|
High Water GIS
A GIS firm in Modestro, Calif. takes on the job of mapping high water
marks before recent flood waters recede. The coordination of image processing
and GIS proves key to quick map updating.
By Kevin P. Corbley Stanislaus County, Calif., rang in the New Year with a visit from the Pineapple Express, a Pacific storm that dumped more than 7 inches of rain in the first two days of 1997 on the already drenched San Joaquin Valley. By January 4, the Tuolumne River in Modesto had risen beyond the 500-year flood stage, obliterating the high water mark from any flood previously recorded.
Even as the water crested and began to recede, local officials had no time to breathe a sigh of relief. Weather forecasts called for another tropical storm within a week. Stanislaus County and the city of Modesto teamed with a local GIS firm to map the new high water mark so they could better prepare for the next predicted floods.
The emergency mapping project proved an auspicious start for the county's nascent GIS, which was pressed into service to assist with the mapping and other disaster relief efforts. And the speed with which the mapping was completed underscored the added dimension that raster image processing capabilities can bring to a GIS.
"If there were any doubts in the county government about the decision to build the GIS, they were washed away by the flood," said Raymond Kinser, president of California CAD Solutions Inc. (CalCAD) of Modesto. "The GIS solidified its reputation even though it wasn't finished."
CalCAD was well positioned to help out during the floods, having been involved with the county's enterprise GIS development project since its inception. The GIS was being built with AutoCAD from Autodesk Inc. of San Rafael, Calif., and ArcView from ESRI Inc. in Redlands, Calif. The Stanislaus GIS office and CalCAD both had extensive experience with their software packages predating the GIS.
However, it was the addition of an image processing package to CalCAD's software arsenal that enabled the firm to create a series of critical flood maps within a few days after the waters crested. Late in 1996, CalCAD had obtained ER Mapper image processing software from Earth Resource Mapping in San Diego to bolster its raster data handling capabilities.
"With 1-meter (spatial resolution) satellites coming in the next year, the writing was on the wall in terms of the value of raster imagery in the GIS environment," said Kinser. "We knew we had to integrate image processing into our GIS capabilities." Developing the GIS
When Stanislaus County began developing the GIS in 1993, flood mapping was not on top of the agenda. The county, which was sharing the GIS data with cities such as Modesto, planned to use the GIS in numerous applications from engineering and emergency response to libraries and irrigation.
"From the outset, we wanted to include as many county and city agencies as possible in the GIS," said Gity Miller, GIS manager in the county Public Works department. "One of our major goals was to streamline the permitting and zoning processes."
Stanislaus wanted to reduce the workload involved in notifying residents of proposed zoning changes. County staff had to page through paper parcel maps to manually compile lists of neighbors who had to be notified of zoning hearings. With a digital parcel layer in the GIS, they would simply click on appropriate lots to generate an address list.
The county took charge of creating the parcel basemap. Existing paper maps were scanned and brought into AutoCAD where the parcels were mosaicked into a single basemap layer. The staff shot about 1,500 GPS points throughout the county to correct the digitized parcel layer and rectify it with other data layers already in the GIS.
CalCAD, which specializes in linking databases to digital maps, was working on completing a new street centerline file for the Stanislaus GIS. An existing centerline file containing correct street names and addresses in a NAD 27 projection was used in this phase of the project. In the course of building the new streets centerlines, CalCAD had to rubber sheet 65,000 street segments to the new parcel basemap, which was created in NAD 83, using GPS control points taken at many road intersections in the county.
Development of the parcel basemap and the street centerlines was only four months from completion when the Pineapple Express roared into the valley. Dealing with Rising Waters
Central California was soaked before the Pacific storm even arrived, and flood records were quickly broken. On January 2, the Army Corps of Engineers had to open the flood gates on the Don Pedro Dam for the first time ever. The dam is just outside the county and up river from Modesto. Water rushed out of the dam at 59,000 cubic feet per second - an incredible flow, but only half the rate that would have occurred had the dam not been there.
On the Tuolumne River in downtown Modesto, flood stage is rated at 55 feet. The waters would ultimately crest there at 70.9 feet at noon on January 4.
Before the crest, Stanislaus had already opened its Emergency Operations Center (EOC) to coordinate evacuations and direct relief efforts. CalCAD had earlier obtained digital floodplain maps from the Federal Emergency Management Agency (FEMA) and offered them to EOC personnel. They viewed the FEMA 100- and 500-year floodplain maps on a laptop computer, but needed hard copies and a street reference to utilize them properly.
"We overlaid the FEMA flood maps onto the nearly completed GIS street centerline map using the river centerline as the common reference point," said Robert Spiva, a CalCAD principal. "This put the 100- and 500-year floodplain into perspective with city streets."
EOC personnel were anxious to get a hold of the overlaid FEMA map because it would show them which streets and properties would be under water at various stages of flooding. This information made evacuating citizens a much more orderly and efficient process. Equipped with hard copy maps in the fields, EOC workers evacuated more than 1,600 properties.
Despite the improved efficiency provided by the maps, EOC quickly realized something was not right. Many areas that should have been under water were not, and even worse, areas that were expected to remain high and dry were deluged.
"The actual flooding had exceeded the 500-year floodplain estimates, which meant the FEMA maps had to be updated," said Spiva. "New mapping had to be done quickly while the water was still there."
Predictions of more storms within a week heightened the urgency of the situation. Mapping the high water mark from the first flood could prove critical in handling the next one. Re-mapping the High Water Mark
As the water was cresting on January 4, Geonex Inc., a Sacramento aerial survey firm, flew much of the county and acquired black and white photos at 1:1000 scale. One hundred and seven 9" x 9" contact prints were provided to the county which in turn supplied them to CalCAD.
"Several mapping vendors told us it would be impossible to process the air photos and rectify them in a month, much less a week," said Spiva. "We decided to do it in our office using our new ER Mapper system."
The firm used in-house scanning equipment to create digital files from the 107 aerial photos. Each was scanned at 600 dots per inch resulting in individual file sizes of 28 megabytes. As the scanning was underway, a local survey firm was sent into the field with copies of the photos to plot ground control throughout the city of Modesto. There, GPS points were post-processed by the surveyor and e-mailed to CalCAD's office.
After three days of scanning and collecting ground control, the county requested that mapping focus only on the Modesto area - the most populous and hardest hit. This meant the mapping would only involve 20 of the air photos, which lightened the work load. The scanned photos were input into ER Mapper for rectification and mosaicking of the image processing and file handling work was carried out on standard Pentium 120 desktop computers with 64 megabytes of RAM. Earth Resource Mapping's San Diego office provided CalCAD with technical support as needed during the project.
"We rectified each scanned photo file individually using the GPS points, and then we mosaicked three or four at a time to keep the process moving," said Spiva.
ER Mapper was well suited for the task of handling extremely large data files very quickly. It uses a unique algorithm concept whereby it stores processing commands and transformation operations independently of the image data. This enables the user to apply multiple formulas, routines and filters in real time directly to the displayed data without creating an intermediate data file. Less disk space is required, and complex processing of large files is completed very quickly.
"The file handling speed made our jaws drop," said Spiva. "Screen regenerations of the entire 20-photo mosaic ran in about three seconds every time we applied a balancing routine or other enhancement function."
After all 20 photos were mosaicked, the file was 400 megabytes. With cropping and feathering, the final file was 214 megabytes. Throughout the mosaicking process, technicians took advantage of automated histogram balancing routines in the software to even out radical differences in contrast along the seams of the photos.
"By the middle of the week, we already knew the second rainstorm would not hit our area," said Spiva, "but we still completed the entire rectification and mosaicking process in about three days."
Since the GIS basemap resided in AutoCAD, the GIS technicians saved the air photo mosaic in a TIF format and exported it from ER Mapper. The flood mosaic was overlaid on the street centerline file to double check its accuracy, and then technicians used heads-up digitizing to delineate the boundaries of the high water mark from the actual flood. The result was a GIS vector layer showing the new high water line.
"To make sure we could determine the exact locations of the high water mark on the GIS basemap, we digitized on one computer screen while viewing the flood mosaic in ER Mapper on another screen because the resolution of the image processing system was so much better," said Spiva.
CalCAD supplied copies of the photo-mosaic to the Stanislaus County GIS Department to make small prints for use by field personnel. In the weeks of clean-up that followed, numerous county and city agencies used the flood prints to locate damage, prioritize the hardest hit areas and direct relief efforts. Preparing for Future Floods
Stanislaus County and the citizens of Modesto are thankful the second wave of flooding never arrived. Although they were ready with an updated high water map within a week after the New Year's flood, the new maps did not have to be used -- at least not at that time.
Miller says the 1997 high water mark will be overlaid on the GIS parcel map to assist in emergency operations related to future flood events. If flood waters rise again on the Tuolumne River, the county will be able to view the new floodplain overlay and know which properties will be submerged at various flood levels. Equipped with this up-to-date information, county officials expect to contact individual property owners to evacuate their homes before flood water reaches them.
"The Modesto flood of 1955 was thought to be the flood to end all floods, but then 1997 came along," said Spiva. "We hope there's never another one like it, but if there is, we'll be much better prepared to handle it." About the Author:
Kevin Corbley is a freelance writer specializing in remote sensing, GIS and GPS. He is located in Denver, Colo., and may be reached at 303-987-3979 or by e-mail at [email protected]
Back
|
