INTEGRATING ITUM LU/LC AND FWW WETLANDS COVERAGES
The following procedures were used to create the integrated land use/land cover layers (LULC) used as the baseline data set for the 1995/97 LU/LC update mapping project. The basic procedure was to start with the (ITUM) land use/land cover (LULC) layer generated has part of the Integrated Terrain Unit Map (ITUM) for each county, and update this layer with the freshwater wetlands (FWW) layer delineated for each county under the New Jersey Freshwater Wetlands Mapping Program.
Note that, while the original FWW coverages are network coverages with both polygonal wetlands and linear wetlands delineated in them, only the polygons were included in the integrated LULC layers. Keeping the linear wetland features in the coverage would have greatly increased the number of polygons created and complicated the coding scheme. In addition, the (ITUM) land use/land cover layer did not contain any similar linear features.
The individual procedures used to create the final layers are as outlined below:
1.) Preparing the FWW coverage:
All of the valid freshwater wetland polygons were reselected out of the county FWW coverage and placed in a separate layer. These polygons already are coded with a numeric CLASS code that was part of the original FWW coverage, the COWARDIN alpha-numeric label that identifies the wetlands type, a comparable LAND-USE code that assigns each wetland type a more general Anderson classification system code, and a LABEL which gives a short description of the general wetland type.
Before this layer could be used to update the land use layer, two major steps had to be undertaken. First, any linear wetland features existing within the polygons, and other linear features used in the FWW coverages (quad boundaries, break lines), had to be eliminated. Each reselected FWW coverage was BUILT as a line coverage. IN ARCEDIT, with EDITFEATURE ARC, arcs with LPOLY# = RPOLY# were selected and deleted. The coverage was then BUILT with the POLY option.
The second problem involved islands of non-wetland areas contained within larger wetland polygons. All such non-wetland areas were simply identified as UPLANDS in the FWW mapping program, no matter what the land cover or land use. Before the reselected FWW polygons could be merged into the ITUM land use/land cover layer, these islands had to be coded with the correct land use/land cover type. This was done by reselecting these islands out of the reselected FWW layer according to the expression: RES CLASS = 0.
This island coverage was then INTERSECTED with the ITUM land use/land cover layer so that all appropriate land use/land cover codes and interior lines would then be moved into these polygons. Note that before INTERSECT is given all items except the default items are dropped from the .PAT of the island coverage. This will enable the item LAND-USE containing the Anderson codes to be properly moved into the island polygons.
Once this was accomplished, the completely coded polygons were used to UPDATE the FWW polygon coverage. Note that in order to use the UPDATE command the .PATís for the two coverages have to be the same. Appropriate attributes were therefore added to the .PATís for both coverages.
2.)Identifying areas in the land use/land cover layer below the Upper Wetlands Boundary (UWB):
One of the major benefits of integrating the FWW layer into the land use/land cover layer was the improvement of delineations of lakes and ponds from the FWW layer. Lakes and ponds in the land use/land cover layer were taken from the USGS DLG files for New Jersey. The source maps for these DLG files are generally out of date and not as accurate as the 1986 orthophoto basemaps used to do the FWW mapping.
A complicating problem in this replacement exists because the FWW mapping project was limited to non-tidal areas of the state. These were defined as all areas above the Upper Wetlands Boundary (UWB). The UWB was a line designated in a previous state mapping program to identify tidally influenced areas of the state from those non-tidally influenced areas. Tidally influenced areas, i.e., areas below the UWB, are under Tidal Wetlands Regulations, while non-tidally influenced areas, i.e., areas above the UWB, are regulated by the Land Use Regulatory Program. When the Freshwater Wetlands Act of New Jersey went into effect, only those areas not already under Tidal Wetlands Regulations were to covered by the regulation and, hence, mapped under the FWW mapping program. The UWB, therefore, provided the seaward limit of the FWW mapping effort. Ponds and lakes and other water features below the UWB were therefore not updated by the FWW mapping and remain USGS DLG delineations.
In addition to the fact that the FWW mapping did not cover areas below the UWB, several of the water features contained in the land use/land cover layers were not correctly coded as being tidal. The original USGS water layer used in the ITUM did not separate out tidal from non-tidal waters. In addition, when the land use/land cover layers were created, the UWB maps had not been digitized so that this information was not available for the land use/land cover mapping until the FWW mapping was undertaken and completed.
So that further processing steps did not completely eliminate any water polygons existing below the UWB, all tidally influenced waters had to be identified and given a unique tidal water classification so that these could be separated from other water polygons in the land use/land cover layer which would be eliminated. Anderson does have a water series of 5400, which is used specifically to identify tidal waters.
The following steps were undertaken to insure that this code was in all water polygons below the UWB:
The UWB lines, which were included in the FWW maps, were selected out and placed in a separate coverage for each county. Each land use/land cover layer was then examined with this UWB coverage. Water polygons below the UWB were checked to make sure that they were correctly coded with a tidal water classification. If they were not correctly coded, the LAND-USE code was changed to the 5400 series. Generally, entire water features were either below or above the UWB. However, in those cases where this was not the case, the polygon in the land use/land cover layer was still given a 5400 code. Processing done in further steps correctly split the polygon at the tidal/non-tidal interface coding the upper portion with the right non-tidal code but leaving the lower portion with the correct tidal water code.
3.)Updating the land use/land cover layer with the FWW layer:
Once a completed FWW layer was created containing only valid wetlands polygons and upland islands with the correct Anderson codes, and a land use/land cover layer was created with tidal and non-tidal waters correctly identified, the two coverages could be merged together. The ARC command UPDATE was used to update the base coverage, land use/land cover layer, with the second layer, the FWW coverage. Linework and codes in the base layer are replaced by the line work and codes contained in the updated layer. Any area in the base layer not covered by a polygon in the update layer remains in the output layer as originally delineated. Areas in the base layer covered by polygons in the update layer are completely replaced by the new polygons.
The only additional preparation done to the two original coverages was to standardize the .PAT files. UPDATE requires that each coverage have the same attributes. Several items had to be added to both the FWW layers and the land use/land cover layers for this. These extra items were not evaluated in either coverage.
4.) Eliminating slivers and non-water polygons:
A major problem in updating the ITUM land use/land cover layer with the FWW layer was that large number (thousands per county tile)of slivers were created in each new composite layer. These arose mainly because of the different mapping specifications used in each project and the different scales of the orthophoto basemaps, 1:12000 for the FWW maps and 1:24000 for the land use/land cover maps. Since most of these are remnants of the processing and not real features, they had to be eliminated. These were eliminated using the ELIMINATE command in ARC, with special procedures undertaken to control which arcs were dropped in the processing. The dominant rule was that no arc defining a polygon from the FWW maps was to be dropped. The FWW arcs are, by definition, more accurate and more detailed than arcs generated in the land use/land cover mapping. So, wherever a sliver polygon existed between a larger land use/land cover poly and a FWW poly, the arc for the land use/land cover poly was to be dropped and the sliver incorporated into the land use/land cover non-wetland polygon.
The dropping of arcs was controlled by coding the FWW arcs with a negative user id. ELIMINATE will not remove an arc with a negative -id. The coding of the FWW arcs was accomplished by running an AML on the UPDATED coverage that first built the coverage as a line coverage, identified whether the left or right polygon of each arc had a valid FWW wetlands code, and if so calculated the -id of these arcs to be negative. In addition, a DELETE flag was evaluated for all polygons. Those below the predetermined sliver size of .25 acres were given a DELETE value of 1, and all other polygons given a value of 0. When ELIMINATE is run, all polygons with DELETE = 1 were reselected for elimination. The arc that was dropped to eliminate the sliver was the arc having a positive -id. This AML was a modification of one written by ESRI in doing sliver analysis and removal when generating the original ITUM for each county.
Additional polygons were also given a DELETE value of 1 before the ELIMINATE command was run. These included all polygons regardless of size that were identified in the land use/land cover as non-tidal water but which were are not water at all when the land use/land cover layer is examined over the photo basemaps. These could be identified by reselecting out those polys that had an Anderson code between 5000 and 5400 (freshwater ponds, lakes, streams, rivers and canals) but a FWW CLASS code that would identify them as uplands. All of these polys were given a DELETE value of 1 and eliminated along with the slivers.
5.) Re-coding incorrectly coded non-tidal wetlands in the Land use/land cover layer:
Again, because of differences in mapping specifications and standards between the ITUM land use/land cover mapping and the FWW mapping, differences existed in the areas identified as non-tidal wetlands in the two data sets. Generally, more wetlands exist in a county FWW layer than are identified in the land use/land cover mapping, and these areas are integrated in to the land use/land cover layer by the update step above. However, there are often areas identified as non-tidal wetlands in the land use/land cover mapping that are outside the boundaries of the polygons delineated in the FWW mapping. It was determined that the delineations in the land use/land cover mapping were in error, and these areas should be identified as uplands and not wetlands.
To identify these areas, polygons were selected that had CLASS = 0 and LAND-USE values of either 6210; 6220 or 6221; or 6230, 6240. Any poly with CLASS = 0 is a polygon not delineated in the FWW layer, since all valid FWW polygons would have a CLASS value > 0. The LAND-USE values identify wetlands in the Anderson classification system. A value of 6210 identifies deciduous wooded wetlands. Values of 6220 identify coniferous wooded wetlands other than those dominated by Atlantic White Cedar, which have a LAND-USE value of 6221. LAND_USE values of 6230 and 6240 identify scrub/shrub and herbaceous wetlands of various types. In INFO, polygons of CLASS = 0 and LAND-USE = 6210 were given a corrected LAND-USE land use/land cover of 4100, the upland deciduous forest code. Polygons of CLASS = 0 and LAND-USE = 6220 or 6221 were given a corrected code of 4200, the upland coniferous forest code. Polygons with CLASS = 0 and LAND-USE = 6230 or 6240 were given a corrected LAND-USE code of 4400, the code for upland brush and scrub/shrub areas.
Once the re-coding were completed, a second AML was run that identified any of the polygons whose LAND-USE values were just corrected that were adjacent to an original land use/land cover poly that had the same LAND-USE value. If a newly coded polygon met this criteria, it could be merged in with the larger polygon with the same classification. The AML examined the LAND-USE values of the left and right polygons of all arcs. If one of the corrected polygons shared an arc with an adjacent polygon of the same LAND-USE value, the arc was given a unique -id. After all such arcs were found and coded with a unique id, they were set to positive values while all other arcs in the coverage were set to negative values.. When the ELIMINATE command was run, polygons whose LAND-USE codes were changed were selected, and the arcs with the positive -idís were dropped. Polygons with dropped arcs were merged with a larger adjacent polygon having the same LAND-USE code. Polygons whose LAND-USE values were changed, but that did not have any bounding arcs with positive -idís were not eliminated.
6.) Correcting miscoded wetlands below the UWB:
The procedure outlined above for correcting miscoded polygons in the land use layer, may have resulted in codes in tidal wetlands polygons below the UWB being incorrectly changed. This occurred because some of these polygons were coded as non-tidal wetlands when they should have been coded as tidal wetlands.
The magnitude of the problem varied from county to county, and the procedures used to correct the errors depended on the relative number of polygons involved. To evaluate the problem in each county, polygons whose original land-use value was between 6200 and 7000 (all non-tidal wetland codes) and whose codes had been changed to a comparable upland category were highlighted and examined with the UWB line displayed. These polygons are still wetlands but should be identified as tidal wetlands and coded as either 6110 or 6120.
If there were small numbers of polygons involved, the code corrections were done interactively on a polygon by polygon basis. Ids of incorrect polygons were noted and codes and labels of these polygons changed to the correct tidal wetlands codes and labels.
When there were large numbers of polygons involved, additional processing was required to re-code the polygons. First, all polygons that were originally code as wetlands in the land use layer that were re-coded to uplands in Step 5 above were reselected and placed in a separate coverage.
Polygons representing the tidal areas of each county were reselected out of the county FWW coverage and also placed in a separate coverage. These polygons were formed by the UWB and sections of the outer county boundaries. An item was added to this coverage, called UWB. It was evaluated as 1 for all tidal polygons and as 0 for any Ďislandí polygons contained in these tidal areas that was not tidal.
The reselected wetland polygon coverage and the tidal polygon coverage were then combined using the UNION command. The result was to move the UWB code from the tidal polygon coverage into each wetland polygon. All polygons with a UWB value of 1 were then reselected out and placed in a separate coverage.
This new coverage again contained wetland polygons that were tidal as well as no wetland polygon islands within them. The wetland polygons were identified by re-selecting those with a UWB value of 1 and a non-zero LAND-USE value. The LAND-USE, LABEL and LEVELI items in the .PAT were checked for these polygons and changed to the correct values identifying tidal wetlands.
As described earlier for creating the county FWW coverage, the non-tidal islands within this coverage, if there were any, had to be coded separately. They were identified by having a UWB value of 0, re-selected and placed in a separate coverage. This coverage was then INTERSECTed with the coverage generated from Step 5 above, so that all the correct codes and internal lines within these non-tidal areas were restored. This fully coded island coverage could then be used to UPDATE the tidal wetlands polygon coverage.
This completely coded tidal wetlands polygon coverage could then be merged with the original ITUM land use/land cover coverage from which the polygons were selected. Again, the UPDATE command was used so that all correct codes and polygon boundaries would be replaced into the integrated coverage.
As with the original UPDATES done with the FWW layer and the land use layer, this updating also caused some slivers to be created that were identified and ELIMINATEd as described in Step 4 above. Again, an AML was run to identify arcs that should not be eliminated for polygons below .25 acres.
7.) Updating the .PATís
The final step in preparing the coverages was to update the attribute files. The processing required that many additional attributes be added for certain steps, most of which could be dropped after the final processing was completed. Those attributes remaining after the -id are:
LAND-USE Anderson classification system value.
CLASS the numeric code used in the original FWW
COWARDIN The specific wetland type the CLASS value
represents, from COWARDIN et al.
LABEL A description of the Anderson land use category.
LEVELI A generalized grouping of the Anderson
LEVEL II/Level III value to Level I.
8. ) Additional comments:
The purpose of producing the integrated LULC data sets was to improve the utility of the two original data layers by merging them together, eliminating where possible the inconsistencies in the two layers, and standardizing the coding between the two data sets. All non-tidal wetland polygons from the FWW mapping have been incorporated into the land use/land cover layer and given a comparable Anderson classification code. Non-tidal wetlands incorrectly coded as such in the ITUM land use/land cover layer have been re-coded as other upland cover types. Finally, non-tidal waterbodies in the original land use/land cover layer that came from outdated USGS DLG files have been removed and replaced. Importantly, all the detail inherent in the original FWW mapping has been retained in these integrated data layers. All of these polygons can be identified by their positive CLASS values and filled COWARDIN field. If the user so wishes, these polygons can be examined in their original detail by using the CLASS or COWARDIN attributes to do their analyses. These polygons can also be viewed in the context of the more general Anderson classification scheme used to describe the rest of the land use/land cover polygons by using the LAND-USE values.
9.) Problems and cautions:
While every attempt has been made to eliminate as many errors as possible, there are several characteristics of both the original data sets themselves, and the processing done to date, that create problems in the combined data layers that warrant special comment.
First, while smaller slivers have been eliminated, many polygons still exist in the combined data layers that are below the minimum polygon size of either original data layer. These polygons are generally greater than .25 acres but less than the 2.5 acre minimum polygon size of the ITUM land use/land cover layer, and as with the small slivers that were eliminated, were created where FWW polygons replaced most but not all of the original land use/land cover polygon.
Examination of many of these polygons indicates that the determination of whether they should be eliminated, left with their original code, left but given a new code, or merged into an adjacent polygon that has a different code, needs to be done on a case by case basis. It was therefore decided to leave these polygons in the merged data layers for the time being. Plans are underway to update all of these land-use/land cover layers with the new digital imagery being produced for the NJDEP, and these problem polygons will be examined in detail at that time. Since they represent a very small portion of the total area of any county, even if the coding of these polygons is in error, their inclusion in any land use/land cover calculations will at most account for only a small error in the figures. While some of this error would be corrected by performing some county wide processing steps similar to those done for removing the very small slivers, as many new errors may be introduced.
The re-coding of the polygons identified in the original land use/land cover layer as wetlands which are, in fact, not wetlands, with a comparable uplands cover type, may have resulted in some small polygons being incorrectly coded. As an example, consider a small area of a roadway that was included in a large wetlands polygon in the original land use layer. While the road is clearly not wetlands, it was not delineated as a separate polygon because it did not meet the minimum size requirements for a mappable feature. When the FWW polygons are incorporated into the land use layer, this road polygon becomes a distinct element because it does meet the minimum size requirements for a mappable feature in the FWW data layer. The general re-coding scheme presented above would result in this polygon being re-coded; however, it would not be re-coded to a roadway, but to some upland forest or brush category. Whether or not it is primarily road surface or a mix of roadway and adjacent upland forest will need to be determined in future updates, and corrected if necessary.
While not incorrectly coded, users will also note that often long narrow extensions of many land use/land cover polygons exist in the updated data layers. These occur where the boundaries of the original land use polygons and the boundaries of FWW polygons replacing them or sections of them never intersect. These areas are, therefore, not small individual sliver polygons that can be eliminated by the processing described in the above sections. These extensions would need to be corrected by performing polygon specific editing that would involve splitting, adding and eliminating various arcs. Again, these types of corrections will be undertaken in future updates.
The most problematic sections of these new data sets are to be found in the areas of each county below the UWB. Since no FWW polygons exist in these areas, no corrections to water feature or wetland delineations were made to the land use layers when they were merged with the FWW layers. When the new data layers are viewed over highly accurate base maps or digital imagery, these errors are particularly evident. Again, correcting these errors, particularly where re-delineating the water features is concerned, is a very labor intensive process beyond the scope of the present project. These errors will be corrected when the updates of these data layers to the digital imagery are done in the future.
County specific comments:
Two counties included here had some additional processing not done for any other county. These are Gloucester and Somerset counties. For Gloucester county, water features below the UWB have been re-delineated and corrected to the 1991 digital imagery resident on the NJDEPís GIS system. This county is being used for a test to investigate the process of using the digital imagery to update land use/land cover layers in a heads-up fashion. In the process, correcting water feature delineations in areas below the UWB was completed. This is a step that will be done for all other counties with tidal areas, when the digital image updates are undertaken.
For Somerset counties, there are some additional coding characteristics that may be confusing. When the Somerset land use/land cover data layer was created as part of the Somerset ITUM project, Anderson Level III codes were used to categorize the polygons. Where in all other counties, for example, forests were coded either 4100 or 4200 for deciduous or coniferous forests, respectively, in Somerset, these were broken done to the next level such as 4110 or 4120, deciduous forest with less than or greater than 50% crown closure, respectively, etc. Wetland types were also delineated to Level III, but in keeping with coding used in the FWW program, the divisions were based on such factors as types of predominate vegetation, whether or not deciduous and coniferous species were mixed, what the water regime associated with the site was, etc, but not percent crown closure. If an area identified as wetlands in the land use layer was in fact not wetlands when the FWW polygons were incorporated into the coverage, the re-coding scheme that was employed in this project did not allow this incorrectly coded wetland polygons to be given a Level III upland code comparable to other polygons in the county. These polygons could only be coded to Level II. Coding to Level III could be accomplished only by examining the photography for each of these sites, which again was beyond the scope of this project. As with other problem polygons identified above, these inconsistencies will be corrected in future land use/land cover updates.
While problems and errors do exist in the integrated data layers created, we do feel that these layers are extremely useful, and their benefits far outweigh any problems they might contain. First, it has always been confusing to many users as to how to interpret the two delineations of non-tidal wetlands that existed in the two original data sets. This problem is eliminated with the present data sets, since the FWW delineations now are included in the land use layer. Any site analysis using either the FWW layer or these improved land use/land cover layers will involve the same non-tidal wetlands delineations. In addition, it is now necessary only to examine this one integrated layer to make use of the information from the two original extensive data sets. This should result in more efficient site analysis and easier data manipulation. Lastly, many positional errors included as part of the delineations of the water features in the original land use layers are corrected by replacing these features with the water features from the FWW coverages.