Predicting Land Use Impacts on Regional Scale Groundwater Recharge and Discharge Ramsis Salama,* Tom Hatton, and Warrick Dawes ABSTRACT The paper presents the development and evaluation of methods that estimate recharge and discharge, water flow, and salt fluxes in rivers. The methods in combination provide an inferential framework to predict dryland salinity and the selection of the appropriate manage- ment scenarios. Hydrogeomorphic Analysis of Regional Spatial Data (HARSD) was used for delineating hydrogeologically homogenous units, and to translate limited hydrological data into hydraulic head surfaces and ultimately a steady state flow net representing recharge- discharge relationships. A complex, physically based water, energy, and carbon model (WAVES) was developed and tested to provide recharge estimates required for the flow net simulations. Remote sensing imagery and analysis techniques involving airborne, advanced very high resolution reflectance (AVHRR) and LANDSAT-Thematic Mapper (TM) data were used to infer the temporal and spatial patterns of leaf area index (LA1) and land-cover type. The techniques were applied in 22 subcatchntents in the Loddon and Campaspe and for the two major catchments. Modeled recharge was consistent with local estimates based on inverse methods and with subcatchment- scale estimates based on stream salt loads under a steady-state assump- tion. The calibrated flow nets for each of the subcatchment have been used to test the sensitivity of this system to changes in recharge resulting from land use change. It is shown, for example, that the Upper Campaspe subcatchment would require the reforestation of key recharge areas totaling 45% of the subcatchment to reduce salt loads from approximately 25000 Mg yr~' down to 18000 Mg yr~' under steady state assumption. I NFERRING the hydrological and hydrogeochemical im- pacts of land use change at the regional scale embod- ies all of the major scientific challenges currently faced by hydrologists: scaling, extrapolation, modeling philos- ophy, and process understanding. For many hydrolo- gists, activities in this area are trans-scientific; that is, an area for which questions can be asked in scientific language, but for which science has no answers. Yet R. Salama and T. Hatton, CSIRO Land and Water, Private Bag, Wembley, Western Australia 6014; and W. Dawes, CSIRO Land and Water, GPO Box 1666, Canberra, ACT 2601. Received 18 Nov. 1997. *Corresponding author (ramsis@per.clw.csiro.au). Published in J. Environ. Qual. 28:446-460 (1999). society faces many pressing and immediate problems that demand advice, and uncertainties that demand augur. In Australia, one such phenomenon is dryland salinization, the result of complex and subtle interac- tions among the atmosphere, the soil, plant, and people. Land managers are demanding advice on the relative and absolute merits of alternate land practices in con- trolling on-farm and off-site impacts, in the full knowl- edge that such advice is inevitably site specific and per- haps scale-dependent. In the absence of experimental evaluation of all alter- natives at all sites, we appeal to modeling. In response, scientists have provided an array of salinity models rang- ing from the loosely empirical and qualitative to the complex and physically based. Hatton et al. (1994) dis- cusses the relative merits of these approaches with re- spect to dryland salinity prediction. Generally, pre- dictive salinity modeling with respect to land use change has been resolved through the application of complex, continuum mechanics-based, spatially distributed, tem- porally dynamic models (e.g., Salama et al., 1993c; Dawes et al., 1997). In common among these models is an ability to predict fluxes of groundwater (and, indi- rectly, salt) as a function of land use if (i) the correct conceptual model for the catchment is available, and (ii) there is sufficient data for the multitude of (often spatially distributed) parameters involved. Unfortunately, there is rarely (if ever) such an abun- dance of local data for land use planning in Australian catchments. At best, there may be a few monitored bores, a regional climate surface, surface elevation data, a broad geological map, and whatever can be seen from above the ground (e.g., land use or vegetation type). In Australia the term bore is commonly used in place of Abbreviations: HARSD, hydrogeomorphic analysis of regional spatial data; AVHRR, advanced very high resolution reflectance; TM, the- matic mapper; LAI, leaf area index; GIS, geographic information systems; HGU, hydrogeomorphological unit; REV, representative el- ementary volume; REA, representative elementary area; UC, Upper Campaspe; UL, Upper Loddon; AHD, Australian height datum; N1R, near infrared. Published March, 1999