STREAMFLOW PREDICTION WITH PHYSICALLY DISTRIBUTED WATERSHED MODEL IN

STREAMFLOW PREDICTION WITH PHYSICALLY DISTRIBUTED WATERSHED MODEL - IN A SNOWMELT-DOMINANT WATERSHED DK Kang, John Choi, Jiyue Zhu, Ed Kim, and Leung Tsang AGU, San Francisco, CA, December 2019 Abstract Study Area: Senator Beck, CO Streamflow Prediction Two dimensional Runoff e. Xport (TREX) Snowmelt-runoff process exabits different hydrological features compared to the rainfall runoff. This paper implements a temperature index snowmelt method to a physically based hydrology model. The model is applied to the Senator Beck site where the recent NASA Snow. Ex snow field campaign was conducted in 2016 -2017. A 10 m resolution DEM is used to extract the watershed along with soil hydraulic properties and land cover to drive the snowmelt-runoff model in overland channel grid cells. The physically based land surface watershed model, the Colorado State University TREX is modified to handle snowmelt as a rainfall when the air temperature is above zero degree celcius in the 10 m grid at each time step. The weather forcing dataset was extracted from a nearby USDA SNOTEL station, named the Red Mountain. With a support from Colorado Avalanche Center, the streamflow observation at the outlet of the Senator Beck watershed is compared with the modeled streamflow at the same outlet location. This study will further to equip an energy balance approach to determine the snow temperature at each time step, and a single snow layer will be modified to have a multi-layer scheme which is relevant to the represent a micro-structure of the layered snow. Fig. 4. Input dataset to drive TREX including initial SWE, 50 m DEM, soil classification, land cover, channel link/node, and a location of the streamflow gauge. 5 th Fig. 3. Schematic view of TREX framework along with an amendment of snow hydrological processes such as ‘snowfall’, ‘snowmelt’, and an addition to the derivative of the water depth with a released water from a snowmelt Objectives * To implement snowmelt-runoff routine to TREX framework to handle a streamflow prediction driven by snowmelt release * To identify an amount and a timing of the peak streamflow associated with a distributed snow water equivalent in a basin and a temperature rise above 0 degree Celsius * To understand snowmelt runoff in a rugged mountain watershed toward a streamflow generation * To demonstrate a further implementation of a snow subroutine to handle an energy balance approach and a multi-layer snow configuration Datasets & Model q Senator Beck Watershed maintained by Colorado Avalanche Center q 10 m DEM, soil classification map, land use map, channel link/node, channel specifications with dimensions q A physically distributed watershed model Fig. 1 Overview of CASC 2 D (previous TREX) Framework Fig. 2 (Top) Overview of the Senator Beck area and distribution of observation networks. Study areas outlined in blue are the (bottom) an arial view of the Senator Beck watershed distributed with weather observations at SBSP and SASP, and a streamflow gauge observation at SBSG. The Senator Beck Basin Area (SBBA) is located in in the Ouray Ranger District of the Uncompahgre National Forest in the west of San Juan Mountains in a southwestern part of Colorado State. With a special environmental permit, a Center for Avalanche Studies (CAS) has maintained environmental observing stations at the outlet of the SBBA including three weather stations and one streamflow gauge. The Swamp Angle Study Plot (SASP) is located just below treeline at 3, 371 meter a. s. l. . The Senator Beck Basin Study Plot is above the treeline at 3, 714 meter a. s. l. , and the Putney Study Plot (PTSP) is within a ridgeline at 3, 756 meter a. s. l. within a private lot. Lastly, the Senator Beck Basin Stream Gauge (SBSG) station is situated in a sharp gorge at the hydrologing outlet point of the watershed. Figure 3 illustrates a method how the snow hydrological processes are implemented to the existing TREX framework. As mentioned earlier, only a hydrology part is turned on in the TREX framework. And, the snow hydrology is attached to existing hydrology subroutines including rainfall, infiltration, and overland/channel water balance. To accommodate a precipitation as a snowfall, air temperature is read from a main input file at each available time step. A temporally interpolated air temperature is adjusted with an adiabatic lapse rate based on an elevation of the weather station, which is also read from the input file as a main argument of the TREX. When the air temperature is below 273. 15 Kelvin, the rainfall becomes snowfall at that time and location. Implementations to handle the snowfall are inserted to a ‘Rainfall. c’ subroutine when a snow option is turned on. To calculate a derivative to update a water depth with a time step to a current water depth, net-rain rate was only used without the snow hydrological processes. But, right before the ‘Overland. Water. Depth. c’, ‘Snowmelt. c’ is implemented to create the melt water based on a temperature difference between the air temperature at that time and location and 273. 15 Kelvin. A coefficient is used with the temperature difference to determine the amount of the released melt-water [Dingman 1991]. hour 12 th hour 24 th hour Fig. 5. Snapshots of water depth in Senator Beck released from snowmelt. 72 hours of simulations are extracted to show that upstream snow starts to melt, and channel is charged with melt water, and finally the outlet has the higher water depth. Summary . 1) TREX is modified to equip a snowmelt runoff scheme by using a temperature index method. GIS dataset provided by Colorado Avalanche Center were utilized to provide the input files. Nearby Snotel station, Red Mountain is used to provide a time series of weather forcing dataset such as air temperature, precipitation, wind speed, air pressure, and etc. 2) The secondary Snow. Ex site for Year 1 was selected to apply the modified TREX to evaluate a streamflow released from a snowmelt-runoff Affiliations 1 Research Scientist, NASA GSFC, Greenbelt MD 20771 dk. kang@nasa. gov 2 Ph. D. Candidate, U. of Michigan, Ann Arbor, MI, 48109 ziyuezhu@umich. edu 1 Physical scientist, NASA GSFC, Greenbelt, MD 20771 ed. kim@nasa. gov 3. Undergraudate Programmer, Ohio State University, Columbus, OH, 43210 choi. 1655@buckeyemail. osu. edu Acknowledgements GIS data and streamflow gage observation are provided by Jeff Derry at Colorado Avalanche Center