Development of high spatial resolution rainfall data for Ghana
Development of high spatial resolution rainfall data for Ghana
复制标题
DOI：
10.1002/joc.5238
复制
发表时间：
2018-03-15
期刊：
影响因子：
--
通讯作者：
Yorke, C.
中科院分区：
文献类型：
--
作者：
Aryee, J. N. A.;Amekudzi, L. K.;Yorke, C.
该国经济的各个部门——农业、卫生、能源等——在很大程度上依赖于气候信息，因此获得高质量的气候数据对于这些部门的气候影响研究至关重要。本文利用分布在加纳4个农业生态区域的113个加纳气象局（GMet）测量网，建立了一个空间分辨率为0.5度x0.5度的月降雨量数据库（GMet v1.0），时间跨度为23年（1990-2012）。首先通过分位数匹配调整对数据集进行均匀化，然后使用带张力参数的最小表面曲率以0.5度x0.5度的空间分辨率对数据集进行网格化，以便对开发的数据集进行全面的空间场评估。之后，使用GMet v1.0对先前由于大数据集而被排除在外的站点的测量数据进行点像素验证。这证明了GMet v1.0的可靠性，在99%的置信度水平上具有高且统计显著的相关性，并且相对较低的偏差和rmse。此外，将GMet v1.0与GPCC和TRMM的降雨量估计进行了比较，发现这两个产品都充分模仿了GMet v1.0，在99%的置信度下具有高相关性，偏差和rmse都很低。此外，与GMet v1.0相比，第90百分位的比值提供了TRMM和GPCC相当相似的极值捕获。最后，基于年降雨量和月变化，对GMet v1.0进行k均值聚类分析，将该国划分为四个不同的气候带。开发的降雨数据一旦正式发布，将成为加纳气候影响和进一步降雨验证研究的有用产品。
Various sectors of the country's economy-agriculture, health, energy, among others-largely depend on climate information, hence availability of quality climate data is very essential for climate-impact studies in these sectors. In this paper, a monthly rainfall database (GMet v1.0) has been developed at a 0.5 degrees x0.5 degrees spatial resolution, from 113 Ghana Meteorological Agency (GMet) gauge network distributed across the four agro-ecological zones of Ghana, and spanning a 23-year period (1990-2012). The datasets were first homogenized with quantile-matching adjustments and thereafter, gridded at a spatial resolution of 0.5 degrees x0.5 degrees using Minimum Surface Curvature with tensioning parameter, allowing for comprehensive spatial fields assessment on the developed dataset. Afterwards, point-pixel validation was performed using GMet v1.0 against gauge data from stations that were earlier excluded due to large datagaps. This proved the reliability of GMet v1.0, with high and statistically significant correlations at 99% confidence level, and relatively low biases and rmse. Furthermore, GMet v1.0 was compared with GPCC and TRMM rainfall estimates, with both products found to adequately mimick GMet v1.0, with high correlations which are significant at 99% confidence level, low biases and rmse. In addition, the ratio of 90th-percentile provided fairly similar capture of extremes by both TRMM and GPCC, in relation to GMet v1.0. Finally, based on annual rainfall totals and monthly variability, k-means cluster analysis was performed on GMet v1.0, which delineated the country into four distinct climatic zones. The developed rainfall data, when officially released, will be a useful product for climate impact and further rainfall validation studies in Ghana.