FLOODING IN CENTRAL INDIANA - 
 A FORECAST CHALLENGE
 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
   
SCOTT CASSIN 
Being the smallest continental state west of the Appalachian Mountains, Indiana is a relatively flat
 state in terms of its topography.  The elevation ranges from a low point of 320 feet above sea level
  in the far south to a high point of 1,257 feet above sea level in the north.  Most of the state
   however averages around 700 feet with a general slope from north to south toward the Ohio and
    Wabash Rivers in the southwestern portion of the state.  Additionally, Indiana is part of the
     Great Midwestern Corn Belt, therefore predominantly covered by vast farm land with small urban
      developments peppered throughout the state.  Although the soil is well-drained, fertile, and
       ideal for agriculture, the very nature of the state's topography can lead to devastation,
        primarily as a result of general and flash flooding from over-flowing rivers and other
         bodies of water.
  
 
   
As with any state, Indiana has an immense network of streams, lakes, and rivers, all of which are
 capable of overflowing their banks and causing widespread damage to property and potential loss
  of life.  The focus of this commentary will be on Central Indiana, home to the state's capital,
   Indianapolis, and the difficulties of forecasting generalized flooding in the region.  The
    emphasis will center on the west fork of the White River, the area's major tributary that
     runs through the city and eventually empties into the Wabash and Ohio Rivers downstream.  The
      ultimate destination of these drainage basins is the Mississippi River and the Gulf of Mexico.
    
Indiana is geographically located in an area subject to a typical trough-ridge pattern throughout
 much of the year.  This pattern and the associated Low pressure systems that frequently ride along
  it, put Indiana in the direct path of these moisture-latent storms, which is even more pronounced
   throughout the northern hemisphere winter months.  Although flooding can, and does, occur in
    any season, it is in the winter and early spring that the problem is often magnified.  It is
     the abundance of moisture coupled with the rapid runoff on non-vegetated and frozen ground
      that leads to a potential significant flooding risk.  Add to this any snow melt, and the
       problem is even further exaggerated.   Unfortunately however, many of the events are
        either under-predicted or not predicted at all in some locations.  These occurrences,
         known for their newsworthiness, generally occur two to three times per year at some
          point along the White River causing widespread damage and hardship.
  
Further analysis of the issue reveals that the problem is not necessarily determining when precipitation
 will occur, or even generally where, but more so with when and where any associated flooding will occur
  as a result.  Currently, the National Weather Service (NWS), which is part of the National Oceanic
   and Atmospheric Administration (NOAA), has the responsibility by law of issuing river forecasts
    and flood warnings.  The NWS, supported by thirteen River Forecast Centers throughout the
     United States, uses multiple sources of data and other inputs when developing its flood
      forecasts.  Most of this data comes from the office of the United States Geological
       Survey (USGS).  The USGS is the principal source of data on river depth and streamflow
        and obtains this information from direct gaging of the rivers.  At this time there are
         around 160 streamflow gaging stations in Indiana.  Of these, around seventy are along
          the White River, all recording parameters such as gage height, discharge rate, and
           water temperature.  This is only the first piece of the puzzle however.  Determining
            the existing condition of local rivers may be the easiest piece, since much of it
             is automated and directly measurable.  It is predicting what the future conditions
              along the river will be is what becomes difficult.
  
 
   
Because much of central Indiana is urbanized and part of the White River watershed and basin, it doesn't
 take much of a precipitation event to cause problems along the river.  In the winter and early spring
  when the ground is still frozen and lacking moisture-absorbing plant life, there is nowhere for the
   excess water to go, filling up small streams, creeks, ponds, lakes, and rivers rather quickly.  An
    early spring with a heavy snow pack anywhere within the basin compounds the problem.  As the
     temperature and dew point rise above the freezing point, the snow begins to melt at a rapid
      pace, thereby raising water levels even further.  Areas to the south of the city of Indianapolis
       are especially prone, since the river flows from north to south.  Huge inputs from a severe
        storm, substantial snowmelt, days of stratiform precipitation, or a combination thereof,
         can put tremendous strains on the river system and send it all rapidly downstream
          to low-lying areas.
  
One of the problems discovered associated with determining when and where a flood will occur
 along the White River is in the process itself.  To develop its forecast, the NWS utilizes
  a complex system of models to determine how the river will respond to a precipitation input.  These
   models are developed for specific forecast points along the river.  As heavy rainfall is predicted
    for the river basin, the forecasted amounts are then added to the model to determine how the river
     will react in response to the input.  In addition, water entering the basin from other
      tributaries upstream and from any snow melt is also taken into consideration.  The river
       forecast is then produced and updated as needed.  However, it is just that, a forecast
        and like any forecast, a slight change in any of the input data can cause a significant
         change in the output.  What has been happening, as it appears, is that the models used
          are slow to respond to changes in area watershed characteristics.  The central Indiana
           region has seen explosive growth over the last twenty years or so and therefore has
            undergone, and continues to undergo, the rapid development of urban areas, drainage
             improvements, and levee and dam construction.  These along with natural changes over
              the years have dramatically changed the way water flows into and down the river
               system.  Additionally, along with urbanization comes problems associated with
                it such as blocked or re-directed drainage systems as well as sometimes adverse
                 political implications.
  
Floods continue to be one of the most frequent and most expensive of natural disasters in terms
 of economic loss and hardship.  Not counting droughts, as much as ninety percent of all damage
  related to natural disasters comes from floods or their associated debris and mud flows.  Since
   we cannot control the weather, we can better prepare ourselves for these catastrophic flooding
    events by adhering to certain standards as well as learning from previous events.  The federal
     government has already taken a significant step by updating the Federal Flood Maps for the
      area in 2007.  These maps through time have become outdated as progress has changed the
       topography.  According to Ed Ferguson, a local planning director for the area, the Federal
        Flood maps have not been updated in over ten years.  These new maps show much more detail
         and outline new areas of significant flooding potential.  With this new data, the NWS
          and its River Forecast Offices will need to update their models.  With the ability 
          to capture and store data being better than it ever has, the process of recalibrating
           models should be done more frequently to ensure better accuracy.  Additionally,
            building codes and local ordinances should be amended to reflect these changes 
            as well by not permitting new construction inside newly identified or existing
             flood plains.  This continuous cycle of evaluating and updating data, maps, and
              models is necessary to limit the amount of damage and loss of life not only along
               the White River basin, but in all basins across the country.
  
References
  
En-Ching, Hsu, and Rao, Ramachandra.  Streamflow Data Analysis.  Indiana:  Purdue University, 2008.
  
Ferguson, Ed.  Personal Interview.  7 March, 2008.
  
Indiana Flood Information Web Page.  08 Feb. 2008.  United States Geological Survey.  14 Mar. 2008. 
   http://in.water.usgs.gov/flood_new.
  
Indiana Water Resource Page.  12 Nov. 2006.  Indiana Department of Natural Resources.  11 Mar. 2008. 
  http://www.in.gov/dnr/water/water_availability/WaterResource/sflow_analysis.htm
  
Maskey, Shreedhar.  Modeling Uncertainty in Flood Forecasting Systems.  Delft, Netherlands:  Delft University, 2004.
  
National Weather Service - SRH.  27 Jul. 2007.  National Oceanic and Atmospheric Association.  
14 Mar. 2008.   http://www.srh.noaa.gov.
  
Stream Gaging and Flood Forecasting.   20 Sep. 1999.  United States Geological Survey.  
11 Mar. 2008.   http://water.usgs.gov/wid/FS_209-95/mason-weiger.html
     
                 
  
  
  
   
 |   
 | 
 
  
 |