DICast : A Dynamic Integrated ForeCast System NCAR/RAL 10/10/2014

Transcription

1 DICast : A Dynamic Integrated ForeCast System NCAR/RAL 10/10/2014

2 The DICast System n An automated point weather forecast system n Provides timely, tuned, worldwide forecasts n Designed to emulate the human forecast process n Applicable to a variety of forecast problems n Uses state-of-the-art scientific and engineering principles n Requires only modest computing systems and common data sources n Custom data sources can add skill

3 Applications of DICast n Lay forecasts for the public n Road Weather Forecasts n Agricultural Soil Forecasts n Wind Turbine Forecasts n Solar Power Forecasts

4 DICast Output and Operations n Can produce a variety of tuned forecast variables Daily Max/Min Temp Cloudiness Probability of Precip Probability of Thunder Precip Amount and Type Probability of Fog Temp & Dew point Visibility Wind u-, v-, speed More

5 DICast Output and Operations n Can be set up in different temporal configurations Extent Resolution Update Freq Long Term 0-16 days 3 or 6 hours 3 hour Short Term 0-4 days 1 or 3 hours 1 hour Near Term 0-24 hrs 1 hour 1 hour

6 DICast Output and Operations n Can produce tuned (with observations) or interpolated forecasts for thousands of locations ;72565;KDEN;3983;-10466;1655;4;DENVER, DENVER INTERNATIONAL AIRPORT;CO;UNITED STATES ;72494;KSFO;3762;-12236;3;4;SAN FRANCISCO, SAN FRANCISCO INTERNATIONAL AIRPORT;CA;UNITED STATES ;72793;KSEA;4744;-12231;130;4;SEATTLE, SEATTLE-TACOMA INTERNATIONAL AIRPORT;WA;UNITED STATES ;72503;KLGA;4078;-7388;6;4;NEW YORK, LA GUARDIA AIRPORT;NY;UNITED STATES ;72530;KORD;4198;-8792;203;4;CHICAGO, CHICAGO-O'HARE INTERNATIONAL AIRPORT;IL;UNITED STATES ;72202;KMIA;2579;-8032;3;4;MIAMI, MIAMI INTERNATIONAL AIRPORT;FL;UNITED STATES ;72658;KMSP;4488;-9323;256;4;MINNEAPOLIS, MINNEAPOLIS-ST PAUL INTERNATIONAL AIRPORT;MN;UNITED STATES ;03772;EGLL;5148;-045;24;6;LONDON / HEATHROW AIRPORT;XX;UNITED KINGDOM ;07149;LFPO;4873;240;89;6;PARIS-ORLY;XX;FRANCE ;47662;RJTD;3568;13977;5;2;TOKYO;XX;JAPAN ;94767;YSSY;-3395;15118;6;5;SYDNEY AIRPORT;XX;AUSTRALIA

7 Basic DICast System Diagram Data Ingest Forecast Module A Forecast Module B Forecast Module C Forecast Module D Forecast Module N Integrator Post Processing Forecast Products

8 Dynamic MOS Data Ingest Forecast Module A Forecast Module B Forecast Module C Forecast Module D Forecast Module N Integrator Post Processing Forecast Products

9 Dynamic MOS Linear regression-based statistical method Similar to NWS MOS, but regressions built dynamically Can be applied to any NWP forecast model fairly easily Uses default equations if statistical model fails 80 Good Regression 80 Bad Regression Max Temp Max Temp Thickness Thickness

10 DMOS Default Equations Default equations are substituted whenever the statistical methods fail to produce a suitable result Default equations are combinations of one or more of the regressors Several regressors were designed specifically as defaults Example: Surface Temp: Model s Vertical Temperature Profile Model s Station Elevation Actual Station Elevation DICAST Estimated Surface Temperature

11 Regression Extrapolation Max Temp Regression MaxT = * CAPE CAPE CAPE Range Application Max Temp CAPE CAPE = 2500 J MaxT = *2500 = 360 F Applied CAPE: 2500

12 Forecast Integrator Data Ingest Forecast Module A Forecast Module B Forecast Module C Forecast Module D Forecast Module N Integrator Post Processing Forecast Products

13 Forecast Integrator Objectives To combine forecasts from a set of models: Discovers the best combination of forecast modules for a given forecast time and location Computationally simple and robust Can easily adapt to the addition of new modules or removal of obsolete modules

14 DICast Forecast Integrator Integrated forecasts (F) are bias-corrected, confidence-weighted sums of the module inputs (f i ): F = ( Σ c i w i f i ) / ( Σ c i w i ) + Bias Confidences (c i ) are determined by the forecast modules themselves Weights (w i ) are adjusted daily in the direction of steepest descent of the error (difference between verification, V, and the forecast) in weight space Δw i = S * ( / w i ) {(V - F) 2 }

15 Forecast Integrator Forecast error as function of W 1 & W 2 1 W 2 W 2 (i) Integration Step 0 0 W 1 W 1 (i) 1

16 Post Processing Data Ingest Forecast Module A Forecast Module B Forecast Module C Forecast Module D Forecast Module N Integrator Post Processing Forecast Products

17 Post Processing Quality Control Range Checks Minimal Inter-variable comparisons Temporal Interpolation Spatial Interpolation Variable Derivation Forward Error Correction Deg F Raw Corrected Lead Time (hours)

18 Forecast Products Data Ingest Forecast Module A Forecast Module B Forecast Module C Forecast Module D Forecast Module N Integrator Post Processing Forecast Products

19 Forecast products n Output Data Formats: u netcdf u ASCII - CSV n Data can plug into other systems n Decision Support

20 DICast Advantages DICast forecasts: Outperform every constituent forecast module Outperform human beyond 12 hours Based on sponsor feedback Are totally automated Are more cost effective than human generated forecasts DICast is scalable Additional sites, NWP models or new forecast variables can be easily integrated

21 Short Range Predictions " (0-96 hours)" Temperature

22 Short Range Predictions " (0-96 hours)" Dew Point Temperature

23 Short Range Predictions " (0-96 hours)" Wind Speed

24 Medium Range Predictions " (0-10 days)" Temperature

25 Medium Range Predictions " (0-10 days)" Dew Point Temperature