Fu's research provides background for understanding the integrated effects of all Metro-Agro-Plex (MAP) units at the regional and global scale. In assessing potential impacts of global warming in China, uncertainties regarding how the climate will change regionally make impact assessments even more uncertain. Two approaches are to use General Circulation Model (GCM) outputs to drive statistical models and to look at historical analogs from long term records to see if there is a relationship between temperature change and other environmental elements.
As for the first approach, Fu has little confidence in using present GCM outputs to make regional assessments because the regional accuracy of GCMs is generally believed to be quite poor. GCM simulations indicate that under doubled carbon dioxide (CO2) concentrations, China will become 1 to 1.5°C warmer than at present with slightly higher precipitation and somewhat lower soil moisture. Model results also indicate that it will be drier in summer, with little change in precipitation in winter. Surface runoff is predicted to be less than at present, related to the drier conditions. The cropping system model of Zhao and Wang (1994) predicts that in the year 2050, under model-predicted climate conditions, there will be a 23.1% reduction in the amount of land in China that is single cropped and a 16.2% increase in the amount of land doubled cropped.
Using the second approach of examining historic records, we can look for relationships between changes in temperature and environmental conditions. The aridity (dryness) in China has had a strong influence on agriculture. Looking at the aridity index for eastern China from 100 stations, there has been an increase in aridity over the past century, with a significant jump in aridity in the 1920s that seems to be closely correlated with the temperature record. Statistics show that the rise in the aridity and temperature records were simultaneous in the 1920s.
Other evidence supporting the hypothesis that China is becoming drier includes recent data on rainfall in eastern China which indicates that rainfall is declining. A third corroboration comes from data on river discharge, which also indicates a decline. Data also indicate that the amount and length of the plum rains have decreased, by 22% and 37% respectively from the 1900s averages to the 1980s averages. Lake levels are falling too, and the amount of desertified land is increasing by 1,560 square kilometers per year.
Drier conditions in China is only part of the larger pattern which seems to follow the global warming scenario. Other parts of this pattern include the fact that there is increased rainfall in the tropics and less rainfall at mid-latitudes and in the subtropics, indicating an intensification of the Hadley circulation. Regarding the transition zone between the arid and moist areas, paleoclimatic data indicates that this zone shifts southward during cooling periods and northward during warming periods. Such climatic changes related to the pattern of the monsoon can be used as historical analogs for climate warming.
Uncertainties for making assessments of the impacts of climate change are quite large. GCM outputs used for regional predictions are presently at a low level of confidence. For example, analysis of several GCMs simulated climatology of precipitation in China reveals similar results, matching the curve of precipitation decrease from the east coast to inland, and the general pattern of becoming drier in winter and wetter in summer. But comparing observations with GCM simulations of precipitation patterns, the spatial distribution of the rainfall pattern is quite different.
GCMs can not even simulate regional climatology, so how can we use their output to deal with regional agricultural questions? There is a need to improve cloud physics, boundary conditions and other elements of the models before we can begin to trust their regional predictions. GCMs are ineffective at simulating regional climate for a number of other reasons. Their resolution is too coarse and they cannot account for ecosystem effects well (that is why their spatial results are so poor). In response, high resolution regional models are being developed and may have more accurate results because they include mesoscale topography and land cover patterns, and vegetation/atmosphere interactions, which are particularly important in monsoon areas in which evapotranspiration and biogeochemical cycles are key.
To improve the regional climate system models, two major advances over GCMs in terms of regional predictive ability have been made recently. One is improved boundary parameterizations through the addition of a near surface layer so the flux between the boundary layer and the soil is transferred to the canopy and the surface layer. This offers better a simulation of the transfer of heat, moisture, etc. Secondly, links have been made between the physical and biological elements of the model, enabling vegetation to be linked directly to surface physical parameters.
Results of this regional model reveal that it captured not only the rainfall belts, as did the standard MM4, but also that it has the capacity to simulate the regional patterns of climate, which will be used as the inputs of impact models. In terms of temperature, the regional model performed better than the GCM regionally. Shifts of rainbelts due to advanced summer monsoons has not been captured in most GCMs, but this general pattern has been simulated by the regional model. Simulating changes in rainfall patterns is the first step toward predicting harvests, as rainfall remains a dominant factor in terms of impact on harvest.
The regional model can also be used to attempt to predict the impact of regional SO2 and SO4 distributions. Will these effects change the regional and global simulations under doubled CO2? Land use patterns can also be factored into the model to simulate regional forcing impacts of changing land cover types. Issues of Metro-Agro-Plexes may potentially be explored using these models.