We first discussed our approach for the Harvard Forest's contribution to regional studies at the LTER All-Scientists meeting in 1990. Working out from the heart of New England, and working with the Harvard Forest LTER theme of natural versus human disturbance, we chose the New York/New England region as the target area. This region encompasses a wide range of climatic zones and forest types, as well as significant gradients in pollutant deposition. We developed an approach (Aber et al. 1993) which combined a high resolution GIS with statistical models to summarize important climate drivers and a simple, lumped parameter model to predict water, carbon and nitrogen dynamics across the region. All of the components of this systems are now in place and have been used to derive site-level and region-wide estimates of forest NPP and water yield under current conditions and those predicted for the next century.
For the New England/New York regional GIS, a 30 arc-second (approximately 1 km.) digital elevation model (DEM, Figure 1) was obtained from the USGS, and a 1 km. land use/land cover (LULC) map (Figure 2) was derived from AVHRR satellite data that identifies current vegetation as hardwood, spruce-fir, pine and mixed forest types (Lathrop and Bognar, Rutgers University). In the absence of a successfully-validated soil WHC coverage (Lathrop et al. 1994), we have used a regional mean value of 12 cm. Mean monthly climate values are determined as functions of latitude, longitude and elevation, using a statistical climate model developed for the region in combination with the DEM (e.g. temperature Figure 3, precipitation Figure 4; Ollinger et al. 1995). Existing data on wet deposition and atmospheric concentrations of dry deposition components were used to derive regional patterns in deposition of all major ions (e.g. nitrogen Figure 5, sulfur Figure 6; Ollinger et al. 1993, 1995). In concert with the development of the regional GIS we began to work on a new forest ecosystem model which would summarize accepted physiological controls on water, C and N dynamics in as simple a structure as possible, requiring only those inputs which could be defined within a regional GIS. The result of this effort to date is PnET, a nested series of lumped-parameter models of carbon, nitrogen and water fluxes in temperate and boreal forest ecosystems (Figure 7). The different versions of PnET are modular and build out from simplest to most complex. Algorithms such as photosynthesis which are common to all versions are identical between versions. Increasing complexity occurs by layering additional algorithms, representing additional processes, over the core processes in simpler or included versions.
PnET-Day uses foliar mass, specific leaf weight, foliar N concentration, temperature and radiation flux to predict daily gross and net photosynthesis of whole forest canopies, and has been validated against daily summaries of eddy correlation carbon balance measurements from the Harvard Forest (Aber et al. 1996). PnET-II adds carbon allocation and respiration terms, as well as a full water balance to predict NPP, transpiration and runoff. An empirical soil respiration terms allows prediction of total ecosystem carbon balance under ambient conditions. This version has been validated against annual NPP and monthly water yield data from the Harvard Forest and Hubbard Brook ecosystems and is used to predict the combined effects of climate change and increased atmospheric CO2 on these processes (Aber et al. 1995). An earlier version (Aber and Federer 1992) was also validated against data from 10 additional forest types across North America, and a recent modification has extended the model to predict effects of tropospheric ozone concentrations (Ollinger et al. 1997).
PnET-CN adds compartments for woody biomass and soil organic matter, as well as algorithms for biomass turnover and litter and soil decomposition to allow calculation of complete carbon and nitrogen cycles. This version maintains the predictions for NPP and water balance used for validation in PnET-II, and also compares well with field data in predicting total annual, mean seasonal, and actual time series rates of nitrate loss in streams (Aber et al. 1997a, 1997b). An additional version of the model (PnET-BGC) is under development. This uses multiple element limitations on NPP and element concentrations in all pools to calculate cycling rates for all ../elements. This version has been combined with the soil chemistry model CHESS (Santore and Driscoll, Syracuse University) and used to predict stream and soil chemistry (Postek et al. 1995). Both the PnET-CN and PnET/CHESS versions represent significant collaborative efforts with Dr. Charles Driscoll and other cooperators from the Hubbard Brook LTER site.
For regional productivity and water balances, PnET has been run for each pixel of the 1 km resolution GIS data base (Ollinger et al. 1998). Predicted outputs include annual net ecosystem production (Figure 8), net primary production, wood production and water yield (Figure 9). Regional validation of water yield predictions have been carried out using data summarized from gauged watersheds (Ollinger et al. 1998, Bishop et al. 1997). Initial assessments of climate change effects were made by Aber et al. (1995). Interactions with O3 have been addressed by Ollinger et al. (1996, 1997). Impacts of N deposition have been discussed relative to the ability of forest ecosystems to retain and cycle N under undisturbed (Aber et al. 1997a) and manipulated (Aber et al. 1997b) conditions. PnET has been used in other settings as well. A methodology similar to that described here has been used to develop soil and climate data planes, and to run PnET regionally to predict potential forest productivity under current and double CO2 conditions for Ireland, a country where afforestation is occurring rapidly (Goodale et al. 1997a,b). At the other extreme in spatial coverage, PnET has been used in conjunction with estimates of canopy chemistry obtained by high resolution remote sensing for the Prospect Hill tract at Harvard Forest (Martin and Aber 1996, 1997). Applications to sites in Japan, France and the Czech republic have been carried out and manuscripts are submitted or in preparation.