BNPP/ASB Functional Value of Biodiversity Project – Phase II 

Appendix 6. Four steps for meso (& micro) scale models to clarify land use change effects on watershed functions

Step 1. Which watershed functions are relevant for which stakeholders in what context?

Step 2. How do the various watershed functions relate to ‘land use and land cover’ on   the basis of the various terms of the water balance?

Step 3. How are the various terms in the water balance reflected in the models?

Step 4. How do the models separate ‘inherent properties of sites’, ‘effects of land cover’ and ‘impacts of land use change’?

Step 1. Which watershed functions are relevant for which stakeholders in what context?

·        see phase 1 report

Step 2. How do the various watershed functions relate to ‘land use and land cover’ on the basis of the various terms of the water balance?

·        see phase 1 report

Step 3. How are the various terms in the water balance reflected in the models?

All models considered here follow a basic ‘water balance’ logic:

Rainfall -> Interception + (Infiltration + Runoff)

Infiltration -> Vegetation water use + baseflow

Runoff + baseflow are routed through a stream network

The models differ in the time steps: yearly for Fallow, monthly (with daily approximation) for WBM, daily for VIC, WaNuLCAS and GenRiver, 4-hourly for DHSVM.

The models also differ in the spatial resolution, but appear to handle ‘interception’ and ‘evapotranspiration’ issues in a similar way, driven by the energy balance (potential evapotranspiration).

Table 1. Order of magnitude estimate of the effect of land use change on total water yield as represented in level 2 models (vegetation-dependant water use, see Table 1 from Implementation Protocol).

Effects of land used change (‘deforestation’) on an increase in total water yield are, relatively speaking, the highest in a climate zone with the lowest annual rainfall. For annual rainfall amounts of over 2500 mm year^-1 the relative change in total water yield is likely to be less than 25% and thus within the likely inter-annual variability making it difficult to observe unless long time-series are available.

While an increase in annual water yield can be positive from a downstream water use perspective, especially if the flow can be temporarily stored in reservoirs in the river, the general fear is that this increased flow will largely be  in the form of ‘peak flow’, directly after heavy rainstorms, while the ‘baseflow’ that is, per m³ of river flow, of much higher potential value downstream may be reduced. To get this effect of land use change correctly predicted, we have to focus on how the ‘infiltration process’ is described in the various models.

The ‘infiltration versus runoff’ partitioning is handled with different degrees of sophistication. All models keep track of current soil water content and the soil recharge capacity that is stimulated by antecedent water use. Some of the models also include options for surface infiltration as a rate-limiting process (but to do so they need rainfall intensity at less-than-daily time scale) and the potential for subsurface of vertical outflow during the rainfall event. Some of the models distinguish ‘soil quick flow’ (water that can infiltrate to soil saturation but can not be stored at field capacity) as an intermediate term between direct runoff and the recharge of the pool that feeds baseflow.

While the textbooks distinguish ‘infiltration limited’ (or Hortonian) overland flow from saturation overland flow (or SOF), in mechanistic models four controls on infiltration of rain into soils can be distinguished:

a. antecedent water use creates a ‘soil water deficit’ below field capacity that can be recharged through rainfall; the faster water is used, the more will be able to infiltrate during the next rainfall event,

b. infiltration potential of the soil surface and its change in soils sensitive to slaking (i.e. fine soil particles can regroup from micro-aggregates to form a ‘sealed’ surface; in some soils algae may further contribute to a ‘hydrophobic’ character of the surface,

c. the time available for infiltration will depend on the slope and opportunities for temporary surface water storage.

d. the difference between ‘field capacity’ and ‘saturation’ (over the whole profile depth) may lead to infiltration at the time of rainfall and either seepage into groundwater or subsurface lateral flow (‘soil quickflow’) during the next 24 hours.

All models include effect a, but only models at level 3 and above (see notes to Table 1)  include effects b, c and d.

Land use change can affect all four of these controls, through

·        difference in water use of vegetation relative to potential evapotranspiration (even though differences are likely to be bigger during a ‘dry season’ due to differences in deciduousness,

·         providing continuous protection of the mineral soil via a litter layer that also stimulates soil biota that increase soil porosity or exposes the soil to sun and rain with opportunity for slaking and sealing,

·        providing more or less temporary water storage opportunities at the soil surface, and thus increasing or decreasing the time available for infiltration,

·        increasing or decreasing macroporosity of the soil, and thus the propensity for ‘soil quick flow’ rather than overland flow.

All models predict a ‘hydrograph’ (daily (or monthly) rate of flow at specific points in the network), and from this the annual water yield and the dry season riverflow can be inferred. Maximum and minimum flows per month or year can always be derived, but the operational definitions used for baseflow and peakflow vary. The ways river networks are represented (routing time, modification of pulse) vary.

Figure 1. Link between patch-level water balance and catchment level hydrological functions; the various models are all based on similar ‘water balance logic’ but differ in the details of the assemblage and filter rules that are used to predict river flows.

Step 4. How do the models separate ‘inherent properties of sites’, ‘effects of land cover’ and ‘impacts of land use change’?

¹Rainfall appears to be adequately known for ‘coarse’ models, but the total input to catchments tends to be underrepresented by non-representative rainfall station locations and high spatial variability of rain; high-resolution models are restricted by ability to generate/obtain spatially explicit rainfall data.

Policy-relevant aspects of land use change on watershed functions:

Design and update: Sandra Velarde 

                                                                                                                                              back to top

Global Partnership Amazon Congo Basin Southeast Asia
Climate Change Biodiversity Sustainability Poverty, Policy and Deforestation
Partners Publications Policy Briefs Gallery Funding Priorities Get Involved
e-mail: asb@cgiar.org   Home Page

Last updated: 28 November, 2003     ©2003 ASB. All rights reserved.