Jump to content

Mid-term Probabilistic Operations Model (MTOM): Difference between revisions

From coloradoriverscience.org
No edit summary
No edit summary
Line 19: Line 19:
The 5-year inflow sequences are provided by the NOAA Colorado Basin River Forecast Center (CBRFC). The 35 sequences are generated by CBRFC's modeling system for [[seasonal streamflow forecasts]]; each sequence begins with the basin moisture conditions as of the MTOM run date and then is driven by the historical temperatures and precipitation for one of 35 overlapping 5-year periods sampled from the 1981-2015 observational record. This method is effectively the same as that used by CBRFC for their ensemble streamflow prediction (ESP) forecasts for seasonal streamflows, just over a longer time period.  
The 5-year inflow sequences are provided by the NOAA Colorado Basin River Forecast Center (CBRFC). The 35 sequences are generated by CBRFC's modeling system for [[seasonal streamflow forecasts]]; each sequence begins with the basin moisture conditions as of the MTOM run date and then is driven by the historical temperatures and precipitation for one of 35 overlapping 5-year periods sampled from the 1981-2015 observational record. This method is effectively the same as that used by CBRFC for their ensemble streamflow prediction (ESP) forecasts for seasonal streamflows, just over a longer time period.  


Upper Basin water demands are not explicitly modeled in 24MS (with three exceptions); instead, the inflow sequences from CBRFC have the expected Upper Basin diversions and consumptive use already subtracted from the flows that would have occurred under natural conditions. The exceptions are the Gunnison Tunnel, Azotea Tunnel, and Navajo Indian Irrigation Project, which are represented as separate diversions in 24MS. Lower Basin water demands are modeled based on monthly schedules provided by the water users.  
Water demands are handled in MTOM the same way as in 24MS. Upper Basin water demands are not explicitly modeled (with three exceptions); instead, the inflow sequences from CBRFC have the expected Upper Basin diversions and consumptive use already subtracted from the flows that would have occurred under natural conditions. The three exceptions are the Gunnison Tunnel, Azotea Tunnel, and Navajo Indian Irrigation Project, which are represented as separate diversions in MTOM. Lower Basin water demands are modeled based on monthly schedules provided by the water users.  


The output from 24MS consists of monthly projected elevations and storage, releases downstream, and hydropower generation (if applicable) for the following 12 reservoirs:
The output from MTOM consists of ~5 years of monthly inflows, projected elevations and storage, releases downstream, and hydropower generation (if applicable) for the same 12 reservoirs as in 24MS. But as indicated above, this "raw" output is seldom distributed in its entirety. Instead, the main operational use of MTOM output is to initialize CRSS for the MTOM/CRSS mid-term risk analyses. For this purpose, only the end-of-year 1 reservoir elevations projected by MTOM are retained; the 35 sets of elevations are passed on to CRSS to be each used as the starting point for multiple CRSS runs that project the system outcomes for years 2-5. In the most recent (January 2021) MTOM/CRSS analyses, two sets of inflow sequences were used for years 2-5 in CRSS: 113 sequences sampled from the 1906-2018 natural flow hydrology (Full hydrology), and 31 sequences sampled from the 1988-2018 hydrology (Stress Test hydrology)
*Fontanelle
*Flaming Gorge
*Taylor Park
*Blue Mesa
*Morrow Point
*Crystal
*Vallecito
*Navajo
*Lake Powell
*Lake Mead
*Lake Mohave
*Lake Havasu
 
As suggested above, 24MS is run monthly with a single inflow sequence, representing the ''most probable'' inflows. The ''most probable'' run in August--specifically, its projected elevations of Powell and Mead on January 1st--is used to set the tier for the Annual Operating Plan. The release from Lake Powell may be subsequently adjusted based on the results of the April run (projected elevations on September 30th).
 
Additionally, four times a year (January, April, August, October), two additional 24MS runs are done, using ''probable minimum '' and ''probable maximum'' inflow sequences provided by CBRFC. These two runs, which represent 90% and 10% exceedance probabilities, help bracket the uncertainty in the main ''most probable'' run. Under the terms of the 2019 Drought Contingency Plans, whenever a ''probable minimum'' run shows Powell going below 3525' at any time in the next 24 months (as happened in January 2021), both of these additional 24MS runs are done monthly instead of four times per year to provide basin stakeholders with additional guidance to prepare for drought impacts.





Revision as of 16:34, 14 April 2021

Overview

The Mid-term Probabilistic Operations Model (MTOM) is one of three river system models developed and maintained by Reclamation to support operational decision-making and planning for the Colorado River Basin. All three models are implemented in the RiverWare modeling platform and share some key features. RiverWare is an object-oriented modeling platform in which the objects may be inflow points, river reaches, diversions, reservoirs, canals, pipelines, and other water-resource features.

MTOM was developed directly from the 24-Month Study Model (24MS) so the two models have very similar structures and representations of operations. Compared to 24MS, MTOM extends the timescale of projection out to 5 years (60-68 months, depending on the starting month), vs. 2 years, and MTOM also allows for the input of a large ensemble (typically 35) of inflow traces to facilitate probabilistic risk analysis, as well as more robust testing of the inputs themselves, vs. the deterministic single-trace runs of 24MS.

MTOM is used for mid-term (2-5 years) system projections for risk-based operational planning and analysis. Operationally, MTOM is paired with the Colorado River Simulation System (CRSS) to create a set of "hybrid" mid-term (5-year) projections that leverage the input and output capabilities of both models. These MTOM/CRSS runs are performed twice per year, in January and August. MTOM is also run on its own without CRSS to create simpler probabilistic analyses; these projections are not readily available but can be requested from Reclamation.

Relevance

The envelope of the MTOM (and MTOM+CRSS) projections of Lake Powell and Lake Mead levels over the 5-year mid-term period provides guidance to Reclamation and basin stakeholders about the probability of undesirable system outcomes over that period (e.g., Lower Basin delivery reductions, reservoir levels falling below the power pool). This information can help water managers and users prepare for those outcomes in accordance with their own risk tolerances.

Model Structure and Operation

Figure 1. The Mid-term Probabilistic Operations Model (MTOM) in object view, including the 12 reservoirs in the bullet list at left, which are shown with the purple triangles. (Source: Reclamation)

MTOM takes a set of projected inflow sequence and runs them through prescribed operational rules at 12 Reclamation reservoirs to project the monthly conditions at those reservoirs over the next 60-68 months, depending on the starting month of the run. A total of 19 inflow points (vs. 16 in 24MS) are used to represent both inflows to those reservoirs and intervening flows between reservoirs. The three additional inflow points allow the automation in MTOM of a process that is done manually in 24MS.

The 5-year inflow sequences are provided by the NOAA Colorado Basin River Forecast Center (CBRFC). The 35 sequences are generated by CBRFC's modeling system for seasonal streamflow forecasts; each sequence begins with the basin moisture conditions as of the MTOM run date and then is driven by the historical temperatures and precipitation for one of 35 overlapping 5-year periods sampled from the 1981-2015 observational record. This method is effectively the same as that used by CBRFC for their ensemble streamflow prediction (ESP) forecasts for seasonal streamflows, just over a longer time period.

Water demands are handled in MTOM the same way as in 24MS. Upper Basin water demands are not explicitly modeled (with three exceptions); instead, the inflow sequences from CBRFC have the expected Upper Basin diversions and consumptive use already subtracted from the flows that would have occurred under natural conditions. The three exceptions are the Gunnison Tunnel, Azotea Tunnel, and Navajo Indian Irrigation Project, which are represented as separate diversions in MTOM. Lower Basin water demands are modeled based on monthly schedules provided by the water users.

The output from MTOM consists of ~5 years of monthly inflows, projected elevations and storage, releases downstream, and hydropower generation (if applicable) for the same 12 reservoirs as in 24MS. But as indicated above, this "raw" output is seldom distributed in its entirety. Instead, the main operational use of MTOM output is to initialize CRSS for the MTOM/CRSS mid-term risk analyses. For this purpose, only the end-of-year 1 reservoir elevations projected by MTOM are retained; the 35 sets of elevations are passed on to CRSS to be each used as the starting point for multiple CRSS runs that project the system outcomes for years 2-5. In the most recent (January 2021) MTOM/CRSS analyses, two sets of inflow sequences were used for years 2-5 in CRSS: 113 sequences sampled from the 1906-2018 natural flow hydrology (Full hydrology), and 31 sequences sampled from the 1988-2018 hydrology (Stress Test hydrology)


Data and Tools

24-Month Study Reports

The output of the primary monthly run of the 24-Month Study Model (using most probable inflows) is released around the 15th of each month. These results are formatted by Reclamation as a standardized set of data tables totaling 17 pages. The Upper Colorado and Lower Colorado regional offices package this set of data tables differently; Upper Colorado adds a 9-page narrative summary for the Upper Colorado reservoirs, while Lower Colorado adds a single cover page with very a brief summary. Note that both report "flavors" contain the results for both Upper Basin and Lower Basin reservoirs.

24-Month Study Reports - as distributed by Upper Colorado Region

24-Month Study Reports - as distributed by Lower Colorado Region

Supplemental 24-Month Study runs

As noted above, two supplemental runs of 24MS are performed either 4 times per year, or monthly, depending on projected reservoir conditions. These runs, based on probable minimum and probable maximum inflows, bracket the uncertainty in the primary run.

Probable Min/Probable Max runs

Additional Resources

State of the Science Report

Chapter 3 of the State of the Science report, Primary Planning Tools, describes the three primary Reclamation system models in detail.

Research Directions

New and Notable Research (2020-present)

[<URL FOR PAPER>]

Summary

Contact

MTOM is maintained and updated by Reclamation Upper Colorado and Lower Colorado regional staff who are based at the University of Colorado Boulder's Center for Advanced Decision Support for Water and Environmental Systems (CADSWES). The RiverWare modeling platform was also developed at CADSWES.

For additional information or questions, contact them at ColoradoRiverModeling@usbr.gov.