Mississippi Basin Carbon Project Science Plan

Reconstructing effects of historical human land use

The reconstruction of predevelopmental conditions (and comparison with the present) is central to our hypothesis that significant amounts of additional carbon are buried during the deposition of sediments in terrestrial environments, and that this burial has occurred in large part because of human acceleration of erosion and modifications of soil and hydrologic systems and nutrient supplies. Moreover, we must estimate historical trends in our calculation of carbon budget terms associated with these processes. We have no direct measurement of the past behavior of these processes, and for many environments, we have no direct measurements of the present. Thus, we must develop and compare parallel syntheses of the present and the past based on our integration of databases and models. There are virtually no precedents for this kind of approach to terrestrial carbon cycle modeling. Thus our approach must be open and flexible enough to adapt to new ideas and to unexpected opportunities and stumbling blocks. We will begin with the following strategies:

1. Mapping of deposition and erosional areas: Depositional areas to be mapped include lakes, wetlands, reservoirs, active floodplains, irrigation systems, paddylands, and areas of aggrading alluvium and colluvium in low-order stream valleys. Many of these features are quite obvious, and reasonable maps can be constructed. Floodplains are mapped. Aggradational sites of alluvium and colluvium are not easily mapped. An approach that we will test involves apportioning sediment that cannot be accounted for by erosional mass balances to alluvium and colluvium. This approach was used by the classic studies by Trimble (1975), Trimble and Lund (1982) and Trimble (1983) for the Coon Creek watershed in Wisconsin. We will use landscape indices such as those of TOPMODEL to produce maps of regions that would have high runoff flows or large contributing areas, but low slope or channel gradient. Sediment would be apportioned into these areas based on existing or newly derived relations for "sediment-delivery ratios"—the ratio of specific sediment yield to upland erosion yield.
2. Depositional rates in wetlands, lakes, and reservoirs: For every wetland, lake, and reservoir, rates of clastic sedimentation and autochthonous carbon production will be estimated. Where direct data are available based on previous compilations or from data compiled by our own efforts, these will be used. However, these data will represent only a small percentage of depositional sites. For other sites we will use average rates extrapolated from similar features in the region compiled from pollen and reservoir data bases (or similar sources). Autochthonous carbon fixation will be estimated from estimates of nutrient loading and eutrophication. We will also use sediment data to extrapolate regional efficiencies of carbon burial (diagenetic effects). Because during predevelopment times there were no reservoirs and high rates of nutrient loading from fertilizer application and precipitation, the past may well be easier to reconstruct than the present.
3. Pre- and post-development soils and vegetation: For predevelopmental conditions, we will attempt to prepare maps of natural vegetation and carbon inventories in soils. In the case of soils, we will make direct comparison with modern soils. The soil maps will be used to estimate carbon concentrations for sediment derived from upland erosion and in reconstructing runoff patterns using landscape hydrology models. These models will be used in turn to estimate patterns of soil moisture that would have existed in the absence of recent colluvial and alluvial accumulation, reservoirs, irrigation systems, and wetland modification. Obviously, this will be an iterative process. Carbon inventories in vegetation will be estimated from other studies. We will try to include the effects of a controlled modern fire regime using data from a few investigations such as those at Konza Prairie and the Colorado Front Range.
4. Pre- and post-development runoff: Our efforts to model runoff will be based on a landscape hydrology model such as TOPMODEL. Presently several regions along the axis of the Mississippi River Basin appear to have a very high ratio of overland flow to total runoff. The amount of overland flow strongly affects the ability of rainfall to erode. It is not clear whether this is caused by soil texture alone or whether compaction due to grazing may play a role. We will examine whether the pattern changes pre- and post-development.
5. Estimation of upland erosion: For today's environment, we will use a combination of direct measurements and applications of erosion-dispersal models such as HUMUS/SWRRB and erosion models such as those developed by WEPP and by groups refining the USLE. Our initial efforts will focus on testing the models in small watersheds and then pilot sub-basins; however, we will also estimate areas of upland erosion and apportion upland sediment as described in (1) above. HUMUS/SWRRB and WEPP use a combination of mechanistic and empirical terms to calculate erosion. Although the USLE approach is strictly empirical (and not truly "universal"), it is based on a large number of measurements within the Mississippi River Basin, and we should be able to apply it to many upland areas within the basin. We will pay particular attention to verification of the modified USLE parameters which represent effects of land use, because these parameters will provide a direct means of comparison between simulations of contemporary and predevelopment conditions. The implementation of "predevelopment land-use" parameters within empirical soil loss equations will be our primary means of estimating predevelopment erosion rates. Wherever possible we will also use measurements of physical erosion at natural sites such as Konza Prairie, but such sites are rare. We will also apply the equilibrium-erosion model presented in Stallard (1995a,b). As described above, this approach estimates equilibrium physical erosion rates based on modern measurements that are usually available as normal water-quality measurements (sodium, silica, calcium, alkalinity, runoff) and not strongly affected by relatively recent erosion of soil A, B, and even soft C horizons in headwaters. Although the concept of "equilibrium" physical erosion cannot be applied too literally to the predevelopment Mississippi basin (particularly in areas of loess erosion), this method can give confidence limits for estimates derived by other means.‹

	Contents
	Modeling landscape mass balance
	Collaborative activities