当前位置: X-MOL 学术J. Hydrol. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Feasibility of model output statistics (MOS) for improving the quantitative precipitation forecasts of IMD GFS model
Journal of Hydrology ( IF 5.9 ) Pub Date : 2024-11-30 , DOI: 10.1016/j.jhydrol.2024.132454
A. Madhulatha, Ashok Kumar Das, S.C. Bhan, M. Mohapatra, D.S. Pai, D.R. Pattanaik, P. Mukhopadhyay

To improve the quantitative precipitation forecast (QPF) over river-subbasin wise, analysis is conducted using the India Meteorological Department (IMD) Global Forecast System (GFS) model output and high-resolution Flood Monitoring Office (FMO) Bhubaneswar (includes ten catchments (sub-basins) observations. Bilinear interpolation technique is performed to extract the GFS model data to nearest grid point of station observations. Model Output statistics (MOS) approach is utilized by performing the multivariate regression analysis between observed QPF and forecast variables from IMD GFS (QPF and various key parameters (Temperature and Relative Humidity at 2 m (T2m, RH2m), Wind components at 10 m (U10, V10), Convective Available Potential Energy (CAPE), Convective Inhibition (CINE), absolute vorticity (850 hPa), precipitable water, vertical velocity (700 hPa)). To understand the relationship between the QPF and basic meteorological variables, correlation analysis is performed between observed Quantitative Precipitation (averaged over sub-catchments) against forecasted QPF (for Day 1), other key parameters (24 h averaged over sub-catchments) for 2022 summer monsoon season. Among the 11 parameters, rainfall (QPF), precipitable water, absolute vorticity (850 hPa) and vertical velocity (700 hPa) showed good correlation (of above 0.5) with observed Quantitative Precipitation. These parameters with significant correlation are further utilized for multi-variate regression analysis and tested for four different combinations (one, two, three and four parameters). Among all, Root Mean Square Error (RMSE) is improved with regressed rainfall using the combination of four highly correlated parameters. Cathment-wise QPF analysis has showed reduced bias, RMSE in regressed QPF compared to forecast QPF against observations. Temporal evolution of QPF during the monsoon season also showed good consistency between observed, forecasted and regressed QPF. For the training period, skill score analysis of total QPF for entire FMO Bhubaneswar covering the10 sub-basins has showed that Probability of Detection (POD), Critical Success Index (CSI) are improved in Regressed QPF with reduction in False Alarm Rate (FAR) and fewer differences in Miss Rate (MR), Critical Success Index (CSI) and Percentage Correct (PC). Furthermore, Categorial analysis over different QPF categories (0–0.1 mm, 0.1–10 mm, 11–25 mm, 26–50 mm, 51–100 mm, >100 mm) has also showed improvement over QPF categories (0.1–10 mm, 11–25 mm, 26–50 mm) for POD and PC. Independent verification for Monsoon 2023 season also revealed improved skill scores in overall QPF and different categories.

中文翻译:


模型输出统计 (MOS) 改进 IMD GFS 模型定量降水预报的可行性



为了改进河流子流域的定量降水预报 (QPF),使用印度气象局 (IMD) 全球预报系统 (GFS) 模型输出和高分辨率洪水监测办公室 (FMO) 布巴内斯瓦尔(包括 10 个集水区(子流域)观测)进行了分析。执行双线性插值技术以将 GFS 模型数据提取到最近的站点观测网格点。模型输出统计 (MOS) 方法通过在观测到 QPF 和 IMD GFS (QPF 和各种关键参数(2 m 处的温度和相对湿度(T2m、RH2m)、10 m 处的风分量(U10、V10)、对流可用势能 (CAPE)、对流抑制 (CINE)、绝对涡度 (850 hPa)、可降水、 垂直速度 (700 hPa))。为了了解 QPF 与基本气象变量之间的关系,对 2022 年夏季季风季节观测的定量降水(子集水区平均值)与预测的 QPF(第 1 天)、其他关键参数(子集水区平均 24 小时)进行了相关性分析。在 11 个参数中,降雨量 (QPF)、可降水量、绝对涡度 (850 hPa) 和垂直速度 (700 hPa) 与观测到的定量降水具有良好的相关性(高于 0.5)。这些具有显著相关性的参数进一步用于多变量回归分析,并针对四种不同的组合(一、二、三和四个参数)进行测试。其中,使用四个高度相关参数的组合,随着降雨量的回归,均方根误差 (RMSE) 得到改善。 根据观察结果预测 QPF 时,预测 QPF 的偏差 QPF 偏差和 RMSE 降低。季风季节 QPF 的时间演变也表明观测、预测和回归 QPF 之间具有良好的一致性。在培训期间,涵盖 10 个子流域的整个 FMO Bhubaneswar 总 QPF 的技能分数分析表明,回归 QPF 的检测概率 (POD)、关键成功指数 (CSI) 得到改善,误报率 (FAR) 降低,漏失率 (MR)、关键成功指数 (CSI) 和正确百分比 (PC) 的差异较小。此外,对不同 QPF 类别(0-0.1 mm、0.1-10 mm、11-25 mm、26-50 mm、51-100 mm、>100 mm)的分类分析也显示,POD 和 PC 的 QPF 类别(0.1-10 mm、11-25 mm、26-50 mm)也有所改善。2023 年季风赛季的独立验证还显示,整体 QPF 和不同类别的技能得分有所提高。
更新日期:2024-11-30
down
wechat
bug