Evaluation of Satellite Precipitation Measurements Considering Physical Conditions of Atmosphere

Transcription

1 平成 28 年度 JAXA/EORC Tokyo.JP Evaluation of Satellite Precipitation Measurements Considering Physical Conditions of Atmosphere 1 Hyungjun Kim and 1 Nobuyuki Utsumi 1 Institute of Industrial Science, the University of Tokyo, Tokyo, Japan Image courtesy of

2 全球水文シミュレーションと申請者の既往研究 大気再解析データ 河道網 河川形状 幅 深さ 氾濫原 河川パラメーター + バイアス 補正 地表面モデル 大気外力 アンサンブル降水データ 流出 観測データ 陸面パラメーター 葉面積 地表面被覆 土地利用 土壌タイプ 河川モデル

3 Uncertainty in Observational Precipitation Ω : Index to measure the similarity of individual time series of ensemble members [Koster et al., 2000] Ω P =0.77 Ensemble members : m 0 n 0 n 0 mn Ω ET =0.89 Evapotranspiration Ω R =0.66 Runoff 1 = Var[ m 2 2 m mσ aˆ σ a 2 σ + ΩNon-linear a = impacts aˆ a ij ] 2 in land surface simulations ( m 1) σ a i= 1 + Uncertainty in precipitation has heterogeneous global distribution σ 2 a = Var[ a ij ] * Kim 2010

4 Domain: 30 N 38 N; 60 E 149 E Precipitation Composite (DJF) Intercomparison for Extratropical Cyclone CL04 TOTAL Class Central Pressure Gradient Product Type S. Res. T. Res. Period CL [hpa/1000km] 1 APHRODITE Gauge 0.25 Daily CL GPCP 1DD Hybrid 1.0 Daily CL GSMaP V Satellite 0.1 Hourly CL JRA25 Reanalysis T106 Monthly 1979 APHRODITE GPCP1DD GSMAP V5 JRA25

5 Weather Systems Spatiotemporal Resolution Weather System Mask - Tropical cyclone - Extratropical cyclone center - Front (temperature) - Others 1.25 degree, 6-hourly (can be downscaled) Timespan (possible extension to 1958) Input dataset - JRA-55 (SLP, U, V, T, Q) - IBTrACS (TC best track) Data format Supplementary Tool TC Binary file * Python module for handling (in Beta) ExC center Front * Utsumi et al., under review

6 Error Estimation by Weather Systems: Sensitivity Tropical cyclone ExTropical cyclone (center) Front Others Radar -AMeDAS + Different algorithms show biases of different amounts and directions.

7 Error Estimation by Weather Systems: Sensitivity (GSMaPv5 TRMM/PR 2A25v7; ) difference [mm/hour] TC All ExC Others difference [mm/hour] Different biases for Different weather systems Different seasons Different regions * Utsumi et al., in preparation; Azariah et al., in preparation

8 Decomposition of River Discharge by Precipitation Induced by Different Weather Systems Tagged precipitation Products TC ExTrop Cyclone Front Others Weather system Classification Method + Forcing data Land surface model with tracer module Hanasaki et al., Tracer module Utsumi et al., 2016 Tagged hydrological variables (e.g. River discharge) Multi-scale analysis for the propagation of extreme signal from climate to local scale + Climate regime shift and long-term water security problem

9 Cloud Type Mask High cloud Mid cloud Low cloud * Satellite Cloud Grid Information over Western North Pacific by JMA * Visible and Infrared (IR) from Himarwari-6 (0.2 x 0.25 deg., hourly)

10 Probability Density of Precipitation by Cloud Types High Mid Probability density Cumulus (Low) Cumulomimbus (mm/hour) * Lower cloud height more serious underestimation by IR retrieval

11 Validation of Sensitivity by Sensors /* Cumulonimbus */ KuPR GMI

12 Concluding Remarks + Uncertainty in forcing data is one of important uncertainty sources of hydrologic simulations. + Multiple precipitation products including satellite measurement show systematic bias by different causal weather systems. + Weather system mask is not only useful to classify measurement error and update algorithms but also to trace reasons of water excess (flood) or deficit (drought) in different spatiotemporal scale. + Satellite retrieval algorithms show different sensitivities to various cloud types. + Algorithms using IR tend to underestimate week precipitation, but its impact is not considerable in IMERG product. + Overall, current satellite precipitation retrievals mostly underestimate precipitation comparing to Radar-AMeDAS.

13 GPyM: a Python Module to Interface TRMM/GPM Data I/O Interface Features + Archive data from G-Portal (SFTP protocol) + Search granules by timespan and spatial domain + Convert and upscale granules to maps + Cached IO (e.g., orbits)

14 平成 28 年度 JAXA/EORC Tokyo.JP Thank you Image courtesy of