Working Projection
Download 423.29 Kb.
|
- Navigate this page:
- Watershed Polygons
- North American Watershed names
- Watershed Lines
| North America Atlas
Basin Watersheds, Rivers, Waterbodies, Political Boundaries, roads, rails. Source Datasets and Process Notes Working Projection: Projection Lambert Conformal Conic Datum NAD83 Zunits NO Units METERS Spheroid GRS1980 Xshift 0.0000000000 Yshift 0.0000000000 Parameters 49 0 0.000 /* 1st standard parallel 77 0 0.000 /* 2nd standard parallel -95 0 0.000 /* central meridian 49 0 0.000 /* latitude of projection's origin 0.00000 /* false easting (meters) 0.00000 /* false northing (meters)
Projection: Lambert_Azimuthal_Equal_Area False_Easting: 0.000000 False_Northing: 0.000000 Central_Meridian: -100.000000 Latitude_Of_Origin: 45.000000 Linear Unit: Meter (1.000000) Geographic Coordinate System: GCS_Sphere_ARC_INFO Angular Unit: Degree (0.017453292519943299) Prime Meridian: Greenwich (0.000000000000000000) Datum: D_Sphere_ARC_INFO Spheroid: Sphere_ARC_INFO Semimajor Axis: 6370997.000000000000000000 Semiminor Axis: 6370997.000000000000000000 Inverse Flattening: 0.000000000000000000
10M Watersheds National_Atlas_updated_October2008.zip National_Atlas_work.gdb (geodatabase of data created or edited in Oct 2008) updated_datasets_lambert (featuredataset) featureclasses… Hydro_rivers_editv1 (includes coastline, lakeshores) NorthAmerica_l_editv1 (watershed boundary lines) NorthAmerica_p_editv1 (watershed polygons) bound_l_editv1 (political boundary lines) bound_p_editv1 (political polygons) ContourNA250m (unused in this process)
road_l_shp (2004) rail_l_shp (2004) Downloaded from Atlas of Canada canadfda (1:1-million watersheds) bndlam (1:1-million political boundaries) PROCESS NOTES Scope of Project The original plan was to update the Canadian watershed boundaries and polygons from an accurate, large-scale dataset. In order to maintain the integration between all NA Atlas layers, several other datasets also had to be adjusted: The hydro lines and polygons were edited a) to fit properly with the new Canadian watershed boundaries, or b) to correct significant errors in the hydro base such as the omission of major reservoirs. A small number of roads and rails needed to be shifted due to edits to the hydro layers. Political boundaries required editing because some segments need to be coincident with watershed boundaries (heights-of-land), but the existing political boundaries were of lower absolute locational accuracy than the new watersheds. Rather than degrade the accuracy of the watersheds, it was decided to upgrade the political boundaries. This will make it easier in the future to integrate larger-scale data from Atlas of Canada or any other sources. Note that throughout this process, only features that fall entirely within Canada were edited for geometry. Features coincident with the Canada/USA border were not edited. Coastline was not edited. The only exceptions were several features with minor corrections to attributes only, the correction of coastline in Alaska to match John Hutchinson’s earlier corrections, and the addition of two islands in the Bering Sea (HUC 19030104). Editing Environment The majority of the work was done in coverages in the ArcInfo workstation environment. In order to make use all resources (editing tools, reference layers, etc) the datasets were converted to the Atlas of Canada’s working projection (Lambert Conformal Conic), and projected back to the North American Altas’ projection (Lambert Azimuthal Equal Area) for final output. There should be no concern about coordinate shifts caused by the two projection processes because the mathematical precision of the projection process is much greater than the coordinate precision of the geodatabase, which is .001 meters. Further, no geometry outside of Canada was altered. WATERSHEDS Watershed PolygonsThe 1:1-million watershed coverage CANADFDA (FDA = Fundamental Drainage Areas) was copied from internal Atlas repository. The Water Survey of Canada’s sub-drainage area dataset was extracted and generalized for 1:10M. After trial-and-error, the generalization process was unsplit all arcs to reduce the number of nodes. (The UNSPLIT command merges lines. Here it is done after each step because nodes (endpoints) do not get generalized, they remain fixed, so they influence the final shape of the line far more than individual vertices. In ArcInfo, lines have a limit of 500 vertices, so each reduction in the number of vertices is an opportunity to combine lines and eliminate fixed endpoints. generalize (pointremove) with small tolerance (50 m) to reduce excessive vertex density unsplit again to further reduce nodes spline with tolerance of 3000 m unsplit again spline with tolerance of 1500 to smooth the line somewhat, reducing its angularity unsplit generalize (pointremove) a final time with the 50 m tolerance to remove redundant vertices in straight-line segments. Located intersections (overlaps) between the generalized watershed boundaries and the hydro lines. Many of these intersections were legitimate (e.g.: a watershed boundary crossing a confluence of streams), but others were the result of the watershed boundary being shifted and now crossing a river or lakeshore. Stepped through these intersections (visual inspection) and corrected the hydro as required. At the same time checked roads, rails and populated places for any conflicts with the shifted hydro. Atlas of Canada frameworks layers at 1:1M were used for reference. (Note: after this step was complete, John Hutchinson, USGS, suggested the use of a different generalization algorithm. This new algorithm was used for subsequent generalization of different datasets, but because the Quality Control of watersheds was complete it was decided not to re-generalize the watersheds.) Some political boundaries are defined as height-of-land, and some of these coincide with watershed boundaries. In these cases, the watershed boundary was made coincident with the generalized political boundary (see section on political boundaries, below). The WSC sub-basins do not match NA Atlas watersheds exactly: there are inconsistencies in watershed definitions across the Canada-USA border, and areas of “internal drainage” are not identified in the WSC’s watershed scheme. To account for these shortcomings, some extra boundaries were imported from the Atlas of Canada’s Fundamental Drainage Areas dataset (CANADFDA). Polygon topology was built. Taking advantage of line/polygon topology, the line attribute TYPE_WSC was checked and corrected. A new value of “9” was added to the extra boundary lines which are not part of the WSC watershed scheme. The Canadian watershed boundary lines were reintegrated into the 2008 geodatabase: The 2008 geodatabase National_Atlas_work.gdb (see section Source Datasets above) was copied to a new geodatabase NA_Atlas_20090904.gdb. The featuredataset updated_datasets_lambert was renamed to data_20090904. The featureclass NorthAmerica_l_editv1 was renamed wshed_l. In wshed_l, the features with COUNTRY = ‘CAN’ were deleted. (Note that features with COUNTRY = ‘CAN USA’ were not deleted.) The attributes EDIT_DATE and EDIT were added to accommodate these new attributes in the new coverage. All features outside of Canada were coded as EDIT_DATE = 20081020 and EDIT = NEW. The line feature classes in the watersheds coverage was imported as a temporary feature class in the featuredataset data_20090904, thus inheriting the projection and (much lower, but at .001 metres, quite sufficient) coordinate precision of the featuredataset. In the temporary feature class, all features with COUNTRY <> ‘CAN’ were deleted. An ArcMap edit session was started. All remaining features (the Canadian features) in the temporary line featureclass were copied and pasted into the featureclass wshed_l. These features were coded with EDIT_DATE = 20090827 and EDIT = EDT and given a UIDENT value. Once the watersheds were back in the geodatabase, it was decided that for the upcoming release of this dataset there should be a way of showing the North American watersheds in a graphic manner similar to the watersheds paper map published in 2006. The attribute MAP_COLOR was added, and the watersheds were selected and coded with a number from 1-18 to represent the colour. This crude coding scheme is based on the legend of the paper map. MAP_COLOR Level 1 Level 2 Level 3 1 Arctic Ocean Seaboard 2 Arctic Ocean Mackenzie River 3 Atlantic Ocean Seaboard 4 Atlantic Ocean St. Lawrence River 5 Gulf of Mexico Seaboard 6 Gulf of Mexico Mississippi River system Mississippi River 7 Gulf of Mexico Mississippi River system Missouri River 8 Gulf of Mexico Mississippi River system Ohio River 9 Gulf of Mexico Mississippi River system Arkansas River 10 Gulf of Mexico Rio Grande 11 Hudson Bay Seaboard 12 Hudson Bay Nelson River 13 Pacific Ocean Seaboard 14 Pacific Ocean Yukon River 15 Pacific Ocean Columbia River 16 Pacific Ocean Colorado River 17 Caribbean Sea Seaboard 18 Internal drainage area
0 Review of geometry (and possible edit) required. For the sake of speed, this work was done in a copy of the watershed feature class, converted to coverage format. Once done, the attributes were joined back to the geodatabase featureclass via the MAP_LABEL field. There were several small areas where the featureclass boundaries did not match those on the map. These may or may not be changed, so they were left with MAP_COLOR = 0. A plot (PDF) highlighting these areas was sent to USGS and INEGI for review. Two other polygons need to be checked as to whether they were correctly assigned to a Major River Basin (Yukon, Mississippi) or if they should be considered “Seaboard”. At this point it became obvious that an attribute for colouring the watersheds was not enough, and that we really need a set of North American Watersheds which have not just national identities but also continental identities which cross international boundaries. North American Watershed namesCopying the legend from the paper map, a spreadsheet was created containing the English, Spanish and French names associated with each map colour (see Appendix 2). These were joined to the geodatabase featureclass via the MAP_COLOR field. This took care of Levels 1,2 and 3. Empty fields for level 4 watershed names were added. In total there are 12 fields for watershed names. For now they are all text fields, 60 characters wide. NAW1_EN NAW1_SP NAW1_FR NAW2_EN NAW2_SP NAW2_FR NAW3_EN NAW3_SP NAW3_FR NAW4_EN NAW4_SP NAW4_FR [NAW1_SP means North American Watershed, Level 1, Spanish]. The attributes for the 4th level, the smallest drainage units, were populated. For Canadian watersheds, the attribute WSCSDA_NAME was copied to NAW4_EN, NAW4_SP, and NAW4_FR. For USA watersheds the source attribute was ACC_NAME, and for the Mexican watersheds the source was REG_RH. For watersheds falling entirely within a single country these names were fine. But where a physical watershed crosses an international boundary, a common name is required for both parts. All watersheds along the international boundaries were inspected and tentative watershed names were overwritten in the NAW4_EN, NAW4_SP, and NAW4_FR attributes. The attribute SHARED indicated the type of decision taken in assigning the tentative names. A table showing all the decisions made is attached as Appendix 3. In most cases the existing name from one of the countries was adopted. In these cases, the attribute SHARED is either CAN, USA, or MEX. Occasionally neither country’s name describe the combined watershed, and a new name was invented (SHARED = NEW). In 2 cases the name was already the same in each country (SHARED = OK). There were a few cases (e.g. Yukon River) where a watershed inside a country (not crossing or touching the boundary) had to be renamed in order that other names made sense (SHARED = NEW). There are 2 pairs of cross-border polygons for which I did not suggest even a tentative name: one near Baja California, and the other involving a possible re-delineation of a watershed. These were coded XX1 and XX2 respectively. Watershed LinesAttributes for lines were assigned in ArcInfo using line/polygon topology, and joined back to the wshed_l feature class. Record FREQ NAW_LEVEL Description 1 8 -9 Frame 2 93 -8 International bdy 3 1955 -7 Coastline 4 11 -6 Not drawn on map 5 249 -1 Lakeshore 6 194 1 Level 1 (Ocean Drainage Areas) 7 117 2 Level 2 (Major River systems) 8 41 3 Level 3 (Sub-basins in Mississippi) 9 1025 4 Level 4 (smallest units) 10 151 9 Boundary of internal drainage area 11 18 100 Bdys of polygons needing review. POLITICAL BOUNDARIES Political boundaries required editing because some segments need to be coincident with watershed boundaries (heights-of-land) or hydro features, but the existing political boundaries were of lower absolute locational accuracy than the new watersheds. Rather than degrade the accuracy of the watersheds, the political boundaries were upgraded. The 1:1-million watershed coverage BNDLAM was copied from the internal Atlas repository. Segments of internal (provincial/territorial) boundaries that were coincident with streams or watershed boundaries were identified and generalized using the ArcGIS “Smooth Line” tool using the PAEK option with a tolerance of 6000 m, followed by some minor manual modification. The coincident watershed boundaries and rivers were edited to match these lines. (Smoothed sections were YK/NT, BC/AB, QC/NL, ON/QC along the Ottawa River, and QC/NB along the Restigouche River.) Because most of the remaining political boundaries were relatively simple straight-line segments, it was decided to replace the entire set of Canadian provincial/territorial boundaries. These boundary segments were generalized or densified to achieve a vertex density of about 3000-5000 metres. The locational accuracy of the internal Canadian boundaries is now much improved, and they shouldn’t need to be revisited in the future. Note that along the International boundary in the Alaska panhandle the boundary is defined as peak-to-peak. About 125 km of this boundary roughly follows edge of the Yukon River watershed. The Canadian political boundary lines and polygons were reintegrated into the 2008 data in the same manner as the watershed lines (see above). The only difference is that no detailed change-detection was required since all internal Canadian lines and polygons had been replaced, and no others. These were simply flagged with EDIT = ‘EDT’ and EDIT_DATE = ‘20090827’ HYDRO Converted the geodatabase feature class “Hydro_rivers_editv1” to a coverage in Lambert Conformal Conic projection. Hydro polygons have not been edited since 2004, but streams were edited in 2008. Shifts in hydro features were required due to changes to the watershed boundaries (see above) but some significant errors and omissions needed to be corrected as well. Because this involved edits to hydro polygons, the 2004 hydro polygon shapefile was downloaded from the CEC website and converted to coverage. The label points (and their attributes) were imported into the hydro line coverage and polygon topology was built. During subsequent editing, corrections to lakeshores automatically propagated to the polygon feature class. The Ottawa and Restigouche rivers are coincident with provincial boundaries (ON-QC and QC-NC respectively) and the geometry for these rivers was made coincident with the generalized political boundaries (see section on political boundaries, above). Used an AML change.aml to detect which arcs were modified (added, deleted, edited, split, moved, or edits to key attributes). These were flagged with new attributes EDIT_DATE and EDIT, and new features were given a new UIDENT value. Line/polygon topology was used to detect which polygons had been edited, and these were similarly attributed. The Canadian hydro lines were reintegrated into the 2008 data in the same manner as the watershed lines (see above). The hydro polygons required an additional step. In this layer, in addition to waterbodies there exist polygons to represent all areas that are not lake or ocean – all the continental land masses, including the islands within lakes. Effectively, lakes are holes in the continental landmass. During the editing process, over 200 Canadian lakes were either added or edited, so reshaping the holes to match the modified lakes was not a good option. Instead, all the Canadian lakes in wshed_p (the 2008 version of the featureclass) were merged with the North American landmass polygon, filling in the Canadian holes. Then the “Erase” tool (ArcToolbox>Analysis Tools>Overlay>Erase) was used to punch holes in the landmass using the new Canadian lakes as the “Erase Features”. Finally, the new Canadian lakes were cut-and-pasted into wshed_p. ROADS AND RAILS Neither of these layers had been edited since 2004. Shifts in the positions of lakes and streams created some conflicts with nearby roads and rails which required edits to a small number of features. The 2004 layers were downloaded from the CEC web site http://www.cec.org/naatlas/maps and converted to coverages in Lambert Conformal Conic. At the same time, copies of roads and rails were placed in the geodatabase featuredataset, thus inheriting the featuredataset’s projection and coordinate precision. These coverages were edited concurrently with the hydro coverages. Once the edits were finished, the AML script change.aml was used to detect which arcs were modified (added, deleted, edited, split, moved, or edits to key attributes). These were flagged with new attributes EDIT_DATE and EDIT, and new features were given a new UIDENT value. The Canadian roads and rails were reintegrated into the 2008 data in a similar manner to the watershed lines (see above). However, since very few features were edited and none were added or deleted, the old arcs were manually selected and deleted, then replaced with the new arcs. Roads and rails were plotted on top of the ~800 hydro arcs which had been changed. (It was assumed that the roads and rails would not conflict with unmodified hydro.) About a dozen minor conflicts were found, and roads or rails were corrected (updating EDIT_DATE and EDIT attributes). POPULATED PLACES This layer had not been edited since 2004. Shifts in the positions of political boundaries, lakes, streams, roads and rails could have created some conflicts with nearby roads and rails. The 2004 layers were downloaded from the CEC web site http://www.cec.org/naatlas/maps and were edited concurrently with the hydro coverages. When edits were complete no populated places had been shifted. Several QC plots were made and visual inspection turned up no conflicts with this round of edits. APPENDIX 1: HIERARCHY Four levels, not three. There is a slight wrinkle in this attractively simple hierarchy which communications people and others may need to be aware of when we release this layer. It’s not that there’s anything wrong, just nature not cooperating with our desire to abstract. The various levels of our watershed hierarchy reveal and make comprehensible the drainage patterns of the continent. In classifying the North American Watersheds in the GIS layer, I copied the hierarchy of the paper map. At first glance there appear to be 3 levels, nicely correlated to the size of their units: 1) ocean drainage areas covering the North American landmass, being subdivided into 2) major rivers and large remnant areas of “Seaboard” drainage, in turn divided into 3) individual local watersheds. At the top level (Level 1, the Ocean Drainage Areas) the units are fairly straightforward: one might argue about where on the Florida coast should the boundary fall between the Atlantic Ocean and the Gulf of Mexico – but I wouldn’t. (The Caribbean Sea drainage area is not well defined here, as only a fraction of it falls within the 3 countries.) At the bottom level (Level 4, the level of the smallest watersheds) the drainage areas in the 3 countries are relatively consistent in size. Although it’s not explicitly defined, it seems reasonable to describe Level 2 as Major River Basins: basins of a certain minimum size that have an outlet at a single point or delta on the ocean shore. The remaining areas of the North American land mass (within CAN, USA and MEX) are classed as “Seaboard” or “Littoral”. The largest of the Major River Basins is the Mississippi River. It’s exceptional in that it’s almost twice the size of the 2nd-largest, the Mackenzie River, and larger than the smallest of the Ocean Drainage Areas (the Atlantic Ocean). On the paper map a third level was created below Level 2, dividing the Mississippi into 4 units. Three of these units are basins tributary to the Mississippi River, and the fourth is the remainder of the Mississippi basin. Each of these units is comparable in size to the other Level 2 units (see Figure 1). If one were building this hierarchy based purely on size, there would be no distinction between the units in Level 2 and Level 3. The units in this middle level would be of comparable size, smaller than Ocean drainage Areas and larger than the local watersheds. The unit called Mississippi River System would not exist, its area being covered by the 4 sub-units. And the definition of this middle level could not include “having an outlet at a single point or delta on the ocean shore”. But it appears that this hierarchy was based on size and on “river basins with an outlet to the ocean”: Ocean Drainage Areas, River, and local watersheds. And that the Mississippi River basin was so much larger than the others that the drainage patterns of this area, a large part of the continent, were hard to see. And that the Mississippi River basin lends itself easily to being subdivided into units of a size compatible with the design of the map. So size was a guiding but not defining factor in the creation of this hierarchy. 
Figure 1. Only areas within Canada, USA and Mexico are counted. Most of the Pacific and Atlantic Ocean Drainage Areas fall inside the three countries. Only a tiny fraction of the Caribbean Sea Ocean Drainage Area is counted (a small piece of Mexico and about one quarter of Puerto Rico). APPENDIX 2: NAMES FOR LEVELS 1,2,3
APPENDIX 3: CROSS-BORDER NAMES FOR LEVEL 4 WATERSHEDS
Cross-border names: In most cases, there were 2 or more watersheds (with different names on each side of the border) to combine into one North American Watershed. In most cases I used the name from one of the countries. Occasionally I had to invent a new name, and in 2 cases the name was already the same in each country. Finally, there were a few cases where a watershed inside a country (not crossing or touching the boundary) had to be renamed in order that other names made sense (e.g. Yukon and Columbia rivers). Yellow highlighting shows where several names from one country were merged into a single name. ====== TO DO (5 Oct 2009) ======
When we release the new watersheds layer, will we also release the rest of the collection? (Last release was June 2004). Policy on UIDENTs (unique identifiers) Some features were split in this editing session (and earlier ones) and their UIDENTs are no longer unique. Need a policy on what to do in this case. What is our standard for spatial accuracy? The standard of +/- 19,000 metres mentioned in the 2008 process notes seems generous! GIS format (1): The 2008 data came to me as shapefiles and a file geodatabase. The 2008 work was done in geodatabase. Given the notoriety of the shapefile format and the demise of the coverage, I decided to stay with the geodatabase – another unilateral decision. Even if we release shapefiles to the public, we need to keep our maintained data as one package of integrated data layers. GIS format (2): The spatial resolution (coordinate resolution) of the geodatabase I inherited was .001 metres. Seems fine to me, but it would be good to get concensus. Longer-term plan for data creation, maintenance, and management. Directory: gad gad -> Robert "Bobby" Jones gad president gad -> Unit A. Gadgets Vocabulary Make sure that you know the words gad -> Committee on the judiciary subcommittee on constitution, civil rights and civil liberties united states house of representatives gad -> Note: almost all of these are now out-of-date. gad -> Date meet location gad -> Georgia District of Kiwanis Board Meeting March 11, 2016 7: 00 P. M. Hampton Inn, Carrollton, Ga gad -> City of atlanta, dolphins swim team gad -> Sixty-seventh session Item 70 (a) of the provisional agenda Download 423.29 Kb. Share with your friends:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||