欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2017, Vol. 33 ›› Issue (19): 101-109.doi: 10.11924/j.issn.1000-6850.casb17030062

所属专题： 农业气象

• 资源 环境 生态 土壤 气象 • 上一篇    下一篇

海南岛日最低、最高气温空间插值研究

张国峰1,2, ，白蕤1,2，李伟光1,2，田光辉1,2，蔡大鑫1,2，邹海平1,2   

  1. 1 海南省气象科学研究所，海口 570203； 2 海南省南海气象防灾减灾重点实验室，海口 570203
  • 收稿日期:2017-03-10 修回日期:2017-05-29 接受日期:2017-05-19 出版日期:2017-07-11 发布日期:2017-07-11
  • 通讯作者: 张国峰
  • 基金资助:
    海南省自然科学基金面上项目“海南岛日最低、最高气温空间插值研究”(20164173)。

Spatial interpolation of daily minimum temperature and daily maximum temperature in Hainan Island

Zhang Guofeng1,2, , Bai Rui1,2, Li Weiguang1,2, Tian Guanghui1,2, Cai Daxin1,2, Zou Haiping1,2   

  1. 1 Hainan Institute of Meteorological Science, Haikou 570203;   2 Key Laboratory of South China Sea Meteorological Disaster Prevention and Mitigation of Hainan Province, Haikou 570203
  • Received:2017-03-10 Revised:2017-05-29 Accepted:2017-05-19 Online:2017-07-11 Published:2017-07-11

PDF (PC)

341

摘要： 为探索适用于海南岛日最低、最高气温空间插值的方法，研究以经度、纬度、海拔、坡度、坡向、海陆距离、NDVI等为环境变量，采用多元线性回归（multiple linear regression，MLR）、地理加权回归 (geographically weighted regression，GWRK)、多元线性回归克里格 (multiple linear regression-Kiging，MLRK) 和 地理加权回归克里格(geographically weighted regression-Kriging，GWRK)等4种方法对海南岛2016年1月1日—6月30日的日最低、最高气温进行了插值。结果表明：4种方法对日最低气温插值的总的平均绝对误差MLR> GWR>GWRK>MLRK，但GWR、GWRK、MLRK对日最低气温插值的总的平均绝对误差十分接近，对日最高气温有相同的规律。MLRK对日最低、最高气温的总体平均绝对误差分别为0.50℃和0.73℃。GWRK、MLRK对逐日最低气温插值的平均绝对误差也十分接近，对日最高气温也有相同的规律。无论是对日最低气温还是对日最高气温，MLRK、GWRK插值空间分布的主要差异均在站点稀疏的山区。因此，在海南岛，宜采用多元回归克里格（MLRK）对日最低、最高气温进行空间插值。

关键词: 日最低气温, 日最高气温, 地理加权回归克里格, 回归克里格, 海南岛

Abstract: The paper aims to find the suitable spatial interpolation method for daily minimum temperature and daily maximum temperature in Hainan Island, multiple linear regression (MLR), geographically weighted regression (GWRK), multiple linear regression-Kiging (MLRK) and geographically weighted regression-Kriging (GWRK) were compared to interpolate daily minimum temperature and daily maximum temperature in Hainan Island on January 1-June 30, 2016 by using longitude, latitude, elevation, slope, aspect, closest distance to the seaside, NDVI as covariates. The results showed that: the mean absolute errors (MAE) of the four methods for all daily minimum temperature samples were MLR> GWR>GWRK>MLRK; however, the MAEs of GWR, GWRK, and MLRK for all daily minimum temperature samples were very close, the MAEs of GWRK and MLRK for daily minimum temperature were also very close; the daily maximum temperature was the same; the MAE of MLRK for daily minimum temperature and daily maximum temperature was 0.50℃ and 0.73℃, respectively; the main differences of the spatial distributions interpolated by GWRK and MLRK for the daily minimum temperature were in the site sparse mountainous areas; the daily maximum temperature was the same. Therefore, in Hainan Island, the daily minimum temperature and daily maximum temperature should be interpolated by using MLRK.

Key words: daily minimum temperature, daily maximum temperature, geographically weighted regression Kriging, regression Kriging, Hainan Island