Water Management Accuracy with iSnobal

M3Works' Rigorous Validation Process

In snowpack modeling for water management, validation is often overlooked, leading to mediocre results that fail to capture real-world scenarios. However, at M3Works, validation is not just a step in the process; it's a philosophy that underpins our approach in producing accurate models.

By validating our iSnobal modeling data with real on-the-ground data, we can pinpoint discrepancies and prevent incremental drift from reality. This meticulous validation process allows us to maintain a close alignment with real-world conditions. Our models provide actionable insights for effective water resource management.

At M3Works, we recognize that data validation should rely on multiple factors, including precipitation, snowpack depth, density, and temperature, and forest conditions. By looking at each of these, we can develop a comprehensive understanding of the complex interactions within the water system.

SWE in the Boise River Basin including locations of data validation.

Our data modeling process

Our modeling process begins with the latest data from the National Weather Service, typically provided in a 3km grid format. However, we don't stop there. To achieve the highest level of accuracy, we further refine the input data down to a 50m grid. Our downscaling technique relies on high resolution inputs for elevation and canopy structure. Our models capture even the smallest nuances in the terrain and weather. The 50m grid allows us to fully utilize on-the-ground measurement networks like the NRCS SNOTEL and California Department of Water Resources CDEC networks for real-time time-series comparisons of snowpack mass, depth, and density, as well as forcing variables like soil temperature, precipitation, and air temperature. High resolution comparisons allow the model state to stay tightly coupled with reality.

We also track data in the field. Our modeling system is designed to take in LiDAR derived depth from the Airborne Snow Observatories, inc., as well as aerial or space-born Spectrometer derived snow albedo. In addition, we utilize the latest high quality LiDAR derived forest height metrics. This partnership of high accuracy measurements and reliable, physically based modeling is a key piece to accurately simulating snowmelt and runoff processes. It is also the foundation for which modeling improvements are being made. By validating precipitation, snowpack state, and forest conditions, among other factors, we create a real-time, physically based, high-resolution model of the snowpack. This model serves as a powerful tool for water resource managers, enabling informed decision-making and proactive measures to address water-related challenges.

The provided image below showcases the precision of our methodology. It presents a side-by-side comparison of the modeled snow water equivalent and the measured snow water equivalent. Through thorough validation of various factors, we have successfully maintained a high level of accuracy, ensuring that the modeled SWE closely aligns with the measured SWE.

Comparison of modeled vs measured Snow Water Equivalent for the Boise River Basin.