Across the American West, water managers continually face a version of the same critical question: What do current snowpack, precipitation, and streamflow conditions mean for our water supply?
For Spheros Environmental water resources engineer Page Weil, that question, not a dataset or a model, is where effective forecasting begins.
“There is no one-size-fits-all solution that can answer what the snowpack or spring rainfall means for every water manager,” Weil explains. “With every project, I spend time with the client learning the nuances of how they make decisions and identifying where there’s room to improve that process.”
Modern water data becomes valuable when it is connected to a decision a manager needs to make. Without that connection, tools can produce more information without producing more clarity.
“The best forecasting model in the world is not useful if it does not help someone make a better decision,” Weil says. “Local water managers understand their systems in ways that a model never fully will. Our job is to build tools that enhance that knowledge, answer the questions they actually care about, and communicate the results clearly enough that they can act on them.”
What is decision-first water supply forecasting?
Decision-first water supply forecasting begins by identifying the operational or planning decision a water manager needs to make, then selecting the data, models, and methods needed to support it.
It is tempting to treat forecasting as a data problem: gather enough streamflow records, snowpack readings, and climate model output, and the right answer will surface. In practice, that approach tends to produce the opposite of clarity. Without a defined decision to work toward, more data does not sharpen a forecast. It can add noise and increase the risk of selecting or emphasizing the information that supports an answer someone already expects.
Starting with the decision helps reduce that risk. Before Weil or his team select a dataset, they define the specific question a client is trying to answer. That may be when an irrigation water right will go out of priority, whether a reservoir will fill this year, or how much water is available for a contracted release. Only then do they work backward to identify which datasets and models are relevant to that question and which are simply available.
This is also why Spheros presents forecasts as ranges rather than single answers.
“We do not say, ‘Here is the answer.’ We say, ‘Based on the information we have, we expect this range of dates, this range of volumes, this range of flow rates, or this likelihood that a particular outcome will occur,'” Weil explains. “The worst thing you can do in forecasting is to be confident and wrong. If you issue a highly confident forecast and miss badly, people quickly lose trust in the tools you are building.”
For reservoir operators, the uncertainty has real operational consequences.
“Reservoir operators often have to decide whether to release water to create space for future snowmelt and reduce flood risk,” Weil says. “Across the western United States, that decision must be balanced against the ongoing risk of drought and the possibility of releasing water that may be needed later. Understanding how an operator makes that decision helps us identify the data and tools that can make the process more informed, confident, and defensible.”
This principle is well established in the research on how scientific information actually gets used in decision-making. Research into the relationship between climate science and public decisions has found that information is more likely to be useful when it is developed through ongoing interaction between the people producing it and the people who will act on it, rather than delivered as a finished product built around whatever data was easiest to gather. Decision-first forecasting applies that principle by allowing the question to define what information is relevant.
Why western water systems require customized forecasting
Understanding why this sequence matters requires understanding how Western water systems actually work and why the same forecast can mean something entirely different to two different clients.
“In the West, half of the water in all of our streams comes from the snowpack,” Weil says. “The snow is our largest reservoir.” In drought years like the current one, which Weil describes as a historic drought, the downstream effects ripple differently across every type of water user, which is exactly why a generic forecast cannot answer a specific manager’s question.
Consider the contrast between two project types Spheros has worked on. The first is what hydrologists call a streamflow-dominated system. In a drought year, “they don’t have enough reservoir storage in their basin to allow them to use water in a business-as-usual way through multiple drought years. They have enough to buffer themselves while they promote aggressive water restrictions during a year like this, but they don’t really have a lot of flexibility.”
Other clients, by contrast, hold more than a year’s worth of demand in storage, giving them a meaningful buffer that allows for longer-range planning and steadier management through lean years. The decision each client needs help with, aggressive near-term restriction versus multi-year planning, is entirely different, even though both start with the same question about snowpack.
The type of water use matters just as much as the volume. Municipal indoor water use is largely returned to the stream through treatment facilities, while agricultural diversions typically result in substantially greater consumptive use.
“The same volume of water can have a very different impact depending on who is using it, where it is used, and how much returns to the stream,” Weil notes. “That context matters when we evaluate a client’s risks and opportunities.”
All of this operates within Colorado’s prior appropriation system, a framework of water rights administration that dates to the 1850s and determines who receives water when supplies run short. Most clients already understand where they sit in priority. The decision that needs support is what that position means for their long-term strategy. Answering that question requires looking beyond an individual water right to understand how neighboring systems, larger users, and senior water-right holders operate around it.
Short-range forecasting vs. long-range water planning
Because client decisions vary so widely, the technical approach to water resources planning generally follows two related but distinct tracks: long-range risk and short-range decision support. Each answers a different kind of question.
Long-range planning draws on historical drought patterns, changing snowpack conditions, and global climate model data to evaluate how hydrology may change over the next 10, 20, 50, or even 100 years.
“You cannot look at one year, or even a handful of years, and assume those conditions are a reliable basis for constructing a reservoir or acquiring new water rights,” Weil explains.
Long-term investments, whether acquiring water rights, constructing storage, or building pipelines, need to be informed by a realistic understanding of how precipitation, snowpack, and water availability may change over decades.
Short-range decision support works on a different timescale and requires a different kind of tool. The driving question is what the current year’s runoff and water-use forecast means for a system right now, in terms a manager can act on. That might mean predicting when irrigation water rights will go out of priority, how much water is available for contracted releases, or when a reservoir is likely to fill or not fill.
“Water users often have a deep understanding of their own systems, built through years or even generations of experience,” Weil says. “Modern forecasting tools do not replace that knowledge. They give people another layer of information so they can make decisions earlier, evaluate a wider range of possibilities, and respond more confidently as conditions change.”
How decision-first water forecasting works
Weil describes his approach to building water decision-support tools as a layered process. The layers are organized around decisions, not data availability. Each layer exists to answer a specific question a client has, and the data gets added only after that question is clear.
What is happening in the system now?
The first layer uses observed station data, including streamflow gauges, climate stations, snowpack measurements, and diversion records to answer the most immediate question a manager asks: what is happening in my system right now?
“The first step is to give clients a clear picture of what is happening around them based on the full range of observed data,” Weil explains. “These are measurements they know and understand because they reflect conditions they see in their own systems.”
On a recent project, Weil helped deploy a web viewer that brings together hundreds of measurement points across a district, giving client staff an accessible way to monitor conditions and explore their own system data in one place.
What is happening between measurement points?
The second layer answers a broader question: what is likely happening between my measurement points and across the wider region I depend on? This is where remotely sensed and gridded data sets come in, including recent rainfall, temperature trends, snowpack conditions, and evapotranspiration estimates that indicate how much water irrigated crops are likely to need. These data sets fill the gaps that station data alone cannot cover.
What does the information mean for the decision?
The third layer answers the question that brought the client to Spheros. This is where the modeling expertise matters most, combining observed and gridded data within forecasting frameworks designed to produce outputs tied to a specific choice.
“That can mean forecasting streamflow, estimating how much water irrigated lands are likely to use, evaluating what a client’s water-rights portfolio can support, or anticipating when watering restrictions may be needed,” Weil says. “The goal is not simply to provide more data. It is to help clients understand what the data mean for the decisions they need to make.”
Building water forecasting capability across Colorado
The Colorado Water Supply Measurement and Forecasting Program is a statewide effort to build forecasting capability across Colorado and put modern tools in the hands of local water managers without losing sight of the decisions those tools need to support.
Weil has supported the program for seven years.
“We know there are powerful tools and data sets out there, but many of them are expensive, highly technical, or difficult for an individual water manager to access,” he explains. “The program helps bring those resources within reach and turn them into decision-support products that people can use in the real world.”
Through regular stakeholder calls with water managers across the state, Weil has developed a clear picture of the questions that matter most on the ground: Which water rights are likely to go out of priority, and when? Is a reservoir likely to fill this year? How much water will be available for contracted releases through the end of the season?
The program begins with those questions and works backward to determine what information is needed and which tools and technologies can turn that information into useful knowledge.
Learn more about the Colorado Water Supply Measurement and Forecasting Program.
Decision-first forecasting at McPhee Reservoir
The Dolores Water Conservancy District (DWCD) in southwest Colorado manages a large multi-use reservoir, and its situation shows what starting with the decision looks like in practice. The DWCD juggles multiple competing obligations from a single storage facility: filling McPhee Reservoir, releasing water for downstream fish health requirements, supporting the boating community, meeting irrigation contracts for various local canal companies, and maintaining municipal supply. “Fundamentally, it’s one large bucket that has to be used to supply all of these competing interests.”
The decision the district needed help with was not “give us more water data.” It was “help us sequence a set of releases against five different obligations with limited lead time.” When Spheros came on board, the DWCD was relying on a batch of spreadsheets more than 20 years old as its core decision-making infrastructure. Modernizing that infrastructure meant building a forecasting framework designed around the sequencing decision itself, not simply digitizing the spreadsheets.
The tool Spheros developed gives district staff enough lead time to make sequenced decisions: when to hold releases, when to release for fish health requirements, and how to structure release schedules for boatable flows while preserving reservoir storage. In a dry year, it helps staff understand whether the reservoir is likely to fill at all and, if not, how to manage competing obligations with limited supply. “They want to do their best to optimize the use of McPhee reservoir. And what our tools are letting them do is give them enough heads up so that they understand if they need to make a short-term release.” As Weil puts it, “the goal is helping clients to make those decisions quickly and confidently.”
How better forecasting supports water security
At a community level, the payoff from decision-first forecasting is measured in water security and smarter infrastructure investment.
For a municipal water provider heading into a drought year, good short-range decision tools mean the ability to communicate clearly and early with customers about restrictions, rather than scrambling to respond after conditions deteriorate. “Giving them tools to make that decision to protect their water security over one year and into the next is one outcome.”
For providers considering capital investments, long-range planning tools inform decisions that carry multi-decade consequences: whether to build a new reservoir, acquire water rights, enter a new agreement, or fund infrastructure upgrades. “The other outcome is, if we are going to help you develop a new project or build new infrastructure, you want to make sure those multi-million dollar decisions are informed by sound science.”
Sometimes the highest-value work is not the most sophisticated. Drought planning, stakeholder communication, and straightforward education about how water rights exposure works can deliver meaningful gains in water security without requiring a major technology investment. “The simplest planning efforts and understanding how groups are exposed to drought and what drought actually means can go a long way to improving water security.”
Whether the final product is a complex forecasting dashboard, a long-range infrastructure plan, or a clear drought communication strategy, the process starts the same way: identify the decision, then build backward to the data and tools that support it. That sequence, not the sophistication of any single tool, is what makes environmental data defensible.
FAQs About Decision-First Water Forecasting
Why should forecasting start with the decision instead of the data?
Starting with data first tends to produce more information without more clarity, and it creates a risk of selecting data that supports a preconceived answer. Starting with a specific decision, such as when a water right will go out of priority or whether a reservoir will fill, defines exactly which data and models are relevant, and keeps the forecasting process focused on an answer a manager can actually use.
What does a water decision-support tool actually do for a water manager?
A well-designed tool translates raw environmental data such as snowpack measurements, streamflow gauges, and satellite-derived evapotranspiration estimates into actionable outputs tied to a specific decision a manager needs to make. The goal is not a single confident answer, but a range of likely outcomes with explicit uncertainty, so managers can act with confidence rather than guessing.
Why can’t water managers just use off-the-shelf forecasting tools?
Publicly available AI-driven forecasting tools and remote sensing data sets have made sophisticated water analysis more accessible than ever. But the bottleneck is the expertise needed to identify the right decision to solve for, evaluate which tools are appropriate for that decision, and connect their outputs to the actual choice a client faces. A forecasting model that cannot be tied back to a specific decision has no practical value, regardless of its technical sophistication.
How does Spheros Environmental build water forecasting tools for clients with complex multi-use water systems?
Spheros starts by identifying the specific decisions a client needs to make, then layers in observed station data (gauges, snowpack, diversions), remotely sensed and gridded data sets, and finally a forecasting framework built to answer those decisions directly. For one southwest Colorado conservancy district managing a multi-use reservoir, this meant replacing 20-year-old spreadsheets with a tool built around the district’s actual sequencing decisions across fish health requirements, irrigation contracts, boating schedules, and municipal supply.
What is the Colorado Water Supply Measurement and Forecasting Program and who does it serve?
Established by state legislation, the program is a statewide effort to build modern water forecasting capability and put those tools directly in the hands of local water managers across Colorado. It identifies promising forecasting technologies, funds their deployment, and supports the translation work between technical model outputs and the on-the-ground decisions real water managers need to make. More information is available at coloradosnow.org.
How does long-range climate planning differ from short-range water decision support?
Long-range planning draws on global climate model projections spanning decades to inform major infrastructure decisions like whether to build a reservoir, acquire water rights, or fund pipeline upgrades. Short-range decision support works on a seasonal or annual timescale, helping managers understand what the current year’s snowpack or drought conditions mean for their specific system right now. Both start from a defined decision, but the decision itself determines which data and tools are appropriate.
About the Author
As a Senior Water Resources Project Manager and Water Resources Modeler at Spheros Environmental, Page specializes in climate science, water resources management, and engineering. Page’s work focuses on developing innovative decision-support tools and water resource models that help agencies make informed, data-driven decisions to address complex water management challenges and build resilience for the future.