Premium
An evaluation of terrain‐based downscaling of fractional snow covered area data sets based on LiDAR‐derived snow data and orthoimagery
Author(s) -
Cristea Nicoleta C.,
Breckheimer Ian,
Raleigh Mark S.,
HilleRisLambers Janneke,
Lundquist Jessica D.
Publication year - 2017
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1002/2017wr020799
Subject(s) - downscaling , snow , environmental science , remote sensing , watershed , terrain , moderate resolution imaging spectroradiometer , meteorology , precipitation , satellite , geography , cartography , computer science , machine learning , aerospace engineering , engineering
Reliable maps of snow‐covered areas at scales of meters to tens of meters, with daily temporal resolution, are essential to understanding snow heterogeneity, melt runoff, energy exchange, and ecological processes. Here we develop a parsimonious downscaling routine that can be applied to fractional snow covered area (fSCA) products from satellite platforms such as the Moderate Resolution Imaging Spectroradiometer (MODIS) that provide daily ∼500 m data, to derive higher‐resolution snow presence/absence grids. The method uses a composite index combining both the topographic position index ( TPI ) to represent accumulation effects and the diurnal anisotropic heat ( DAH , sun exposure) index to represent ablation effects. The procedure is evaluated and calibrated using airborne‐derived high‐resolution data sets across the Tuolumne watershed, CA using 11 scenes in 2014 to downscale to 30 m resolution. The average matching F score was 0.83. We then tested our method's transferability in time and space by comparing against the Tuolumne watershed in water years 2013 and 2015, and over an entirely different site, Mt. Rainier, WA in 2009 and 2011, to assess applicability to other topographic and climatic conditions. For application to sites without validation data, we recommend equal weights for the TPI and DAH indices and close TPI neighborhoods (60 and 27 m for downscaling to 30 and 3 m, respectively), which worked well in both our study areas. The method is less effective in forested areas, which still requires site‐specific treatment. We demonstrate that the procedure can even be applied to downscale to 3 m resolution, a very fine scale relevant to alpine ecohydrology research.