Inversion of land surface temperature in China
Land surface temperature (LST) is important value for applications in the field of agrometeorology, climatology and environment. It is also an important parameter for energy balance model. There are many models to inverse LST using NOAA-AVHRR data, including theoretical model, semi-theoretical and semi-empirical model and statistical model etc.. The statistical model based on split window algorithm is one of practical methods. This model has been applied to inverse LST (Parata, 1993, Becker et al 1994, Ottle, 1992, Price, 1984, Sobrino et al, 1994 and Li, 1993). Greater error may be come into being if the model was used to inverse LST in large area because the split window algorithms were developed at special climate condition. For practice, it is necessary to build a set of data bases of a priori knowledge in order to assure accuracy of the models. The model used in this paper was fitted to similar climate condition and land cover according to climate planning and land cover type. The different models in winter and summer for some land type were also used.

Calculation of monthly roughness length of land surface in China
The exchange intensity of energy, momentum and mass between land surface and atmosphere relates to land cover type, season, vegetation height, vegetation canopy density etc.. So roughness of land surface is an important parameter for determining exchange intensity of energy which describes turbulence exchange intensity between land and atmosphere.

It is different to determine monthly roughness length with conventional climate models. Development and practice of remote sensing, especially advantage in temporal and spatial observation in large area become possible to build database of land surface feature and to calculate roughness length. The theoretical and statistical models for computation of roughness can be used (Monteith, 1973 and Jarvis, 1976). For practical and application, the statistical model was applied to calculate monthly roughness length of land surface type, which supported by a series of databases.

Calculation of evapotranspiration (ET) of land surface in China
Energy balance model is main model for calculating ET (moteith,1973), for example Bartholic model, Bowan ratio-energy balance (BREB), Brown-Rosenberg model (Brown, 1973), measurement technique of vegetation physiology, improved Verma – Rosenberg model (verma, 1976), 3D model of energy balance (Martsolf, 1975), complementary relationship model (Morton, 1983 and Xu, 1999) and Penmen-Monteith model etc.. Considering possibility of data acquisition and practice the different models were selected according to the regional features in China. Evapotranspiration in paddy field and marsh swamp area was calculated by potential evaporation models (Dickinson, 1986). C a 6 o empirical model was used for calculating latent flux of snow surface. The evaporation in lake area, river basin, desertification and desert was calculated with help of complementary relationship model. The Penman- Monteith model was employed to calculate ET in dry land, soil, forest area.

Results and Discussion

Evapotranspiration distribution of land surface in China
Many factors affect ET of land surface, including climate factor, geographical factor, physical characteristics of land cover etc.. The most important factor is the climate element. The actual ET distribution in China was studied with climate region as the basic unit. The monthly ET in different climate region and yearly amount was calculated (table 1). Figure 1 indicated change chard of monthly ET in 9 climate regions. The monthly ET in different climate regions changed greatly, difference between maximum and minimum of monthly ET was about 100 mm in a year, and maximum difference of yearly ET extended more than 1000 mm, the minimum was about tens mm. The discrepancy of climate made their regular distribution and amount of monthly ET.
Table 1 monthly evapotranspiration in different climate regions (mm)



Figure 1. Monthly change of evapotranspiration in 9 climate regions
The temporal and spatial distribution of ET in different climate region in China possess the following features:

• Monthly ET in summer was greater than that in winter for seasonal distribution. ET in east part of China was greater than that in west part and ET in high latitude area was less than that in low latitude area for spatial distribution.
• The discrepancy between Eastern and Western in winter was not great due to dry climate influenced by northwest cold air mass, but ET increased with air temperature going up and enough rainfall after spring. It reached maximum in summer (June – August), the ET in Western was less because the west part located inland or plateau which related effect of southeast and southwest monsoon on it.
• ET in high latitude area in winter was less than that in low latitude area. The discrepancy of ET in summer was not significant.
• Change range of ET in coastal area of China was greater than that in central part of the country due to water budget in winter and summer. The less change of ET in whole year in Western and Northwestern was found.