Kirchner et al., 2013

Talk/Poster

Estimating forest snow accumulation with LiDAR derived canopy metrics, southern Sierra Nevada, California.

Kirchner, P., R.C. Bales, and T.H. Painter (2013)
American Geophysical Union, Fall Meeting 2013, abstract #H13J-1507  

Abstract

Water resources from mountain snow-melt are becoming more difficult to predict as demand for them increases. Thus more effective methods of determining the spatial and temporal distribution of mountain snow is of utmost importance, particularly in forested terrain where remote sensing is limited by canopy cover and few sensor networks exist. We present an analysis of LiDAR derived digital elevation and canopy surface models as emerging metrics for quantifying snow depth accumulation in forest covered terrain. Our analysis compares hourly snow depth sensor measurements, collected over four water years from 26 stratified locations, to canopy metrics measured with airborne scanning LiDAR and effective LAI measured in-situ, using LAI - 2000 and hemispheric photos. The period of record represents dry, normal, and wet years where peak mean snow depths were 73 - 165% of the 30 year April 1st mean. Normalized fractional totals were calculated at each location for the 20 largest monthly continuous precipitation events. Mean values within a 2- 40 m radius of six site specific canopy characteristics: mean canopy height, standard deviation of canopy height, canopy surface ratio, fractional canopy cover, and coefficient of variation were determined from 1 m gridded LiDAR surface models at each snow sensor location. In all cases at least twice the variability was present at radii < 20 m. Minimum gap distance to canopy >2 m, estimated from LiDAR altimetry, and effective LAI, estimated from in-situ measurements, were also determined for each location. The predictive capacity of canopy metrics for snow depth variability was found to be dependent on the radius of integration with the highest coefficients of determination at radii between 8 - 12 m. When compared to mean fractional snow accumulation at all sensors, we found: r2 and p in parenthesis, mean canopy height (0.58, < 0.001), standard deviation (0.51, < 0.001), canopy to ground surface ratio (0.48, < 0.001) and fractional canopy cover (0.45, < 0.001) showed the highest coefficients of determination and maximum canopy height (0.40, < 0.001), coefficient of variation (0.38, < 0.001) minimum gap distance (0.25, < 0.001) and both methods of effective LAI (0.38, 0.002 and 0.27, 0.01) the lowest. These metrics, measured as part of a single or ongoing airborne snow observation campaign, provide a basis for estimating snow depth accumulation in locations where few ground measurements are made and limited direct LiDAR snow depth altimetry is available.

Citation

Kirchner, P., R.C. Bales, and T.H. Painter (2013): Estimating forest snow accumulation with LiDAR derived canopy metrics, southern Sierra Nevada, California . American Geophysical Union, Fall Meeting 2013, abstract #H13J-1507.

This Paper/Book acknowledges NSF CZO grant support.