The upper Cross River basin (UCRB) fits a true description of a data scarce watershed in respect of climatic data. This paper seeks to determine the best-fit probability density function (PDF) of annual maximum rainfall for the UCRB using the Climate Forecast System Reanalysis (CFSR) precipitation data. Also, to evaluate the performance of the Intergovernmental Panel on Climate Change (IPCC) Coupled Model Inter-comparison Project (CMIP3) Fourth Assessment Report (AR4) Global Circulation Models (GCMs) in simulating the monthly precipitation in the UCRB considering 1979–2014 data. For the determination of the best-fit PDF, the models under review included the generalized extreme value (GEV), normal, gamma, Weibull and log-normal (LN) distributions. Twenty-four weather station datasets were obtained and subjected to frequency distribution analysis on per station basis, and subsequently fitted to the respective PDFs. Also, simulated monthly precipitation data obtained from 16 AR4 GCMs, for weather station p6191, were subjected to frequency distribution analysis. The results showed the percentages of best-fit to worst-fit PDFs, considering the total number of stations, as follows: 54.17%, 45.83%, 37.50%, 45.83%, and 50%/50%. These percentages corresponded to GEV, Weibull, gamma, gamma, and LN/normal, respectively. The comparison of the predicted and observed values using the Chi-square goodness-of-fit test revealed that the GEV PDF is the best-fit model for the UCRB. The correlation coefficient values further corroborated the correctness of the test. The PDF of the observed data (weather station p6191) and the simulations of the 16 GCMs computed using monthly rainfall datasets were compared using a mean square error (MSE) dependent skill score. The result from this study suggested that the CGCM3.1 (T47) and MRI-CGCM2.3.2 provide the best representations of precipitation, considering about 36 years trend for station p6191. The results have no influence on how well the models perform in other geographical locations.

中文翻译：

---

利用CFSR降水资料评估年度最大降水的区域最佳拟合概率密度函数

克罗斯河上游流域（UCRB）符合气候资料方面数据稀缺的分水岭的真实描述。本文力求使用气候预测系统再分析（CFSR）降水数据确定UCRB的年度最大降雨量的最佳拟合概率密度函数（PDF）。此外，要评估政府间气候变化专门委员会（IPCC）耦合模型比对项目（CMIP3）第四次评估报告（AR4）全球环流模型（GCM）在模拟UCRB中的月降水量（考虑1979-2014年数据）的效果。为了确定最合适的PDF，所审查的模型包括广义极值（GEV），正态，伽马，威布尔和对数正态（LN）分布。获得了二十四个气象站数据集，并在每个气象站的基础上进行了频率分布分析，然后将其拟合为相应的PDF。此外，还对从气象站p6191的16个AR4 GCM获得的模拟月降水量数据进行了频率分布分析。结果显示，考虑站点总数，最适合与最不适合PDF的百分比如下：54.17％，45.83％，37.50％，45.83％和50％/ 50％。这些百分比分别对应于GEV，Weibull，γ，γ和LN / normal。使用卡方拟合优度检验比较预测值和观察值表明，GEV PDF是UCRB的最佳拟合模型。相关系数值进一步证实了测试的正确性。使用依赖于均方误差（MSE）的技能评分，比较了观测数据的PDF（气象站p6191）和使用月降雨量数据集计算的16个GCM的模拟。这项研究的结果表明，考虑到p6191台站约36年的趋势，CGCM3.1（T47）和MRI-CGCM2.3.2提供了最好的降水量表示。结果对模型在其他地理位置的性能没有影响。