Download presentation
Presentation is loading. Please wait.
Published byDelilah French Modified over 9 years ago
1
Great Basin Verification Task 2008 Increased Variability Review of 2008 April through July Period Forecast for 4 Selected Basins Determine what verification statistics say about our forecasting efforts by: Determine which tools might work better than others Look for patterns that indicate better performance or indicate a wasted effort Look at patterns of variability in data and forecasts. Look for Patterns in both the Observed and Forecasts Records
2
Big Cottonwood Creek Near Salt Lake City, UT Historical Stream Flow and Forecasts Increased Variability? 1899-2008 1979-2008
3
Big Cottonwood Creek Near Salt Lake City, UT Scatter Plot of Stream Flow Forecasts Over Forecasting Lows Under Forecasting Highs 1991-2008
4
Big Cottonwood Creek Near Salt Lake City, UT Historical Stream Flow and Forecasts Scatter Plots POR 1979-2008 Coordinated Forecasts are Consitently Too High
5
Big Cottonwood Creek Near Salt Lake City, UT Rank Histogram 3 times the calculated probability
6
Big Cottonwood Creek Near Salt Lake City, UT Mean Absolute Error Lead Time 14% 26% 6%
7
Bear River – Near Utah, Wyoming State Line Historical Stream Flow 1992-2008 1942-2008 1992-2008
8
Bear River – Near Utah, Wyoming State Line Scatter Plot of Stream Flow vs Forecast Values 1992-2008 2001-2008
9
Bear River – Near Utah, Wyoming State Line 1942-2008 1986 1977 1992-2008
10
Bear River – Near Utah, Wyoming State Line Skill Score – 2 periods 1992-2000 2001-2008
11
Bear River – Near Utah, Wyoming State Line Rank Histogram 2001-2008 1992-2000
12
Logan River Near Logan, Utah Historical Stream Flow & Forecasts 1983-2007 1900-2008 1958-1970 2000 1986
13
Logan River Near Logan, Utah Historical Stream Flow & Forecasts 1990-2008: Forecast POR 2001-2008: SWS seems to be over forecasting.
14
Logan River Near Logan, Utah Scatter Plot of Forecasts 1991-2008 1979-2008 Split Distribution of Values Distribution move evenly space going back to 1971
15
Logan River Near Logan, Utah Rank Histogram 1991-2008 1991-2008 Under Forecasting more Frequently in the Region of the 10% Exceedance Probability Over Forecasting Less Frequently in the 90% Exceedance Region than in the 10% Region.
16
Logan River Near Logan, Utah Rank Histogram 2001-2008 Better than Predicted forecasting of low flows Worse than Predicted Values when Forecasting High Flows. 2001-2008 Improved Stats Using the most recent years
17
Logan River Near Logan, Utah POD Above / POD Below Threshold for POR Consistently Lower POD “Above Threshold” - No Matter the Month Good POD of Volumes Below Threshold, except SWS in Jan, Feb.
18
Logan River Near Logan, Utah Mean Absolute Error – Lead Time 44kaf 28kaf 25kaf 23kaf 18kaf 13kaf
19
Weber at Oakley, Utah Historical Stream Flow and Forecasts 1990-2008 7 of Top 10 A-J Flows Occurred Before 1922
20
Weber at Oakley, Utah Historical Forecast Plots 2001-2008 ~ Using New Averages Seems To have Improved our Forecasting Efforts 1990-2000 ~ Prior to ESP and New Averages Under Forecasting was a Consistent Problem
21
Weber at Oakley, Utah Scatter Plot of Forecast vs Observed 1990-2008, no forecast were made prior to 1990 Clearly Under Forecasting the High Years
22
Weber at Oakley, Utah Rank Histogram
23
Weber at Oakley, Utah POD Above and Below Threshold
24
Weber at Oakley, Utah Mean Absolute Error Lead Time 38kaf 9kaf
25
Conclusions Great Basin Verification Task 2008 Increased Variability Findings indicate we detect low flows better than high flows with current and often over forecast the 10% exceedance values. Some Statistics Suggest that SWS often forecast too high in comparison to NWS, NRCS and ESP Forecasts That coordinated forecast numbers are skewed and result in over forecasting. That many of the data sets from 1991-2008 have a binary split indicating that more extremes have occurred in observed flows during the last 20 years. MOST IMPORTANTLY That our forecasting efforts have improved since 2001.
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.