Anusplin is a common method that considering terrain effect during the temperature interpolation and it also has a high accuracy for temperature interpolation in the alpine area. But it is not clear whether there will be different accuracy based on the temperature data at different time scales by this method, especially for mountain temperature interpolation, which is very important for mountain science research. In order to determine the accuracy of temperature interpolation in mountain area at different time scales, we took the most complex mountain area(Hengduan Mountains, a mountain range in Sichuan and Yunnan) as study area. Based on the temperature data recorded by 51 meteorological stations during 1960-2014 and SRTM in the study area, Anusplin was used to interpolate regional temperature at different time scales (decadal, annual, seasonal, monthly, and daily). With the cross-validation and comparison method, the interpolation accuracy of temperature at each time scale was compared by mean absolute error (MAE), mean relative error (MRE) and root mean square error (RMSE). The results showed that:(1) The interpolation of summer temperature shows the minimum value in MAE, MRE and RMSE(0.41℃, 0.03℃ and 0.66℃), while the interpolation of winter temperature shows maximum value in MAE, MRE and RMSE (0.79℃, 0.81℃ and 1.02℃). The MAE value and RMSE value of decadal, annual and autumn temperature interpolation are similar. (2) The temperature interpolation by using Anusplin can well express the climate trends at decadal, annual and seasonal scale. The trend obtained from interpolation data are consistent with the tendency calculated from observed data, and their change rate shows no discrepancy[error < 0.1℃·(10a)-1]. (3) The MAE, MRE and RMSE value for daily and monthly temperature interpolations in summer are also minimum while in winter are maximum. In addition, the MAE, MRE and RMSE value of daily temperature interpolation are higher than the monthly temperature interpolation. Especially for the mean relative error in winter, the value of daily temperature interpolation error is about twice than that of monthly temperature interpolation error. It is worth noting that the inconsistency of temperature interpolation accuracy at decadal, annual and seasonal scale is mainly attributed to the difference in the relationship between temperature and elevation at different time scale. Moreover, the accuracy error of elevation data and the regional special climatic condition can also cause error for temperature interpolation by using Anusplin.
[1]Cui P, Jia Y, 2015.Mountain hazards in the Tibetan Plateau:research status and prospects[J].National Science Review, 2(4):397-399.
[2]Hock R, 2003.Temperature index melt modelling in mountain areas[J].J hydrol, 282(1):104-115.
[3]Hijmans R J, Cameron S E, Parra J L, et al, 2005.Very high resolution interpolated climate surfaces for global land areas[J].Int J climatol, 25(15):1965-1978.
[4]Hutchinson M F, 1991.The application of thin-plate smoothing splines to continent-wide data simulation[J].BMRC Res Rep Ser, 28.
[5]Hutchinson M F, Xu T, 2013.Anusplin version 4.4 user guide[M].Canberra:The Australian National University.
[6]Kohavi R, 1995. A study of cross-validation and bootstrap for accuracy estimation and model selection[C]//Ijcai, 14(2): 1137-1145.
[7]McKenney D W, Pedlar J H, Papadopol P, et al, 2006.The development of 1901-2000 historical monthly climate models for Canadaand the United States[J].Agricultural and Forest Meteorology, 138(1):69-81.
[8]Régnière J, 1996.Generalized approach to landscape-wide seasonal forecasting with temperature-driven simulation models[J].Environmental Entomology, 25(5):869-881.
[9]Thomas A, 2002.Seasonal and spatial variation of evapotranspiration in the mountains of Southwest China[J].J Mount Res, 20(4):385-393.
[10]Guo J P, Liu H, An L C, et al.2016.Study on variation of snow cover and its orographic impact over Qinghai-Xizang Plateau during 20012012[J].Plateau Meteor, 35(1):24-33.DOI:10.7522/j.issn.1000-0534.2014.00140.<br/>郭建平, 刘欢, 安林昌, 等, 2016.2001-2012年青藏高原积雪覆盖率变化及地形影响[J].高原气象, 35(1):24-33.
[11]Gao S Y, Zhang J, Luo Q, 2017.Estimation of the heat transfer coefficient over inhomogeneous underlying surface on the Qinghai-Tibetan Plateau[J].Plateau Meteor, 36(3):596-609.DOI:10.7522/j.issn.1000-0534.2016.00060.<br/>高世仰, 张杰, 罗琦, 2017.青藏高原非均匀下垫面热力输送系数的估算[J].高原气象, 36(3):596-609.
[12]Li Y C, He Z M, Liu C X, 2014.Review on spatial interpolation methods of temperature data from meteorological stations[J].Progress in Geography, 3(8):1019-1028.<br/>李月臣, 何志明, 刘春霞, 2014.基于站点观测数据的气温空间化方法评述[J].地理科学进展, 3(8):1019-1028.
[13]Liu Z H, Yang Q K, Li R, et al, 2008.Interpolation for time series of meteorological variables using ANUSPLIN[J].Journal of Northwest A&F University(Nat.Sci.Ed.), 36(10):227-234.<br/>刘志红, 杨勤科, 李锐, 等, 2008.基于ANUSPLIN的时间序列气象要素空间插值[J].西北农林科技大学学報(自然科学版), 36(10):227-234.
[14]Liu Z H, Yang Q K, Li R, et al, 2006.Modeling spatial distribution of pan evaporation based on quinat-variate thin plate spline function[J].Science of Soil and Water Conservation, 4(6):23-30.<br/>刘志红, 杨勤科, 李锐, 等, 2006.基于5变量局部薄盘光滑样条函数的蒸发空间插值[J].中国水土保持科学, 4(6):23-30.
[15]Liu Z J, Yu X X, Wang S S, et al, 2012.Comparative analysis of three covariates methods in Thin-plate smoothing splines for interpolating precipitation[J].Progress in Geography, 31(1):56-62.<br/>刘正佳, 于兴修, 王丝丝, 等, 2012.薄盘光滑样条插值中三种协变量方法的降水量插值精度比较[J].地理科学进展, 31(1):56-62.
[16]Qian Y L, Lü H Q, Zhang Y H, 2010.Application and assessment of spatial interpolation method on daily meteorological elements based on ANUSPLIN software[J].J Meteor Environ, 26(2):7-15.<br/>钱永兰, 吕厚荃, 张艳红, 2010.基于ANUSPLIN软件的逐日气象要素插值方法应用与评估[J].气象与环境学报, 26(2):7-15.
[17]Shen Y, Xiong A Y, Shi X H, et al, 2008.Development of the grid-based ground water vapor pressure over China in recent 55 years and its accuracy evaluation[J].Acta Meteor Sinica, 66(2):283-291.<br/>沈艳, 熊安元, 施晓晖, 等, 2008.中国55年来地面水汽压网格数据集的建立及精度评价[J].气象学报, 66(2):283-291.
[18]Tan J B, Li A N, Lei G B, 2016.Contrast on ANUSPLIN and Cokriging meteorological spatial interpolation in southeastern margin of Qinghai-Xizang Plateau[J].Plateau Meteor, 35(4):875-886.DOI:10.7522/j.issn.1000-0534.2015.00037.<br/>谭剑波, 李爱农, 雷光斌, 2016.青藏高原东南缘气象要素Anusplin和Cokriging空间插值对比分析[J].高原气象, 35(4):875-886.
[19]Zhang H L, Deng Z W, 2002.A method of spatial simulating of temperature based digital elevation model(DEM) in mountain area[J].J Mount Sci, 20(3):360-364.<br/>张洪亮, 邓自旺, 2002.基于DEM的山区气温空间模拟方法[J].山地学报, 20(3):360-364.
[20]Zhang Y P, Zhang S H, He Y L, 2004.Distribution of climatic elements in the upper reaches of Minjiang river[J].J Mount Sci, 22(2):179-183.<br/>张一平, 张昭辉, 何云玲, 2004.岷江上游气候立体分布特征[J].山地学报, 22(2):179-183.
[21]Zhang C C, Li D L, Wang H, et al, 2017.Characteristics of the surface sensible heat on the Qinghai-Xizang Plateau in the spring and its influences on the summertime rain fall pattern over the eastern China[J].Plateau Meteor, 36(1):13-23.DOI:10.7522/j.issn.1000-0534.2016.00028.<br/>张长灿, 李栋梁, 王慧, 等, 2017.青藏高原春季地表感热特征及其对中国东部夏季雨型的影响[J].高原气象, 36(1):13-23.
[22]Zhou P P, Zhang M J, Wang S J, et al, 2017.Variation and its influences of 0℃ isotherm height in summer over high Asia[J].Plateau Meteor, 36(2):371-383.DOI:10.7522/j.issn.1000-0534.2016.00048.<br/>周盼盼, 张明军, 王圣杰, 等, 2017.高亚洲地区夏季0℃层高度变化及其影响特征研究[J].高原气象, 36(2):371-383.
