Compared to inland areas, meteorological stations in the island regions appear scarce and unevenly distributed, which leads to noteworthy uncertainty in detailed characterisation of various meteorological elements. For the Zhoushan Islands, located in Southeast China, there exist many islands and islets, and the local terrain is quite complex. Therefore, different interpolation strategies usually generate diverse gridded results, which largely influence the reliability and accuracy of operational climate monitoring and diagnosing. Under the background of climate change, the Zhoushan region is frequently invaded by cold waves in recent years, so how to scientifically choose an interpolation scheme to reasonably represent spatial distribution characteristics of temperature becomes an important issue in local climate operations. To solve this problem, based on the index of root mean square error (RMSE), the interpolation effect of Ordinary Kriging (OK), Inverse Distance Weighting (IDW), and ANUSPLIN (ANU) are comparatively analysed for 8 cold wave processes influencing Zhoushan during 2014-2021. Two subdivided indices, i.e., temporal RMSE (TRMSE) and spatial RMSE (SRMSE) are further designed to evaluate the interpolation results on temporal and spatial dimensions respectively. Eleven stations are randomly selected from the total 53 meteorological observational stations to test the interpolation results of OK, IDW, and ANU for the minimum temperature, reduction of daily minimum temperature and daily-mean temperature in the 8 processes. It can be found that the bias in the ANU case is higher than that in the OK and IDW cases. To explain such a phenomenon, 3 interpolation experiments with dense surrounding stations, sparse surrounding stations, and specific distribution of examining stations (all the examining stations are not distributed in the main island of Zhoushan) are further designed. The results demonstrate that the performance of the ANU strategy is closely linked to the spread situation of peripheral stations. When the surrounding stations are concentrated, the interpolation bias of ANU is usually smaller than that of OK and IDW. However, if the surrounding stations appear sparse, the bias of ANU exhibits much larger. In the scenario of dense peripheral stations, regardless of the examining sites distributed over the main island or not, the ANU solution can always get the optimal interpolation results, which implies that the impact of topography on the performance of ANU in temperature interpolation is of less importance. Also, the influence of horizontal resolution for interpolation seems secondary. When the horizontal resolution for three interpolation schemes falls down to 1 km×1 km from 30 m×30 m, the change of RMSE is generally less than 0.1 ℃ for most circumstances, so the impact of interpolation resolution can be neglected.