Digital Ecological Model (DEM) is a platform developed with Java. It consists of six components: DEMGIS, DEMTSA, DEMSTA, DEMMOD, DEMVIEW, and DEMAPPLET. DEMGIS features major functions of geographic information system (GIS), such as building digital elevation model, managing geo-referenced database, translating vector data into raster data, and generating geographic graphs with different projections. DEMTSA is used to interpolate the scattered climatic data into raster data, by means of trend surface analysis (TSA) method and interpolation method. As a plug-in for GIS, DEMSTA provides some widely used statistic methods. DEMMOD is a platform for building process-based landscape model. It provides a visual interface-Visual Programming Interface of Digital Ecological Model (DEMVPI) for ecologists to 'write' and record the models in an interpretation language-Ecological Description Language of Digital Ecological Model (DEMEDL). Ecological Model Interpreter of Digital Ecological Model (DEMEMI) is responsible for compiling the programs written in DEMEDL, running the model and displaying the results. DEMVIEW is a tool for viewing and editing some geographic graphs. DEMAPPLET can link a Java applet with geo-referenced database and display the simulation results on the Internet. All the codes of DEM were compiled into Java application programs, and some of the programs are available on the Internet as Java applets. As a case study, amended Penman's method was used to calculate the potential evapotranspiration and aridity index of China, under present situation and three prescribed climate scenarios, which include raising mean temperature by 1.5, 3.0 and 4.5℃, and raising precipitation by 10%, to assess the potential impacts of global climate change on China water condition.

A model of stomatal conductance was developed to relate plant transpiration rate to photosynthetic active radiation (PAR), vapour pressure deficit and soil water potential. Parameters of the model include sensitivity of osmotic potential of guard cells to photosynthetic active radiation, elastic modulus of guard cell structure, soil-to-leaf conductance and osmotic potential of guard cells at zero PAR. The model was applied to field observations on three functional types that include 11 species in subtropical southern China. Non-linear statistical regression was used to obtain parameters of the model. The result indicated that the model was capable of predicting stomatal conductance of all the 11 species and three functional types under wide ranges of environmental conditions. Major conclusions included that coniferous trees and shrubs were more tolerant for and resistant to soil water stress than broad-leaf trees due to their lower osmotic potential, lignified guard cell walls, and sunken and suspended guard cell structure under subsidiary epidermal cells. Mid-day depression in transpiration and photosynthesis of pines may be explained by decreased stomatal conductance under a large vapour pressure deficit. Stomatal conductance of pine trees was more strongly affected by vapour pressure deficit than that of other species because of their small soil-to-leaf conductance, which is explainable in terms of xylem tracheids in conifer trees. Tracheids transport water by means of small pit-pairs in their side walls, and are much less efficient than the endperforated vessel members in broad-leaf xylem systems. These conclusions remain hypothetical until direct measurements of these parameters are available.

A linear, semi-theoretical relationship between the coverage change of plant communities due to spatial processes and a partial patchiness index of the community distribution patterns in a grassland landscape was established by partitioning the overall coverage change into a spatial increment caused by species migration and a local increment due to local ecological processes. This relationship implies that patchiness of grassland landscapes can accelerate either recovery or degradation of a community, depending on the environmental conditions depicted by a parameter termed as gradient strength. The established relationship also has potential applications in simulating pattern dynamics of plant community distributions for a grassland landscape using a spatially homogeneous patch-scale model. The derived linear relationship was applied to a one-hectare alkaline grassland observatory in northeast China. Gradient strengths of two major plant community types were determined via linear regression from simulation results for selected subregions of the grassland. The calibrated linear relationship was then applied to the rest of the grassland landscape. Preliminary comparisons with complete spatial simulations and observations indicated that using this linear relationship with a patch-scale model can simulate the coverage changes as accurately as using a comprehensive spatial simulation model.

The spatiotemporal variations of vegetation biomass of the ecological transect in northeast China were simulated. State variables of the model included green biomass and nongreen biomass of 12 vegetation categories and water contents of three soil layers. The simulated monthly green biomass was converted into NDVI, or Normalized Differential Vegetation Index of AVHRR (Advanced Very High Resolution Radiometry). A comparison between the modeled and the observed NDVI was made at 10' spatial resolution. Atmospheric CO2 concentration and monthly precipitation were used as two driving variables for global change simulation. Effects of precipitation increments on percentage sunshine, relative humidity, radiation, evapotranspiration, and eventually soil water and plant growth, were considered. Two levels of CO2 concentration (present, doubled) and seven levels of precipitation increments (0, 0.05, 0.1, 0.15, 0.2, 0.25, and 0.30) were prescribed for a total of 14 simulation runs. A steady-state solution was obtained for each simulation run. The results of simulation showed that with the present climate conditions, doubling atmospheric CO2 concentration led approximately to a 20.3% increase in green biomass, 11.0% increase in nongreen biomass, 19.0% increase in green NPP, 12.8% increase in nongreen NPP, and 14.9% increase in overall average NPP at steady state. These increases go, respectively, to 32.9, 13.9, 30.0, 20.1, and 23.4% when a 30% precipitation increase was superimposed on the doubled CO2 concentration.