Model specifications for place-specific sediment regulation models developed for the Puget Sound, Madagascar, and Rocky Mountains case studies
Sediment source: Areas where soil erosion occurs, and water transports sediment to a downstream locations.
Sediment sink: Areas where sediment is deposited. In the ARIES sediment sink models, deposition is typically modeled in floodplains and reservoirs.
Sediment beneficiaries: For the preventive benefit case, sediment use models map the location of potential beneficiaries of avoided 1) detrimental sedimentation, 2) detrimental erosion, or 3) excessively turbid surface water. In the provisioning benefit case, sediment use models identify regions that could benefit from sediment deposition.
Sediment flow: Sediment flows are routed across the landscape through topography and stream networks. Once sediment is in a stream, it can be deposited in floodplains, reservoirs, or lakes and coastal zones. Sediment delivery can be attributed to upstream sediment sources, and deposition can be attributed to upstream sediment sinks, which provide the ecosystem service of sediment regulation.
Model structure and assumptions
Case studies 1. Western Washington State 2. Madagascar 3. Rocky Mountains
Sediment source models. Other ecosystem services researchers have also attempted to map sedimentation values. While Eade and Moran (1996) and Tallis et al. (2011) modeled sedimentation using the USLE, Egoh et al. (2008) and Wendland et al. (2010) used other proxy data. Egoh et al. combined local estimates of soil erosion potential with expert rankings of the ability of tree canopy cover to prevent erosion. Drawing on Quinton et al.’s (1997) work in semiarid Spain, Egoh et al. note that soil erosion is slightly reduced at about 30% tree canopy cover and significantly reduced at about 70% tree canopy cover. By combining areas of high erosion potential and <30, >30, or >70% tree canopy cover, they estimate spatially explicit values of vegetation for erosion control. Wendland et al., noting the established link between forest cover and sedimentation for Madagascar (Albietz 2007), map upstream forest cover from population centers, irrigated rice fields, and mangroves – areas that benefit from sediment-free water. We use these authors’ work to inform our sedimentation models. We designed probabilistic sediment source models that could complement the deterministic (RUSLE) models where their results are expected to be less reliable. For each model, we set annual sediment loss as a function of runoff or rainfall erosivity, vegetative maturity (i.e., vegetation characteristics that affect runoff), and soil erodibility. We set vegetative maturity as a function of vegetation type and percent tree canopy cover for all case studies, and added successional stage as a further influence on vegetative maturity for Western Washington and the Rocky Mountains. We set runoff as a function of annual precipitation and tropical storm probability for Madagascar. For Western Washington, we used annual precipitation as a direct influence on annual sediment source value. Where spatial data for rainfall erosivity are available at an adequate spatial resolution, these data will replace the runoff node. Finally, we set soil erodibility in all models as a function of hydrologic soils group, soil texture, and slope. In Western Washington, we added slope stability as a fourth influence on soil erodiblity, since these data were available for the region. We discretized percent tree canopy cover using Quinton et al.’s (1997) breakpoints of 30% and 70% cover discussed above and used Jenks natural breaks to discretize all other continuous variables. In the Rocky Mountains, we set the presence of a burned area as an influence on soil erodibility, successional stage, and canopy cover, with wildfire reducing canopy cover, setting back the successional stage, and increasing soil erodiblity. We estimated priors based on their corresponding spatial datasets for each case study. For Madagascar, we completed the contingent probability table for vegetative maturity by “pegging the corners” (Marcot et al. 2006) for highest vegetative maturity under conditions of very high tree canopy cover and forest/wetland vegetation type, the lowest vegetative maturity for very low tree canopy cover and cropland/developed vegetation type, and interpolating intermediate values. We set forests and wetlands as having the highest maturity, followed by degraded forests, savannas, and cropland/developed land. For Western Washington, we set conditional priors for vegetative maturity by ranking the order of importance of child nodes, with vegetation type most important and successional stage and percent tree canopy cover progressively less important. We set the greatest vegetative maturity (all else being equal) for Western Washington for forests and wetlands, followed by shrubland and grasslands, followed by cropland, barren, and developed. For the Rocky Mountains, we set forests, grasslands, shrublands, or wetlands to the highest vegetative maturity, followed by invasive vegetation, barren or urban, and agriculture. We set the contingent probability table for runoff in Madagascar by pegging the corners for high precipitation and/or high tropical storm probability leading to the greatest runoff and vice versa, with intermediate values interpolated. We set values in the contingent probability table for erodibility as greatest on steep, coarse soils with high infiltration potential (hydrologic soils group A) and erodibility (for Western Washington), and vice versa, with intermediate values interpolated. We set the top node, the annual sediment erosion source value, at zero for all soils with very low erodibility, set it at its highest on very erodible soils with very high runoff/rainfall erosivity and no vegetative maturity, and interpolated intermediate values.
Sediment sink models. Erosion sinks are areas where sediment accumulates as it flows downhill or downstream. We only consider the deposition of sediment in floodplains and reservoirs, as opposed to sediment carried and then deposited by overland flow before reaching a stream. We define sediment deposition (“Floodplain sediment sink”, measured in tons of sediment per year) to be a function of three stream and floodplain variables – stream gradient, floodplain tree canopy cover, and floodplain width – plus dams that cause sediment deposition in reservoirs. The erosion sink models for the four case studies are identical, except that we set prior probabilities for artificial infrastructure much lower for less developed settings (e.g., Madagascar) than in more developed settings (e.g., Western Washington, Rocky Mountains). We discretized floodplain tree canopy cover using Jenks natural breaks and stream gradient using breakpoints of 0-2% for low gradient streams, 2-5% for moderate gradient streams, and >5% for high gradient streams. We based priors for all nodes on relevant spatial data for each case study. We set the contingent probability table for annual sediment sink by assuming deposition to be greatest in low-gradient streams with wide floodplains and high levels of tree canopy cover, and lowest under the opposite conditions, with intermediate values interpolated. The presence of reservoirs, which create slack water flow conditions, leads to high deposition levels in all circumstances.
Sediment use models. While not explicitly presenting ecosystem service flow model results, both Tallis et al. (2011) and Wendland et al. (2010) incorporate beneficiaries in their sedimentation models. Tallis et al. map the locations of reservoirs where avoided sedimentation is a benefit, while Wendland et al. map human population density (for drinking water), mangroves (for avoided sedimentation of fish habitat), and rice fields (for avoided crop damage). Mapping these beneficiaries in ARIES can often be done with a single spatial data layer or simple GIS operations rather than Bayesian networks. For instance, we can map: 1) the location of reservoirs, drinking water intakes, and navigation infrastructure (where high turbidity or excess sedimentation are undesirable), 2) floodplain farmers (where sedimentation may be beneficial or undesirable, using a simple overlay of floodplains and farmland), or 3) erosion-prone farmers (where erosion is undesirable, by simply intersecting erosion sources and farmland).
Sediment flow models. The source and sink models estimate the annual quantity (in tons or kg of sediment per hectare) of sediment that could potentially be eroded from one part of the landscape (in the source model) and deposited in another (in the sink model). For the preventive benefit case, use models map the location of potential beneficiaries of avoided 1) detrimental sedimentation, 2) detrimental erosion, or 3) excessively turbid surface water. In the provisioning benefit case, use models identify regions that could benefit from sediment deposition. As discussed in the introduction to this chapter, flow models are not necessary to calculate the benefit of avoided erosion: we simply estimate the erosion source value with and without vegetation in order to determine the effects of vegetation on reduced erosion. For the other beneficiary classes, the flow models describe the amount of beneficial or detrimental sediment delivered or the amount of sediment carried in flowing water (i.e., turbidity). Since we do not model wind-based erosion, sediment flows are modeled using a relatively simple hydrologic model. We use hydrography and SRTM elevation data to derive flow direction to route water across the landscape and through waterways. During flood events, sediment can be deposited in floodplains, thus floodplain extents and the presence of levees are used in the water and sediment routing models. Finally, dams are included in the flow models, because essentially all sediment will be deposited into a reservoir as the speed of flowing water slows dramatically when a stream empties into a reservoir. Whenever sediment is deposited on the landscape, its effect, whether beneficial or detrimental, is assigned to any users in the same spatial location as the sink. If no human users (people or assets) are present at the sink site, then no service is accrued by sediment deposition in that location. Key outputs from the flow models include: 1. Possible sediment flow: The downstream movement of sediment when not accounting for sediment sinks. 2. Possible sediment source: Areas which, based on the flow pattern of sediment but disregarding the effects of sinks, provide sediment which reaches downstream users who either benefit from or are damaged by sediment delivery. 3. Possible sediment deposition beneficiaries; Possible reduced sediment deposition beneficiaries; Possible reduced turbidity beneficiaries: Beneficiaries who receive beneficial or detrimental sedimentation when accounting for flow paths but not sinks. 4. Actual sediment flow: The downstream movement of sediment that accounts for sediment sinks. 5. Actual sediment source: Areas which, based on the flow pattern of sediment and accounting for sinks, provide sediment to downstream users who either benefit from or are damaged by sediment delivery. 6. Utilized deposition: Depositional areas that undergo sedimentation, receiving upstream sediment and actively performing a sediment trapping function. 7. Actual sediment deposition beneficiaries; Actual reduced sediment deposition beneficiaries; Actual reduced turbidity beneficiaries: Beneficiaries who receive beneficial or detrimental sedimentation when accounting for flow paths and sinks. 8. Absorbed sediment flow: The flow of sediment that does not reach downstream beneficiaries who benefit from either avoided detrimental sedimentation or beneficial sediment delivery. 9. Negated sediment source: Areas which, due to deposition occurring in downstream sink areas, do not provide sediment to downstream users who either benefit from or are damaged by sediment delivery. 10. Lost valuable sediment; Blocked harmful sediment; Reduced turbidity: Beneficiaries who either receive less beneficial or detrimental sediment as a result of sinks.
|Models||Data theme||Source||Spatial extent||Spatial resolution||Year|
|Source – All models||Slope||Derived from global SRTM data||Global||90 m||n/a|
|Source – Madagascar||Annual precipitation||WorldClim||Global||30 arc-seconds||1950-2000|
|Source – Western WA||Annual precipitation||PRISM/Oregon State Univ.||United States||800 m||1971-2000|
|Source – Madagascar||Average annual runoff||SAGE/Univ. of Wisconsin||Global||0.5 degree||1955-1990|
|Source – Rocky Mountains||Burned area||GeoMAC fire perimeter data||United States||vector polygon data||2000-present|
|Source – Madagascar||Hydrologic soils group||Gately (2008) using FAO soils data||Global||0.083 degrees||n/a|
|Source – Rocky Mountains, Western WA||Hydrologic soils group||SSURGO soils data||United States||30 m||n/a|
|Source – Rocky Mountains||Rainfall erosivity||USGS Effects of Energy Development in the Rocky Mountain Area Project||Colorado, New Mexico||30 m||n/a|
|Source – Madagascar||RUSLE factors & avg. annual soil loss||Yang et al. (2003)||Global||0.5 degree||2000|
|Source – Western WA||RUSLE factors & avg. annual soil loss||U.S. EPA (2010)||United States||Vector polygon data (HUC 8 watersheds)||Not available|
|Slope stability||Washington Dept. of Natural Resources||Washington State||30 m||n/a|
|Source – Madagascar||Soil texture||FAO soils||Global||0.083 degrees||n/a|
|Source – Rocky Mountains, Western WA||Soil texture||SSURGO soil data||United States||30 m||n/a|
|Successional stage||BLM/Interagency Vegetation Mapping Project||Western Washington & Oregon||25 m||1996|
|Source – Rocky Mountains||Successional stage||LANDFIRE||United States||30 m||2008|
|Source – Madagascar||Tree canopy cover||GLCF/Univ. of Maryland||Global||1 km||2000|
|Source – Rocky Mountains, Western WA||Tree canopy cover||NLCD 2001||United States||30 m||2001|
|Source – Madagascar||Tropical storm probability||CIESIN/Columbia University||Global||2.5 minute||1981-2000|
|Source – Madagascar||Land cover||FTM (Madagascar National Mapping Agency)||Madagascar||Vector polygon data||Mid-1990s|
|Source – Rocky Mountains||Land cover||Southwest Regional GAP Analysis (SWReGAP)||AZ, CO, NM, NV, UT||30 m||1999-2001|
|Land cover||Northwest GAP Analysis (NWGAP)||CA, ID, MT, OR, WA, WY||30 m||1999-2001|
|Source – Western WA||Land cover||NLCD 2001||United States||30 m||2001|
|Sink – Madagascar||Floodplain tree canopy cover||GLCF/Univ. Maryland tree cover & Dartmouth Flood Observatory||Global||1 km||2000|
|Sink – Western WA, Rocky Mountains||Floodplain tree canopy cover||NLCD 2001 & FEMA Q3 Flood Data||United States||30 m||2001|
|Floodplain width||FEMA Q3 Flood Data||United States||Vector polygon data||Varies|
|Sink – Madagascar||Stream gradient||BD500 (Madagascar infrastructure data) & SRTM slope||Madagascar||90 m||n/a|
|Sink – Western WA||Stream gradient||DNR hydrography & SRTM slope||Western Washington||90 m||n/a|
|Sink – Rocky Mountains||Stream gradient||NHD hydrography & SRTM slope||United States||90 m||n/a|
|Use – Madagascar||Land cover||FTM (Madagascar National Mapping Agency)||Madagascar||Vector polygon data||Mid-1990s|
|Use – Western Washington||Land cover||NLCD 2001||United States||30 m||2001|
|Use & flow – Madagascar||Floodplain extents||Dartmouth Flood Observatory||Madagascar||Vector polygon data||Based on 2003-2010 flood data|
|Use & flow – Western Washington, Rocky Mountains||Floodplain extents||FEMA Q3 Flood Data||United States||Vector shapefile||Varies|
|Sink, use & flow – Madagascar||Dams||BD500 (Madagascar infrastructure data)||Madagascar||Vector point data||Not available|
|Sink, use & flow – Rocky Mountains||Dams||National Hydrography Dataset||United States||Vector polygon data||2005|
|Sink, use & flow – Western Washington||Dams||Oak Ridge National Laboratory||United States||Digitized reservoir locations for Western WA||2005|
|Flow – All models||Elevation||SRTM||Global||90 m||n/a|
|Flow – Madagascar||Hydrography||BD500 (Madagascar infrastructure data)||Madagascar||Vector line data||n/a|
|Flow – Rocky Mountains||Hydrography||National Hydrography Dataset||United States||Vector line data||n/a|
|Flow – Western Washington||Hydrography||Washington DNR||Washington State||Vector line data||n/a|
|Flow – Rocky Mountains||Levees||FEMA||United States||Vector line data||Varies|
|Flow – Western Washington||Levees||County GIS offices||King, Lewis, Pierce counties||Vector line data||Varies|
Albietz, J. 2007. Watershed protection for ecosystem services in the Makira Forest Area, Madagascar. Tropical Resources Bulletin 26: 21-30.
Eade, J.D.O. and D. Moran. 1996. Spatial economic valuation: Benefits transfer using geographical information systems. Journal of Environmental Management 48: 97-110.
Egoh, B, et al. 2008. Mapping ecosystem services for planning and management. Agriculture, Ecosystems and Environment 127: 135-140.
Gately, M. 2008. Dynamic modeling to inform environmental management: Applications in energy resources and ecosystem services. MS Thesis, University of Vermont, Burlington, VT.
Marcot, B.G., et al. 2006. Guidelines for developing and updating Bayesian belief networks applied to ecological modeling and conservation. Canadian Journal of Forest Research 36: 3063-3074.
Quinton, J.N., et al. 1997. The influence of vegetation species and plant properties on runoff and soil erosion; results from a rainfall simulation study in south east Spain. Soil Use and Management 13: 143-148.
Renard K.G., et al. 1996. Predicting Soil Erosion by Water: A Guide to Conservation Planning with the Revised Universal Soil Loss Equation (RUSLE). Handbook 703, US Department of Agriculture, 404 pp.
Tallis, H.T., et al. 2011. InVEST 2.0 beta User’s Guide. The Natural Capital Project: Stanford. U.S. Environmental Protection Agency (USEPA). 2010. EMAP-West Metric Browser. Accessed March 3, 2010 from: http://www.epa.gov/esd/land-sci/emap_west_browser/EMAP-West_Metric_Browser.htm.
Wendland, K.J., et al. 2010. Targeting and implementing payments for ecosystem services: Opportunities for bundling biodiversity conservation with carbon and water services in Madagascar. Ecological Economics 69: 2093-2107.
Yang, D., et al. 2003. Global potential soil erosion with reference to land use and climate changes. Hydrological Processes 17: 2913-2928.
Acknowledgements and additional contributors
Dave Batker, Jim Pittman, and Paula Swedeen provided data and reviewed models for the Western Washington case study. Miro Honzak provided data and reviewed models for the Madagascar case study. Initial ARIES data and models for the Rocky Mountains were developed by students participating in a graduate level ecosystem services modeling course taught in the University of Denver’s Department of Geography in the fall of 2011; James Reed, Darius Semmens, and Todd Hawbaker assisted with further data and model refinement.