Abstract: As the climate field reconstruction(CFR)data have been used for important dynamical insights and its utility and significance is widely recognized, more research on CFR methodologies is therefore urgently needed. In this paper, a new technique for climate field reconstruction was developed, i.e. Mode Approached Empirical Orthogonal Function(MA-EOF)method. This method allowed appropriate modes being preserved at different iteration steps, which utilized the intrinsic connections of EOFs, so that the reconstructed data over earlier times was approaching to the "true state" step by step.We collected the tree-ring data mainly from the International Tree-Ring Data Bank(ITRDB)and PAGES 2k network, and screened them by the correlation coefficients with the observed annual temperature in their respective neighboring grids(exceeding the threshold values at 95% significance level). Finally, 549 temperature sensitive tree-ring chronologies were screened out, including 269 records from ITRDB, 267 from PAGES 2k network, 7 from China Meteorological Science Data Sharing Service Network and 6 from some published literatures related on regional temperature signal. The screened chronologies were divided into 17 groups according to their covered period and the starting year of the groups were set every 50 years from 1800A.D. to 1000A.D.The validation results verified the reliability of the reconstructions based on 1~12 group, and demonstrated the efficiency of the MA-EOF method particularly in reconstructing decadal variability of the instrumental data. Therefore, the gridded reconstructions of the first twelve groups were averaged as the final gridded reconstruction, i.e. we reconstructed the spatial pattern of the extratropical Northern Hemisphere(22.5°~90°N) temperature during the period 1250~1850A.D.based on the MA-EOF method.In order to highlight the performance of different methods using for large-scale temperature spatial pattern reconstruction, we also utilized the existed method(Regularized Expectation Maximization, RegEM)based on the same tree-ring dataset to reconstruct another version of extratropical Northern Hemisphere temperature field. Compared with the RegEM-based reconstruction, the MA-EOF temperature field reconstruction showed that the spatial pattern of the cold anomaly in the Little Ice Age(1400~1700A.D.)was much more evident, being more close to the ensemble mean of the multi-model simulations. Additionally, we selected several typical years of volcanic eruptions with different intensity according to the value of total stratospheric sulfate aerosol injection, and compared the temperature spatial pattern of the MA-EOF reconstruction in these volcanic years. The results showed that obvious spatial differences of cooling were found in strong(i.e. 1452A.D., 1783A.D., and 1815A.D.), moderate(i.e.1600A.D., 1641A.D., and 1809A.D.)and relatively-weak(i.e.1584A.D., 1693A.D., and 1719A.D.)volcanic years. However, they also displayed a general characteristic, i.e. the high latitude region was more sensitive to the volcanic forcing than the middle and low latitudes, which was similar to the findings deduced from model simulations. The results above all demonstrated the efficiency of this new method and indicated a great application potential in climate field reconstruction.