

The initial implementation permitting you to use points in a vector object in your simulation is nearly complete, but the release of V6.70 of the TNT products could not be delayed for this latest TNTsim3D feature.  Watch the special TNTsim3D pages at www.microimages.com/documentation/CP67tntsim3D.htm for information about the release of a TNTsim3D with this feature, which can then be downloaded from that same page.

When these tables can be used in TNTsim3D, you will select the points, attributes, and styles from a vector object and add them to your views in the Landscape Builder.  Thus, you will need to download the corresponding Landscape Builder that has already been modified to prepare these tables.  Note, however, it has not yet been determined when, if, and how pins might be added, moved, or edited during a simulation either interactively, by editing the table, or dynamically.

Map Projections and Coordinate Systems.

“Michigan GeoRef” coordinate system used in Michigan statewide mapping is now supported.

“North Sahara” datum used in Algerian mapping is now supported.

“ELD-79” datum used in Libya and Tunisia is now supported.

“Hartebeesthoek 94” datum used in South Africa is now supported.

“New Zealand Geodetic Datum” is now supported.

Raster Extract.

Reinstate as an option the older pre-V6.60 auto-naming system applied when performing extraction of multiple raster objects by vector polygons.

Raster Import/Export.

JPEG2000 (JP2).

JPEG2000 *.jp2 files can be imported and exported with georeferencing.  This is discussed in detail in another section of this MEMO.

TIFF/GeoTIFF.

If a GeoTIFF file contains control points but no projection, use the projection set by the user as the default for import instead of “arbitrary.”

Export to GeoTIFF will now automatically default to the closest available GeoTIFF datum rather than none at all.  For example, GeoTIFF only specifies a single NAD27 datum code.  TNTmips supports multiple versons of this datum with the various transformation parameters used in different geographic locations (nations, provinces, counties, …) to meet higher local accuracy requirements.  Any of these custom versions of NAD27 will now revert to the single NAD27 datum during their export to GeoTIFF.

Import and export signed integer 16- and 32-bit TIFF/GeoTIFF files to/from raster objects.

Import TIFF/GeoTIFF files having more than three bands, such as 4-band multispectral QuickBird images, into raster objects.

During the import of CMYK (Cyan–Magenta–Yellow–Black) bands from a TIFF file, they are automatically converted to separate RGB raster objects.

ER Mappers’ ECW.

An option is now provided for exporting to ER Mapper’s ECW file using their “Optimize for Internet Display” setting.

ASTER-HDF.

Import the metadata for the ASTER-HDF file and use the georeference it contains as it is more accurate than the standard HDF georeference for ASTER images.

Vector Import/Export.

W3C’s SVG.

Vector objects can be exported to Scalable Vector Graphics (SVG) files.  SVG files and their pseudo export or conversion from TNT map layouts and all the different kinds of objects and groups they contain is discussed in detail in another section of this MEMO.

ESRI’s Coverages.

The ARC/INFO coverage selection procedure has been improved.  In V6.70 and earlier it is used to select individual *.adf files in a coverage directory.  It then assigned the name of this ADF file as the name of the new vector object.  After V6.70 was completed this import process was revised so that now you select only the coverage directory, the ADF files in it are automatically used, and the vector object is named to match the coverage directory.  This change enables multiple coverages (which means., multiple directories) to be selected and then all imported at once.  The “INFO” directory is not considered a “coverage” and can not be selected as an importable coverage file. These modification were made after the V6.70 CDs were produced and you will need to obtain a patch from microimages.com to use it.

CAD Import/Export.

W3C’s SVG.

CAD objects can be exported to Scalable Vector Graphics (SVG) files.  SVG files and their pseudo export or conversion from TNT map layouts and all the different kinds of objects and groups they contain is discussed in detail in another section of this MEMO.

MapInfo’s TAB.

The import and export of various MapInfo formats has been improved and exports can now be in either “feet” or “meters.”

Surface Modeling.

Improved TIN Topology.

Delaunay triangles are the basis for the topology in a TNT TIN object.  The 3 points making up a triangle form a Delaunay triangle if, and only if, the circle that passes through them contains no other vertex of any other triangle.  Preserving Delaunay triangles insures a good representation of the surface with a minimum of triangles and other useful properties such as “fat triangles all striving to become equilateral.”  This topology also provides the basis for accurate operations in other related TNT processes.  V6.70 improves the TIN computation process to assure this topology is always maintained.

In V6.60 some unusual local point distributions could cause local triangles to be formed that were not Delaunay triangles.  For example, in an unusual scenario, lidar flight lines have very dense geopoints along the lidar lines and lines that are widely distributed.  In this situation, there can be multiple points in the raw data that are very close to each other, fall nearly on a straight line, and yield very long sliver triangles in creating a TIN in V6.60.  This required that the points be prethinned somewhat to eliminate this condition.  However, this situation is now handled in V6.70.  In typical data distributions, these sliver triangles were absent or few in V6.60 and caused only very local anomalies in subsequent processing.  For example, a local small point of inflection would occur in a contour line or slightly dimpled DEM.  However, this in turn propagates into the watershed physiography computations.  In the special cases where this topology was incorrect, the process computing Voronoi polygons would not finish, as they are formed by connecting the bisections of the Delaunay triangle edges.

Better Breaklines.

Breakline insertion and preservation in TIN objects has been improved. However, you must download and install the latest post V6.70 patch to use this improvement. 

Breaklines represent point, line, and polygon features of a surface whose elevation profile is known.  Lake margins, island coasts, drainage lines, ridge lines, mountain peaks, and similar features are representative of breaklines.  When a TIN is built from points and these breaklines are used in the process they become hard edges for the triangles making up the TIN.  In other words, they must be preserved in the triangle edges of any TIN object and through any subsequent processing.  When the TIN is used to represent a surface for contouring, displaying, Voronoi triangles, and so on, these hard edges must lie on the surface and force it to have the necessary inflection such as a drainage channel.

Hard edges can be created in a TIN object in V6.70 and earlier from features in a vector object.  However, some of the hard edges could be lost or violated in subsequent manipulation of that TIN object. This condition has been corrected.  Also when a TIN object’s topology is rebuilt after an edit or other modifying operation, there is a strong tendency for sliver polygons to form along the hard edge of existing triangles as these sides can not move in X-Y position.  This tendency is now controlled so that hard edges are preserved if the TIN is rebuilt.

Harmonic Series Analysis of Multidate Rasters.

Background.

First of all, it must be clearly stated that this is an experimental procedure.  Its addition to TNTmips was sponsored by a client who had already assembled an appropriate dataset.  However, it has general interest in studying subtle ecological, climatic, land use, and other changes with imagery and other kinds of spatial data over seasonal, annual, or even longer time intervals.  Its potential power is that it integrates together many successive sets of spatial observations to look for change or changes in the rate of change.  It is not using simple gross change analysis between 2 dates.  Its disadvantage is that assembling that kind of database can be expensive, time consuming, and tedious.

The sample data provided to MicroImages to test the development of this new process was a set of Project Files containing 55 NDVI (Normalized Difference Vegetation Index) rasters.  These were prepared from a collection of SPOT images processed to provide the NDVI rasters at 10-day intervals over a period of 2.5 years.  Certainly this was a costly set of images to acquire, and it took some time for the client to assemble these Project Files in TNTmips. For brevity, the set of rasters made up of the property of interest for each image date will be referred to hereafter as a CMT (Calibrated Multi-Temporal data set).  This new analysis procedure handled the computations on this 3 gigabyte CMT efficiently by both Fourier methods discussed below.

The following paper was used as a reference during the implementation of this new process.

Harmonic Analysis of Time-Series AVHRR NDVI Data. by Mark E. Jakubauskas, David R. Legates, and Jude H. Kastens.  Photogrammetric Engineering and Remote Sensing.  April 2001.  Vol. 67, No. 4.  pp. 461-470.

Abstract.  Harmonic analysis of a one-year time series (26 periods) of NOAA AVHRR biweekly composite data was used to characterize seasonal changes for natural and agricultural land use/land cover in Finney County in southwest Kansas.  Different crops (corn, soybeans, alfalfa) exhibit distinctive seasonal patterns of NDVI variation that have strong periodic characteristics.  Harmonic analysis, also termed spectral analysis or Fourier analysis, decomposes a time-dependent periodic phenomenon into a series of sinusoidal functions, each defined by unique magnitude and phase values.  The proportional variance in the original time-series data set accounted for by each term in the harmonic analysis can also be calculated.  Magnitude and phase angle images were produced from analysis of the time-series NDVI data and correlated with information on crop type and extent for the region to develop a methodology for crop-type identification.  Crop types occurring in southwest Kansas, including corn, winter wheat, alfalfa, pasture, and native prairie grasslands, were characterized and identified using this technique and biweekly AVHRR composite data for 1992.  For crops with a simple phenology, such as corn, the majority of the variance was captured by the first and additive terms for the harmonic analysis, while winter wheat exhibited a bimodal NDVI periodicity with the majority of the variance accounted for by the second harmonic term.

Concept.

This procedure performs a 1-dimensional Fourier analysis of the irregular curve formed by many sequential time varying observations of some biophysical property of a single ground cell.  It decomposes this irregular curve for that single cell into the many sine waves of differing magnitude and phase that would need to be combined together to reconstruct that irregular curve’s shape.  This process computes these properties independently for each cell in a 2D spatial array of ground cells, usually, but not necessarily, derived from a series of multi-temporal images.  It outputs a series of rasters of matching cell size each of which contains the magnitude and phase of one matching period of this collection of sine functions. 

This process, using either Fast Fourier Transform (FFT) or the slower Fourier Transform (FT) analysis discussed below, will create many, new raster objects.  By default it will create 3 of these new raster objects.  Optionally, it can create more output rasters, each for a sine function with frequencies increased by a power of 2.  The maximum number of output raster objects is the number of input rasters in your CMT.  Creating more than 3 does not effect the time to compute FFT or FT as they are always all computed anyway.  It merely means you need more drive space and a little more time to write each to the drive.

Every output raster is a complex number raster object containing the magnitude and phase components of the sine wave for each cell.  The first raster contains the average value for all the observed values for each cell as its magnitude component and its phase is zero.  The second raster contains these properties for a sine wave of the largest period, the next has a period of 1/2 the first, the next 1/4 of the first, 1/8, 1/16 and so on (1/n to the power of 2 where n =  1, 2 ,3 , 4 …). 

You are primarily interested in the displaying and interpreting the magnitude of these rasters.  For a CMT spanning 1 year, the magnitude in the first raster is the average value for each cell for all of the time interval represented in the CMT.  The second raster contains the magnitude/phase for the annual cycle.  The third raster is the magnitude of 1/2 the annual cycle. The fourth is the magnitude of 1/4 the annual cycle, which might be of particular interest from a seasonal viewpoint, and so on. Things that cause changes in the ground cells during these cycles will control the relative magnitudes of the results for each cell.

Sources of Input Data.

Typical image sources that can be frequently and economically assembled for preparation of a CMT would be AVHRR, MODIS, or meteorological images since they are routinely collected everywhere. AVHRR images have been used by various research labs for these and related kinds of temporal analyses.  AVHRR is collected daily and can be used to compute a vegetation index and assembled into a CMT.  For example, many years of NDVI rasters can be assembled from AVHRR images for 10-day intervals.  If a particular required date is cloudy, then use the image of the day before or after, adjusting the next interval to be longer or shorter to compensate.  If a series of sequential days are partially cloudy, then images for several sequential days can be composited into 1 using a common AVHRR cloud identification procedure.  If more frequent images can be assembled from aircraft or ground sensors for just one season, such as at 3-day intervals, they can also be used to prepare a CMT to study agricultural or climatological changes (apply only the FT method described below as the longest annual period is not complete).

Preparation of Input Data.

This process requires the collection and assembly of many raster objects in a CMT each of which represents 1 set of observations of a time calibrated variable that changes in a periodic fashion with time.  A typical starting point would be to assemble a collection of multi-spectral satellite images of a site at frequent sequential dates throughout 1 or more annual cycles. Another criterion, which must be met in your CMT at this time in this process for computational reasons, is that the time interval between each image must be approximately equal. These images should be processed so as to be coregistered with a common extent, projection, and ground cell size.  Next the multispectral image values for each cell in each image must be converted from image values to the biophysical property of the surface you wish to study. 

Examples of biophysical properties of a surface that can be computed or at least reasonably estimated from multispectral satellite images are vegetation indices (for example, NDVI6, NDVI7, and various others), absolute reflectance in 1 band, surface albedo, real or radiant surface temperatures, and so on.  Harmonic analysis permits you to study how these properties change over time.

It is important to understand that raw image pixel values, cell or pixel radiance, classified cell values, and similar uncalibrated rasters can not be used in a CMT or this analysis.  These cell values may vary nicely within an image but they have no known relationship between images as a function of time.  Or put another way, if a cell has a value of 100 for one of these properties today and a value of 200 tomorrow, this does not necessarily mean that the surface has changed at all, perhaps its only a change in the gain of the imaging device (for example, an aperture change) or some subsequent processing transformation.  Attempting to use a CMT with these kinds of multi-date properties will produce totally meaningless results.

Choosing A Fourier Analysis Computational Method.

What’s the Difference?

The Fast Fourier Transform has stringent requirements in the structure of the data whether it is applied in a 2D spatial, single image analysis or as in this case, to a one-dimensional analysis of the multi-temporal values for each cell.  The Fourier Transform does not impose these special restrictions and is computed using trigonometric floating point computations. If these restrictions are met, and FFT can be applied, its computation is entirely by addition and subtraction, which is hundreds of times faster on any desktop computer.

Fast Fourier Transform (FFT).

The Fast Fourier Transform was developed by engineers as a practical implementation of the slow Fourier Transform for application in electronic signal processing and earlier image analysis.  While an FFT is fast, it also places some more restrictions on any data sets to which it is applied.  If it is applied in the spatial analysis (2D) of a raster (which means, using the Fourier Analysis process elsewhere in TNTmips) the rows and columns of that raster must be equal to a power of 2 and equal to each other.  When FFT is applied in this multi-temporal or 1-dimensional analysis of a CMT for harmonic analysis these conditions must be met.

The number of multi-temporal values for each cell must be a power of 2.  An option to create this condition in a CMT is discussed below.  It resamples the time-variant real values for each cell to create a power of 2 new interpolated value.

The CMT must contain 1 or more complete periods for the longest periodicity present in the CMT.  For many applications this will be 1 or more annual cycles, which start at any calendar date and must end at approximately the same calendar date in the final year. Do not include any partial periods of the largest periodicity included in the CMT. A linear trend in all the multi-temporal values for each cell represents an incomplete longer period cycle.  Compensating for this is why the linear trend removal option discussed below is required for a FFT.

Fourier Transform (FT).

The FT is much more tolerant than the Fast Fourier Transform and does not require the restrictions outlined above for the FFT approach.  FT does not require that the raster making up the CMT total to a power of 2.  The FT also does not require that an integer number of the largest period be represented in the data range of the CMT.  It can be applied to several complete annual cycles that do not begin on the same calendar date or season.  It can also be applied to part of an annual cycle.  Thus linear trend removal may or may not be useful in the FT approach.  However, even the FT results will not be meaningful if applied to a CMT containing a small number of rasters.

Unfortunately, even on fastest desktop workstations the FT requires a significant amount of computation time for each cell in the CMT. For this reason it had not even been previously implemented as a function in the TNTsdk before this process was implemented.  As a result, and depending upon the size of your CMT, do not start an FT process unless you can dedicate TNTmips to it for a few hours.

A practical approach to using this kind of analysis might be to combine the FFT and the FT.  First experiment with your CMT in the FFT method and its various approximations if your CMT has enough temporal range so that it can be temporarily used to cover at least 1 complete annual cycle.  The FFT method is fast enough so you can experiment with it and the makeup of your CMT.  When the results are promising make a slow FT run without resampling required by the FFT to get the best cell by cell results for the entire multi-temporal range covered in your CMT.

Linear Trend Removal,

Your CMT may contain a long term linear trend that is longer than the principle periodicity of the CMT (which means, longer than an annual cycle).  This trend appears to the FFT to represent an incomplete multi-temporal cycle, which is not permitted in the FFT process. If your CMT represents several annual cycles, the trend might be due to global warming, the sun spot cycle, a gradual change in average precipitation, or something else.  A trend in a CMT representing a period of only 1 year might result from a gradual deterioration in the imaging system.

Use the Linear Trend Removal option to remove the trend from your CMT during the computation of the FFT.  It will compute the average of all the multi-temporal observations available for a cell and then statistically determine the linear trend in all these values.  Each original multi-temporal value for the cell is then adjusted by adding or subtracting the difference between the average value and linear trend value for that cell.  If the trend value is greater than the average, this difference is added to that cell’s original value.  If the trend value is less than the average, this difference is subtracted from that cell’s original value.  This has the effect that the average for that cell does not change but the linear trend is no longer present.

Smoothing Filter.

Anomalous changes or noise may occur in the observation for any single cell on a specific date.  For example, all images have some noise present even if it is not obvious.  Smoke from a local fire might obscure some local cells on one of the available dates as would atypical, temporary areas of standing water just after a rainstorm.  

In harmonic analysis you are not interested in these 1-date anomalies for each cell, which can occur anywhere on any date (which means, they represent numerous spurious high frequency events). This smoothing option fits a local sliding curve to the multi-temporal observations for each cell during the FT or FFT processing and uses this curve to recompute the adjusted value for each cell.  Note that this approach to smoothing is another reason why the time intervals between values should be approximately equal.

Resampling in the Time Domain.

If the number of rasters in a CMT is not a power of 2, then these values must be created for each cell by interpolation between the real observations.  This is accomplished during the FFT processing by choosing the option Linear or Cubic, which will fit a linear or cubic spline to the real observations spaced at the intervals needed to create a new interpolated set of multi-temporal values totaling the next power of 2 greater (for example, 58 real observations become 64 interpolated values).  If the number of real values is equal to a power of 2, then selection of Linear or Cubic will be ignored.  This interpolation process does not have any knowledge of how the multi-temporal observations for the cell might be spaced in time and assumes they have equal time intervals.  This is another reason to make sure the time intervals between the rasters in your CMT are as close to equal as possible.

Results.

Using either the Fast Fourier Transform or the slower Fourier Transform, this process will by default create 3 new raster objects from the CMT.  You may elect to create more output rasters; the computation time is the same, but a little more time is required for writing the additional rasters.  Each of these new rasters is a complex raster object containing a magnitude and a phase component for every ground cell. The magnitude of these rasters is of the most interest and when displayed, show the changes taking place in the ground area in that property for an annual cycle, a seasonal cycle, or some other shorter cycle.   

The first raster, usually referred to as the 0 component, has an magnitude for each cell equal to the average of all the values for that cell and a phase everywhere of zero.  A display of the magnitude of this raster object would compare the accumulation or loss of the observed value over the time interval involved.  For example, displaying the average of the NDVI would indicate how much green vegetation biomass each cell produced over the year relative to all other cells.  Displaying the average temperature would indicate climatic differences in each cell.

The second raster object contains the magnitude/phase of the longest period sine wave in the observed values.  If your CMT covers just 1 year, then displaying this magnitude shows the amount of variation in that observed parameter over 1 year.  For example, did the temperature of each cell fluctuate more or less widely than all the other cells.  If your CMT covers 3 or 4 years, this can be a longer period such as an 11 year sun spot period if it gradually effects the temperatures of all the cells values all the time in a subtle fashion.  Alternately, you could make a CMT for each successive and equal annual period, analyze each separately, and then compare like magnitudes for the annual periods to examine change from year to year.  For example, you could display 3 successive years magnitudes in red, green, and blue to enhance the changes from year to year.

The third raster contains the magnitude/phase of the sine wave that has a period of 1/2 that of the second raster.  If the CMT covers just 1 year then the period of this sine wave is 1/2 a year.  Its varying magnitude represents the variation in the observed property with a period of 1/2 year after the annual variation is removed.

The fourth raster’s magnitude/phase are for the 1/4 annual cycle, the fifth for 1/8, the sixth for 1/16 and so on.  The magnitude of the fourth  (which means, period = 1/4 annual period) may be of particular interest related to changes in seasonal variability between the cells. 

Possible Improvements.

Improvements could be made to this process by MicroImages if you can assemble a CMT that needs them and if you understand the general application enough to interpret your results.  The most important would be to add a date and time to each raster in the CMT.  This would permit improved time domain smoothing methods to be applied to the multi-temporal observations, such as splining each cell’s observation spaced at their actual and unequal intervals.  This would permit the creation of suitable CMTs from rasters that span a suitable period of time in sufficient frequency, but do not automatically have nearly equal time intervals.  Assembling a CMT with nearly equal intervals is possible with low resolution satellite imagery, such as meteorological, AVHRR, or MODIS images.  These are collected daily permitting shifts of 1 or 2 days for cloudy conditions.  Such conditions can knock out a whole interval for less frequent orbital coverage devices, such as Landsat (for which there are many years of multi-seasonal imagery archived).  When these unequal time periods are involved, the curve fitting could take into account the actual time interval and the curve interpolated to produce equal time intervals of a power of 2 for the FFT or without for the FT approaches.  

* Buffer Zones.

You have reported that the TNT buffer zone process accomplishes tasks of a magnitude and complexity that could not be completed in competing products at all, let alone maintaining topology.  As a result, this process is continually being challenged and tested where their buffer zone problem cases are imported into TNTmips, buffered, and exported back to them.  Since the computation of buffer zones is a key process in many geospatial analyses, it has been rewritten for V6.70 to make it even more powerful and flexible by the addition of many new, unique features.

New Interface.

Control Panel.

The Buffer Zone Analysis dialog has been completely revised.  Tabbed panels are now used to present the control/selection options separately for Points, Lines, and Polygons as well as a Preferences panel.  This provides a basis for controlling how the new features are applied to each element type while simplifying the appearance of this dialog.

Integrated View Window.

A standard View window is now opened in this process to preview the elements you have selected to buffer from a vector object by any method.  As usual, this is a standard View so you can use all the visualization, multiple object overlay, interactive or query based tools, and other standard features to select elements for buffering.  You can use this view to visually confirm that you have selected all the elements as you planned.

Next you can apply your buffer analysis to these elements and the buffer results will be displayed superimposed on the selected elements.  In this fashion, you can immediately preview the results to determine how the buffer polygons fit the selected elements and how they inscribe or match features in any other layer in the view such as an orthophoto. Or you could use the attribute table inspection tools to view the new records, tables, and their relational tree-like structure. 

Use this new View window and the new control dialog to interactively refine your selection and buffer zone operation, and fine turn your results.  Each time you apply the changes in your procedure the view is refreshed and the computed buffer zones saved to a temporary file.  When you are satisfied, you can save the temporary buffer zone object as a CAD or vector object.

Advanced Procedures.

None, All, or Selected are now options to select the elements to buffer.  Buffer distances can be derived for each element as All Same or By Script.  Using a script permits an independent distance to be computed or logically selected (for example, theme ranges) independently for every individual element.  Using a computed field to derive an attribute also provides a means of varying the distance by element.  These are only a few of the many new ways in which buffer zones can now be specified.  The color plate entitled New Buffer Zone Features illustrates and discusses additional options.

Selection of the elements to buffer by query or other methods can now use any of the attached attributes.  V6.60 permitted only the selection of elements by the key field, which often required a lot of unnecessary table manipulation.  This also permits the creation of buffer zones for all elements attached to records with the same value for the selected attribute(s).

Buffer zones can be merged for all elements or kept as separate and overlapping polygons for elements with different attributes.  Merging them will amalgamate overlapping polygons into single polygons regardless of their attributes.  Separating individual element’s buffer zones by attribute into independent polygons with different attributes is useful for CAD applications where topology is not wanted and separate, overlapping polygons are common.  On the other hand, overlapping polygons saved to a vector object will have their topology formed.  In other words, 2 overlapping polygons will become 3 with the new polygon representing the area of overlap.  In all these cases a variety of record/table management capabilities are provided to control how attributes are transferred to the new vector or CAD objects.  These and other combinations are illustrated in the attached color plate entitled Separating Buffer Zones by Attribute.

* Polygon Fitting.

Polygon Fitting, Buffer Zones, and Surface Modeling are all examples of important first steps in geospatial analysis and data mining.  These processes and others convert point observation to areas that can then be tested for their association with other spatial objects (which means, images, maps, surfaces, …) by multivariable analysis, vector combinations, visual inspection, and other techniques. 

Similar to the other TNT point-to-area conversion methods, the points used for fitting can now be selected by the select tool, the GeoToolbox, by query, or by record from an attribute table.  This process can now, in a single pass, create sets of associated polygons in one or separate vector objects (one for each set) with attributes.  In a single pass, it can also create sets of multiple polygons in a CAD object with attributes or be saved as separate CAD objects.

Topological Considerations. 

The term “sets of associated polygons” means that each set of disjoint points assembled by a query or other means can be fit with one or more disjoint polygons.  Since your query may assemble several sets of disjoint points you can get several sets of potentially overlapping polygons from one fitting operation.  If these results are saved to a CAD object, the polygons making up each set are saved as a single element called a multi-polygon.  In a CAD object, the multi-polygons can overlap but can subsequently be accessed by group or as individual polygons.  If you choose to save all sets of associated polygons into a single topological vector object, all the individual polygons must be intersected together creating many new non-overlapping polygons in that object.

Fitting By Attribute.

A database may contain a complex relational structure associated with the point observations it represents.  Polygon Fitting can now mine that database directly in a single pass.  The attribute selected for separating polygon groups can be a simple field or a computed field that makes use of complex queries to define the relationships that must exist between the points if they are to be included in forming each set of polygons.  You can also use any means available, including complex queries, to select the points used for polygon fitting.  The majority of the points tested can be completely omitted by the query.  Thus, the polygons that do result represent a few sets of points with simple or complex interrelationships.

When you apply a fitting operation, the results are saved to a temporary file.  These sets of polygons are also displayed immediately over all the original point positions (omitted or included) or any other image, vector, or other object, and you have access to the TNT visualization tools. Change your query or other standard point element selection procedure until you are satisfied with your results.  Use Save As… to save the contents of the temporary file to a CAD or vector objects(s). The attached color plate entitled Polygon Fitting By Attribute illustrates a simple example of this kind of application. 

Sample Application.

A simple example will illustrate how this works to mine a database. Suppose the relational database has 10,000 individual geopositions of where 100 different identified trucks have been (for example, from cell phone or GPS locations).  The process will pass through this database once and create 100 sets of polygons (1 or more polygon in each set) based on the selected attribute (for example, license plate) that will separate each truck’s position records by its identity and fit 1 or more polygons to that vehicles’ locations.  These sets of polygons represent the area(s) where each of these vehicles operate and can be saved to a vector or CAD object(s).  This object(s) can then be combined with other spatial variables to determine why these vehicles operate in their respective areas.

If the polygons are written out to a CAD object, you will have 100 collections of multiple but disjoint polygons in that single object.  They can overlap, since this is a CAD object. However, each group (multi-polygon) will have its own shared attribute associating it with the vehicle and whatever additional attributes were brought along for later use in a subsequent analysis.  If you choose to save the result in a single topological vector object you may create many new polygons with multiple vehicle records attached to each for every area of overlap.

Transferring Attributes.

Attributes were not transferred in the Polygon Fitting process in V6.60.  Thus, it required some tedious manipulations to get attributes organized and attached to these polygons in a vector object.  V6.70 now transfers these attributes to the new polygons.  This is illustrated in the attached color plate entitled Transferring Attributes in Polygon Fitting.  However, care should be taken or you can quickly create some complex sets of attributes when you create a single topological vector object.  For example, if your polygon fitting yields many sets of polygons with common areas of overlap, then the many new overlap polygons formed in a vector object could have many records attached to each. 

Saving a result into a single topological vector object is most effective when you select elements and attributes that tend to produce sets of polygons that isolate areas.  This is why it is important to view the results of your choice of fitting algorithm and point selection before saving them.  If you want to save sets with lots of areas of overlap, consider their future use and the possibility of saving them as a CAD object.

Transfer Attributes.

Lines to Polygons,

A Split At Border operation has been added for use when transferring attributes from lines to polygons.  The standard line attributes table that is attached to the polygons by the operation is modified so that the length reported is only the length of each line that falls within that polygon.  For example, you make a grid cell vector object and transfer attributes from lines to polygons using the Split at Border operation to have the length of the roads in each grid polygon attached to that polygon.  You must download and install the latest post-V6.70 patch to use this improvement.

Handles High Vertex Polygons.

The transfer of attributes has been impractically slow when the process encountered individual polygons with 100s of thousands of vertices.  Yes, it has even been applied to single polygons with millions of vertices.  Under these circumstances, the process was not efficient and took hours. It has now been streamlined and is now practical to use on these messy polygons.  However, you must download and install the latest post-V6.70 patch to use this improvement.

* Spatial Data Editor.

Improved label positioning and editing are the principle new features added to the Spatial Data Editor.  These new methods use a combination of an automated preliminary positioning followed by an easy inspection and manual repositioning by you.  This 2-step procedure is the most efficient approach to high quality label placement with today’s software tools.  A new Tutorial booklet entitled Advanced Vector Editing has been provided to introduce these procedures in detail.  As a result, they will only be outlined here in concept and their use is expanded upon in this booklet and in the attached, illustrative color plates.

Automatic Polygon Labeling.

Soil, geologic, vegetation, image classification, and similar vector objects can contain many polygons.  Automatically positioning labels in these polygons for the various font size and display scales you might select is a complex undertaking.  This is a task that you can manually accomplish, but it is tedious at best and very time consuming, as a single vector object for one map could have a thousand polygons.  Current automatic label placement techniques can place many of the labels in a satisfactory position but can not provide for 100% quality placement for electronically generated products. There are simply too many conditions that can occur due to the wide variety of display layers, scales, fonts, polygon merging and dividing, and so on that might be used in a geospatial analysis system. 

Automatic label generation now attempts to fit the polygon label inside the polygon at its widest horizontal position but with a bias toward the centroid.  If the label will not fit in this width, it selects one of the adjacent (common boundary) polygons for the label and adds a leader line.  As a last resort, it will place the label over the center of the polygon regardless of its width (a common result for very small polygons).

Your selection of label size can have a dramatic effect on the success of this preliminary automatic placement.  Turn on the Preview option to experiment with label size and font type.  After you have achieved the best automatic placement and appearance possible, save the placement with the vector object and move on to reviewing and repositioning those labels that still have an unacceptable position.  This automated step and some illustrated results are contained in the attached color plate entitled Auto-Generating Vector Polygon Labels.

The Auto Generate Label operation now also provides a Clip Under option.  This will clip open a hole for the label text in the polygon fill and every other element in that layer.  Be sure the text style used for the label has Vertical Alignment set to Center for correct placement of the clipping box.

Automatic Line Labeling.

The automatic labeling of lines also has some new features to improve the quality of your display and maps.  All of these are illustrated in the attached color plate entitled Auto-Generating Vector Line Labels.  Again you should think of this automated scheme as doing most of the work so that you can use the new line positioning features to refine and finalize your line labels.

Urban areas have many long streets that are broken by the nodes for crossing streets into many short line segments.  Drainages also have many short reaches broken by nodes at connecting links. These are two of the many network-oriented vector objects that in V6.60 would produce multiple labels along adjacent lines with the same attribute.  The illustration at the top of the color plate shows this earlier result.  Now you can assemble these into longer virtual lines by their attributes and treat them for the purposes of labeling as a single long line with the label placed near its center.  This improved, new, thinned placement is illustrated on the same color plate.

The way labels conform to an irregular line shapes at the automatically located position can now be selected to be

• Exact, where the labels flow along the line’s shape with tilting characters,

• Spline, where the reach of the line spanned by the label is fit with a spline, which the label follows with tilting characters, or

• Straight, where an inclined straight line is fit to the vertices in the line covered by the span of the label.

The vertical placement of the line’s label can now be selected as Top, Bottom, or Center relative to the line.  When the label is centered on the line, it can be bisected by the line or the Clip Under option used to open a gap for the label in the line and any other lines in that object.  The size of the gap in the line before and after the label can be set by the Clip Distance entered in the new Advanced Options dialog used for setting text styles.

The Set Line Label tool assistant now allows the selection of a table and field for the ‘Z’ value, rather than assuming the internal ‘Z’ coordinate value.  This and the new gap clipping for the label are particularly useful in preparing a contour map.

Screening Labels.

Collision or overlap errors can occur in automatic line and polygon label positions.  Certainly it would be possible for the automatic placement process to identify label collisions in one object, but rectifying them is a difficult issue and multiple layers can be involved.  Labels can also be positioned outside the extent of the vector, which you may or may not desire.  These are simply some samples of the label positions you will want to manually adjust. 

If you are preparing a professional map or view layer, you will want to conduct a quality control inspection of the placement of every label regardless of whether they are placed automatically, by you, or someone else.  You could do this by manually panning about and looking for problems.  However, a more productive approach is to zoom in to a scale that focuses upon one label and its relationship to its feature.  Next select this label and edit its appearance and position as desired.  When this is complete, use the Select Next icon on the Active Element Information panel. This will pan you at that same zoom to center on the next label on the list for its inspection.  In this fashion you can visit and give a very controlled inspection to every label in the layer and adjust those that need it.  The attached color plate entitled Screening for Label Collisions outlines and illustrates this approach.

Sliding Line Labels.

The position of a line label attached to a line can now be easily changed via a new feature of the Edit Element tool.  Simply select the label, keep the mouse button down, slide it in either direction along the line, and drop it at a new position.  The curved or straight alignment and the top, center, or bottom baseline property will be maintained.  This easy approach to editing the position of a line label is illustrated in the attached color plate entitled Interactive Editing of Line Labels. 

Drag and Drop Polygon Labels.

Individual polygon labels can now be interactively selected and repositioned.  Once a label has been selected, it can be dragged with the mouse button down to any new position in the polygon or out of it.  If the label is dragged out of the associated polygon, a straight leader line will automatically appear leading back to its original position.  For the most suitable new, external label position, this leader line may pass through some other label or feature.  If you wish, you can select a position on this leader line and pull it out, rubber band fashion, so that it will have a dogleg around that location.  You also have the option of grabbing the end of the leader line and moving it to some other position within the polygon.  The attached color plate entitled Interactive Editing of Labels and Leader Lines illustrates and discusses these procedures.

Changing a Label’s Appearance.

A new, interactive polygon text label size and orientation tool is available.  When a label is selected, it now presents some “handles.”  Selecting with the left mouse button down within the label text will drag it to a new position.  If this position is anywhere outside its polygon, a leader line appears as described above.  Dragging the new (+) plus handle with the left button down resizes the label larger or smaller.  Dragging the new ( [] ) box handle at either end of the baseline of the label will rotate it about the opposite end.  These altered individual labels can be saved with these new properties.  These elastic text operations are illustrated at the bottom of the attached color plate entitled Interactive Editing of Labels and Leader Lines.

A specific point, line, or polygon label can now be selected and custom restyled.  You can change its font and colors.  Its font characteristics can be changed to bold, italics, outline, or underline, which can also be controlled by the new italics angle, boldness, and other settings discussed in the Map and View Legends section of this MEMO.  Multiple line labels can also be created with the various alignment options and word wrap.

Map Layouts.

Each release of TNTmips for the past 2 to 3 years has introduced new features to assist you in the preparation and publishing of professional and cartographically accurate maps in paper and electronic form.  Effort in this direction continues with this release by providing

• improved automatic label placement and interactive label position editing tools,

• better font management and font appearance in PDF files,

• completely new SVG electronic map distribution format,

• advanced text styling and text justification for legends, and

• legend samples for individual elements rendered by script.

Using all these features, including some only available in V6.70, the TNT map layout process can now prepare topographic and geologic maps that closely match those of the United States Geological Survey.  The general procedures used to layout these sample maps have been documented in the 2 new Applications booklets entitled Making Geologic Maps and Making Topographic Maps.  Several other geospatial analysis systems could layout similar maps from the geodata used for producing them in TNTmips.  They will differ from TNTmips in the procedural approaches they use, how they access and analyze source materials, layout and templating procedures, the electronic publishing formats supported, and the number and integration of the products needed.

Preparing complex maps in any product is labor intensive and expensive.  Thus, it will be the efficiency and interactive ease of use in preparing the map that can best distinguish TNTmips from competing products. Some additional cartographic tools are still needed in the preparation of TNT map layouts.  However, MicroImages’ development efforts are gradually shifting toward improving the process to reduce the complexity of completing a layout (for example, the improved label placement/editing in V6.70).  Some examples of these possible future procedural improvements would be a graphical or WYSIWYG text editor and a schematic display of group relationships providing measurements and interactive group placement and positioning.

Reliability Testing.

Improvements and adjustments are made daily to some of the myriad components used in the TNT products, many of which are integrated into the TNT map layout process.  Just one example is the continual modifications to the TNT Geospatial Rendering Engine (GRE).  This has caused you frustrations when layouts and templates made in earlier versions of the TNT products can not be reused.

Another less obvious area of map layout activity for V6.70 was the implementation of an automated map layout testing procedure soon after the release of V6.60.  It works something like this.  A growing group (currently 33) of complex and representative map layouts and the Project Files used in them were assembled.  The sample geologic map and topographic map in the new Application booklets are part of this collection.  Each layout was used to produce a raster object containing that map, which was carefully scrutinized for correctness and then saved.  Starting several months ago and continuing, every one of these test map layouts is automatically recomputed several times a day to produce the same raster object.  This recompilation uses the most recent compile, or build, of TNTmips.  This new raster object is then automatically compared pixel by pixel with the original test raster.  If any pixel is different, these differences are traced down and rectified.   

This testing procedure has been effective through this V6.70 development cycle in pinpointing changes and errors that affect complex map layouts.  Since complex map layouts use many TNT component operations, this procedure also continually tests these operations as well.  However, it is not possible to design layouts that exercise all the features and approaches that you can use in a layout. MicroImages will add additional sample layouts to test other TNT components and will daily test any of the complex layouts you may wish to contribute to this activity.

Map and View Legends.

Most of these features were added at your request as you make ever more complex maps in a wider variety of your languages.  These refinements are gradually making this a truly interactive and international map layout process.

CartoScript Styled Legends.

You can now interactively create legend samples and descriptions for vector elements you have styled using a script.  The procedure described below applies whether you are using a simple style script (for points, lines, or polygons) or a CartoScript to create complex symbols for points or lines.

Background.

Line and point styles rendered by CartoScript depend upon the varying attributes associated with the elements.  Their application is specifically designed to permit you to vary the style of element throughout a view or map to convey something about the element that also varies.  For example, the varying flow rate of a river (or road or other network) could be attached to the line segments making up a river system.  You could create a separate line width style and color for every different flow rate.  However, with a single CartoScript you can draw all river segments in a width and color controlled by this flow rate attribute.  Another application of CartoScripts is illustrated in the color plate entitled Orient Point Symbols Using CartoScripts.  In this example, wind velocity vectors with varying orientations are created by a single CartoScript.

Selecting Legend Samples. 

CartoScripts are not new and their use is described in detail in a Tutorial booklet entiled Using CartoScripts.  However, in V6.60 you could not fully exploit the power of custom styling with CartoScripts as the varying conditions they represent could not be defined in our LegendView or map legends.  V6.70 adds this important feature.  However, creating a sample legend entry for a point or line type rendered by a CartoScript is not automatic.  You must interactively choose the line element or the sample point from the view that you wish to represent that line or point style in the legend. 

To add a representative element to LegendView, use the selection tools in the GeoToolbox to:

• select the representative element,

• expose the menu provided by the right mouse button,

• choose Add Active Polygon to Legend (or line, point, …),

• type in the text for this new legend sample in the Legend Element Label dialog box, which is automatically exposed, and

• the new legend element appears in the LegendView.

This procedure is illustrated in the attached color plate entitled Legend Samples for CartoScript Styles.

You would repeat this procedure to provide a sample for each different drawing style created by the style script.  Each legend sample is drawn in LegendView by your script using the database attributes of the element you selected for it.  If you change your mind about a representative element you have added, simply reselect the element in the vector object and use the right mouse button menu to delete the sample from the LegendView.  The sample descriptions and labels are stored in a database table associated with the elements, so you can select these elements easily or change the text if necessary.  In the river flow example you could create several different legend samples for this one CartoScript, and label each with the flow rate it represents.  This would provide a legend for the actual flow rates of all the rivers that are being rendered by this CartoScript in varying widths.

Dynamic Applications.

A powerful feature of styling with a CartoScript is that styles update automatically if the attributes of the line segments involved are periodically changed.  In the river example, a new view or map can be prepared each time a new flow rate table is obtained due to a rainfall event.  The wind vector map or view can be revised automatically using a map or display layout template substituting only each new hourly, daily, or monthly velocity/magnitude table.  The CartoScript approach automatically uses these new attributes to adjust its rendering. 

Your CartoScript legend sample is rendered by that actual CartoScript from the attributes for the element and position you selected.  If either of these are changed, that legend element may or may not be appropriate depending on your intentions.  For example, you may wish the width of the river legend sample to widen if the flow rate attribute for the sample position increases.  On the other hand, you may have used several river width legend samples as noted earlier each with text denoting a specific flow rate.  In this case you do not want the sample width to change because the associated interpretive text will be incorrect.

Example Maps. 

Rendering lines and points by CartoScripts is important in the design a geologic map.  Now you can now prepare appropriate legend entries for these lines.  This is illustrated clearly in the attached color plate entitled Geologic Map of the Granite Gulch Study Area, Inyo County, California and also in the sample geologic map prepared in the new Applications booklet entitled Making Geologic Maps.

Inserting Text into Polygon Samples.

Soil, geologic, vegetation, and similar maps present many different types of polygons.  Often there are so many that they can not be accurately identified in your legend by their fill color, hatching, or other symbology.  For these complex situations, the text label in the polygon is also used in the legend to insure that each type of polygon can be uniquely identified in the legend. 

V6.60 did not provide any means for inserting these text labels into the sample of each type of polygon displayed in the legend.  You can now choose to have the text label for each polygon type inserted into its legend sample in LegendView or a map layout.  The color plate entitled Text Labels for Polygon Legends illustrates this new procedure. V6.70 also provides improved techniques for locating these labels within every polygon.  This is discussed in detail in the Spatial Data Editor section of this MEMO.

Complex Legend Text.

The text with legends used for the complex polygon maps noted above and others can contain complex descriptions, text styles, and formatting.  Meeting these requirements creates a fully internationalized page layout capability embedded within the map layout process.  V6.70 provides you with more control and features to apply to text in these complex map legends.

New Text Styling.

New text styling features are available so that your legends can now be more attractive and cartographically precise in your language.  The Text Layer Control dialog provides access to a new Advanced Options dialog.  The text block width can now be entered in any of the TNT supported measurement units.  This dialog also allows you to specify boldness for bold text, the angle for italic text, thickness for TNT’s special enhanced text, shadow offset distance and angle for shadowed text.  The stroke width and offset can now be set for text underlines.  The clip distance for labels inserted into lines and polygons can also be set.  This is the distance before and after the label characters where the line or fill is clipped when the label is inserted.  You can combine these special effects, but some are mutually exclusive such as the basic styles of bold, enhanced, outline, and shadow.  The attached color plate entitled Advanced Text Features illustrates these special text styling options.

Text Alignment.

Alignment in text blocks can now be selected as Left, Right, Center, or Justify (which means, full justification) in the Text Style dialog.  You must toggle on Word Wrap on the new Advanced Options dialog to select Justify for your text block.  Word wrapping has to pick the best break positions using the spaces between whole words.  To use word wrap, make sure you only use the line return or paragraph enter key at the end of a paragraph.  You can insert hyphens or tabs as break points to refine the appearance of your word wrap.  Word wrapping is automatically enabled for justified text in multi-object legends where you have inserted a vertical right column guide. The attached color plate entitled Alignment Control for Legend Text illustrates these new justification results.

At this time when you enter and edit your text in the Properties dialog it will always appear left aligned regardless of the alignment selected.  It will only render the other alignment options when applied in the Legend Layer Controls window.  This is due to the different customs required for entering the many languages the TNT products support (for example, those that are typed right to left).

Straight Versus ZigZag Lines.

Automatic imposition of zigzag line samples in line legends may be popular with some, but often it is not appropriate for complex line styles.  Admittedly, it is a good marketing ploy as it certainly identifies the product as being created with ESRI’s ArcView.  However, cartographic applications, especially map layouts, are sometimes improved if straight lines are used for legend line samples so as to better represent the complex styles they contain, especially those styled by CartoScript.  You now have the option of choosing either zigzag or straight for the line samples that appear in your LegendView or your layouts (the default is zigzag).  This is also illustrated in the attached color plate entitled Legend Samples for CartoScript Styles. 

View PDF Version (941 Kb)

Go to Next Section

 

©MicroImages, Inc. 2010 Published in the United States of America
11th Floor - Sharp Tower, 206 South 13th Street, Lincoln NE 68508-2010   USA
Business & Sales: (402)477-9554  Support: (402)477-9562  Fax: (402)477-9559
Business info@microimages.com  Support support@microimages.com  Web webmaster@microimages.com

---