Run-off and Land Use Change
This project involved building a first approximation of expected rainfall runoff as a function of rainfall events, runoff estimates, and mapped land use changes over three intervals for Burlington, VT.
Runoff and Land Use Changes
Method, Data, and Error
For this project five adjacent residential lots were documented for land use changes over two dates of orthophotography (Fig 1-3). For self consistency between projects, an area was chosen with visible, but moderate changes on N Willard Ave, 3 blocks from from the intersection of Riverside Ave and Intervale Rd.
These adjacent plots were scaled to nearly the same size and imported into a 1mm x 1mm grid system in Inskscape. Land use was color coded into six categories: Houses/Roofs, Sidewalks, Parking (paved), Parking (dirt), Lawn/Trees, and Fences/Walls. Summed grid coverage by land use type was normalized by total image area to derive relative land use percentages (Fig. 4)
Each of these land use types was categorized by land type according to Urban Hydrology for Small Watersheds (1986) with a corresponding runoff estimate percent (curve numbers) based on this land use classification, as follows:
Lawn/Trees - Open Space Good Condition (75%+ grass) Group A
Houses/Roofs - Parking Lots, Roofs, Driveways
Sidewalks - Streets/Roads (paved)
Parking (paved) - Parking Lots, Roofs, Driveways
Parking (dirt) - Dirt Road, Group D
Fences/Walls (compact soil) - Open Area Poor Condition (50%- grass)
Weighted averages for runoff percentage were calculated for each time-frame's land use of the study area, with an assumed full cover of wooded area in 1600CE. These were then scaled by 2 year, 1 hour, 100 year 1 hour, 2 year 24 hour, and 100 year 24 hour expected rainfall events (US. Department of Commerce and Weather Bureau, 1959) (See Runoff Calculation Spreadsheet). The results are displayed in Fig. 5. Note the extrapolation from trends derived from n=3 data points are necessarily approximate, confidence intervals are displayed to illustrate better expected ranges of runoff over the projection.
Sources of error include estimates of runoff percent (Urban Hydrology, 1986), poor estimations of land use type or soil quality, extrapolation of 100 year rainfall estimates (USDC & WB, 1959), not diverting roof rainfall percent to the slope adjacent land use it would naturally drain to, and not extending the site margins to include 1/2 of the roads for land use percentage calculations.
Discussion
Water infiltration and surface run off both depend on ground porosity and the interconnectedness of those pores, which in turn have defining factors such as mineral composition, animal and root paths, plant uptake, density and compaction. As humans tend to cultivate their surroundings for specific purposes, our land use has a large impact on water infiltration. As populations grow, and the density of these populations do as well, development of housing and parking addresses these needs, but usually at a cost of natural green spaces. These green spaces - forests, woods, grassland, even lawns - have high rates of water infiltration, whereas compact dirt, gravel, or paved areas are rather impermeable and drive surface runoff which leads to saturation, erosion, and landslides.
In comparing the development of of the 5 adjacent lots from 1998 and 2018, we can see a few trends in the land use. Paved and unpaved parking are universally expanded at the expense of green space and roofs (especially those that appear to be carports), which results in a 6.5% increase in rainfall runoff over this area (See Runoff Calculation Spreadsheet).
In contrast, we can also see the relatively stable levels of runoff generated ~1600CE assuming a fully wooded area in precolonial times. The development of roads (which should have a larger land use impact than measured), and buildings grow with settled permanent population. A more expansive study should show Fig. 5 with the additional axis of population growth in the Burlington area for comparison to a non-linear regression of rainfall runoff trend. In all the 1998 calculated estimated runoff is 2.25 times higher than assumed preindustrial/precolonial amount, while the 2018 calculated estimate of runoff is 2.39 times higher than from 1600.
This higher runoff is especially worrisome due to the increased parking that is used. Vehicles can leak small amounts of complex chemical compounds while functioning, and larger amounts when encountering issues. The increased runoff in Burlington drains either to the Winooski, or directly into Lake Champlain.
- Urban Hydrology for Small Watersheds. (1986). Chapter 2: Estimating Runoff. Technical Release 55. Second Ed., June.
- US. Department of Commerce and Weather Bureau. Rainfall Intensity-Frequency Regime. (1959). Part 4 - Northeastern United States. Technical Paper no. 29. U.S. Government Printing Office, Washington 25, D.C. May.