Diagnosis of inconsistencies in multi-year gridded precipitation data over mountainous Diagnosis of inconsistencies in multi-year gridded precipitation data over mountainous areas and related impacts on hydrologic simulations areas and related impacts on hydrologic simulations 1. Introduction 1. Introduction Multi-year gridded Quantitative Precipitation Estimates (QPE) were derived for 1988-2002 for the western basin experiments of the Distributed Model Intercomparison Project (DMIP 2). This data set was subsequently extended to 2006 as more recent data became available. However, we uncovered consistency errors in the extended data set, requiring diagnostic analysis of the underlying gauge data. In this poster we: • Present a method to diagnose consistency errors in high spatial-temporal resolution QPE grids. • Illustrate the importance of temporally consistent QPE for hydrologic modeling. Naoki Mizukami 1,2 , Michael Smith 1 1 NWS/OHD, 2 Len Tech Office of Hydrologic Development, NOAA National Weather Service 1325 East-West Highway, Silver Spring, MD 20910, U.S.A. e-mail [email protected] 2. Background 2. Background AGU 2010 Fall Meeting Dec 11- 17, San Francisco, CA 3. Consistency Check - Gridded Double Mass 3. Consistency Check - Gridded Double Mass Analysis Analysis 4. Diagnosis of QPE Inconsistency 4. Diagnosis of QPE Inconsistency 6. Summary 6. Summary Plot of ‘reference’ cumulative time series R against deviation of cumulative P at target cell from cumulative time series R • DMIP2 Sierra Nevada basins - North Folk of American (NFAR) - East Folk of Carson (EFCR) • Generation of gridded precipitation and air temperature for hydrologic model forcing (Moreda et al. 2006) -Resolution: 4km and hourly -Period: WY1988 - WY 2006 -Spatial interpolation of hourly gauge data 1) Generating historical temporally-consistent QPE grid can be challenging given temporally varying gage networks as well as inconsistent data at each gauge. 2) Developed a method for consistency detection for high spatial and temporal resolution QPE. 3) More consistent QPE leads to more consistent error trends within the simulation, making it easier to improve the simulation with further model calibration H23A-1172 -150 -50 50 150 250 350 450 550 650 750 Oct- 88 Oct- 89 Oct- 90 Oct- 91 Oct- 92 Oct- 93 Oct- 94 Oct- 95 Oct- 96 Oct- 97 Oct- 98 Oct- 99 Oct- 00 Oct- 01 Oct- 02 Oct- 03 Oct- 04 Oct- 05 Date Accum ulated Error,m m O riginal D M IP2 M AP Adjusted D M IP2 M AP Accum ulated error= Σ(Q sim -Q obs ) 5. Hydrologic Simulations with Corrected Data 5. Hydrologic Simulations with Corrected Data Each point corresponds a specific measurement time (e.g. year, month etc.) Time when inconsistency occurs Reference time series: R i n i G G P R k j j m j ij j j i ,..... 1 1 2 1 2 n: the number of time steps ρ j : correlation coefficient between two time series at a target gage and at a neighbor gauge j G ij : time series of a neighbor gauge j, G j : mean of the series at the target gage. P: mean of the time series at the neighbor gauge j. Inconsistency found at groups 4 through 8 in Mar 2003 (upper portion of basin) S S S S S S S S S S S S S S S S S S S S S S S S S S S S S 9 7 6 5 4 3 2 1 82 81 79 77 76 75 73 72 71 70 69 68 67 66 65 64 61 60 58 57 56 54 52 50 48 47 46 45 44 43 42 40 39 38 37 34 33 31 29 28 27 26 25 24 23 22 21 20 19 18 16 15 13 11 10 H igh :4274 Low :-84.8521 N FA R EFCR • Need to diagnose the error trend change for NFAR. • Started investigation of consistency of gridded QPE time series. (consistency: error magnitude and sign is consistent over the period) Target -> time series at pixel groups 1 thru 8 (average P over 4 pixels) Reference -> time series based on 4 reference gages Target -> time series at lower and upper zones (average P over each zone) Reference -> time series based on 4 reference gages DMIP2 Sierra Nevada basins, CA DMIP2 Sierra Nevada basins, CA Streamflow Simulation Results Using Gridded SAC-SMA and SNOW-17 Streamflow Simulation Results Using Gridded SAC-SMA and SNOW-17 Inconsistency found at upper elev. zone The streamflow simulations made with corrected QPE illustrate improvement of the error trend. The simulations below were generated by running the SAC-SMA and SNOW-17 models in two elevation zones for North Folk of American Deviation of Acc. precip. of target cell from Acc. Reference precip. Σ(P i -R i ) 0 Acc. precip. of reference ΣR i Consistent time series Σ(P-R) Σ(P-R) Σ(P-R) ΣR ΣR ΣR History for the individual gauges used to generate the gridded QPE indicates temporally varying gauge network. Discontinuation of the Lake Spaulding (ID-58) caused gridded QPE inconsistency Jan- 89 Jan- 90 Jan- 91 Jan- 92 Jan- 93 Jan- 94 Jan- 95 Jan- 96 Jan- 97 Jan- 98 Jan- 99 Jan- 00 Jan- 01 Jan- 02 Jan- 03 Jan- 04 Jan- 05 Jan- 06 Jan- 07 1 2 3 7 10 11 13 16 19 20 23 27 28 37 38 42 43 44 46 51 52 54 56 57 58 66 67 68 69 73 79 Application to QPE over NFAR and Results Application to QPE over NFAR and Results Overview - Double Mass Analysis (DMA) Overview - Double Mass Analysis (DMA) Gauge numeric ID Lake Sp aulding Square symbols indicate months with more than 80% of hourly or daily data in one month available as valid data for that month. Mar 2003 Mar 2003 Evaluation of average QPE over 4- pixel groups Mar 2003 Mar 2003 Mar 2003 Mar 2003 Mar 2003 -700 -600 -500 -400 -300 -200 -100 0 Oct- 88 Oct- 89 Oct- 90 Oct- 91 Oct- 92 Oct- 93 Oct- 94 Oct- 95 Oct- 96 Oct- 97 Oct- 98 Oct- 99 Oct- 00 Oct- 01 Oct- 02 Oct- 03 Oct- 04 Oct- 05 Date Accum ulated Error,m m EFCR NFAR Accum ulated error= Σ(Q sim -Q obs ) E ast Folk of Carson ast Folk of Carson North F olk of American Error trend changes from under- to overestimation Consistent error trend Simulation with corrected QPE Simulation with original QPE Evaluation of average QPE over elevation zones Inconsistency correction: Use correction factor based on two slopes. (NWS/HL On-line documentation of Interactive Double Mass Analysis (IDMA) User's Guide . [http://www.nws.noaa.gov/oh/hrl/idma/html/dma_home_frame.htm] ΣR Moreda, F., Cong, S., Schaake, J., and Smith, M., 2006. Gridded Rainfall Estimation for Distributed Modeling in Western Mountainous Areas. Poster H23A, Spring Meeting of the AGU, May 23-27, Baltimore, MD. Reference Reference