30-Arcsecond monthly climate surfaces with global land coverage
30-Arcsecond monthly climate surfaces with global land coverage
复制标题
DOI：
10.1002/joc.3829
复制
发表时间：
2014-06-01
期刊：
影响因子：
--
通讯作者：
Sharp, Kendra V.
中科院分区：
文献类型：
--
作者：
Mosier, Thomas M.;Hill, David F.;Sharp, Kendra V.
提供了全球所有陆地区域的每月总降水量和平均温度气候表面，网格化为 30 角秒（赤道处大约 1 公里）。这些数据集是通过 Delta 降尺度方法生成的，使用 30 角秒 WorldClim 气候学来缩放每月异常网格。对于月平均气温，异常是根据气候研究单位 (CRU) 和威尔莫特和松浦 (W&M) 0.5 度时间序列数据集构建的，而对于月降水量，还使用全球降水气候中心 (GPCC) 数据。然后将 0.5 度异常插值到 30 角秒分辨率。与双线性和三次样条插值相比，使用分段三次 Hermite 插值多项式 (PCHIP) 对异常网格进行插值会产生更具物理代表性的 Delta 缩小表面。将达美航空缩小规模的产品与分布在全球的六个测试区域的全球历史气候网络 (GHCN) 台站记录进行比较。在此分析中，使用 W&M 时间序列数据集生成的 Delta 网格比使用 GPCC 或 CRU 生成的网格表现更好。使用美国俄勒冈州作为测试区域，将 Delta 缩小数据集与独立坡度模型 (PRISM) 数据集的参数高程回归进行比较。对于月降水量，PRISM 的表现优于三个 Delta 降尺度数据集，但对于平均温度，两个 Delta 降尺度数据集均优于 PRISM。通过计算 WorldClim 气候学中 GHCN 台站描绘误差与 Delta 缩小尺度的 W&M 数据之间的皮尔逊积矩相关系数，结果表明 Delta 网格的性能与参考气候学的性能密切相关。因此，本文中描述的 30 角秒 Delta 网格的未来改进与所有全球陆地表面的高分辨率气候数据的进步密切相关。本文讨论的 Delta 缩小数据集是开源的，并在 http://www.globalclimatedata.org 上免费分发。
Monthly total precipitation and mean temperature climate surfaces, gridded to 30-arcseconds (approximate to 1km at the equator) and available for all global land areas, are presented. These datasets are generated with a Delta downscaling method, using the 30-arcsecond WorldClim climatologies to scale monthly anomaly grids. For monthly mean temperature, the anomalies are constructed from both the Climate Research Unit (CRU) and Willmott & Matsuura (W&M) 0.5 degree time-series datasets, whereas for monthly precipitation Global Precipitation Climatology Centre (GPCC) data are also used. The 0.5 degree anomalies are then interpolated to the 30-arcsecond resolution. Use of piecewise cubic Hermite interpolating polynomials (PCHIP) to interpolate the anomaly grids results in more physically representative Delta downscaled surfaces, compared to bilinear and cubic spline interpolation. The Delta downscaled products are compared to Global Historical Climatology Network (GHCN) station records for six test regions distributed globally. In this analysis, the Delta grids produced using the W&M time-series dataset perform better than grids produced using GPCC or CRU. Using Oregon, USA as a test region, the Delta downscaled datasets are compared to the Parameter-elevation Regressions on Independent Slopes Model (PRISM) datasets. For monthly precipitation, PRISM performs better than each of the three Delta downscaled datasets, but for mean temperature both Delta downscaled datasets outperform PRISM. Through computing the Pearson product-moment correlation coefficient between GHCN station delineated errors in the WorldClim climatologies and the Delta downscaled W&M data, it is shown that performance of the Delta grids corresponds strongly to performance of the reference climatologies. Therefore, future improvement of the 30-arcsecond Delta grids described in this article is strongly tied to advances in the high-resolution climatological data for all global land surfaces.The Delta downscaled datasets discussed herein are open-source and freely distributed at http://www.globalclimatedata.org.