Abstract: The free-scale climate data plays an important role as basic data in projecting future forest growth and distribution, and analyzing the responses and adaptation of forest ecosystems on climate change. In this paper, by using multi-source long term static scale climate data, the free-scale climate data model in China was established in terms of a new downscaling algorithm by combining the bilinear distance-weighted interpolation and elevation adjustment; the baseline climate data was validated by historical climate data from 302 in the nationwide. The results indicated that the correlations between geographic variables and temperature variables were high, fitting effects of elevation adjustment functions were all significant(R² greater than 0.91). Through downscaling, the baseline data not only improved the spatial resolution, but also the data accuracy. Since the relationships between precipitation variables and geographic variables were not strong enough to develop elevation adjustment functions, they were only adjusted by bilinear distance-weigthed interpolation, but the improvements still remained. The algorithm overlaied the historical and projected climate anomalies to free-scale baseline data ensured the data integrity in the missing values regions and maintained the accuracy. However, attentions should be paid to the fact that, given the randomness and volatility of the climate environment as well as the influnces of micro-climate and vegatation, the estimated accuracy of average value in a long term would be much more reliable.