Assessment of the snow conditions in the major watersheds of Afghanistan using multispectral and passive microwave remote sensing

Abstract

Since the winter season of 2004–2005, annual snow assessments have been conducted for the major watersheds of Afghanistan using multispectral (AVHRR and MODIS) and passive microwave (SSM/I and AMSR‐E) remote sensing technologies. Because of limited ground‐based observations of precipitation and snow pack conditions, remote sensing provides a unique opportunity to assess these conditions at different scales offering an appraisal of current conditions from a historical context. This paper describes the methodology that has been developed over the past seven winter seasons, wherein bi‐weekly snow products and assessments are produced including the following: current snow‐covered area (SCA) at regional and watershed scales; estimation of SCA by elevation band; current snowpack volume [snow water equivalent (SWE)] for each watershed with a historical perspective (1987–present); snow condition outlook by watershed; general summary of snow conditions based on remote sensing products and limited ground‐based observations; and if warranted, a snow melt flooding advisory. Further analysis of these snow products and assessments are presented. Moreover, comparisons between both available passive microwave estimates of SWE (SSM/I and AMSR‐E) for all watersheds differ in magnitude yet exhibit similar historical trends. The SSM/I versus AMSR‐E historical differences in SWE for each of the major Afghanistan watersheds are quantified and show a strong relationship (R2 = 96%) to the mean elevation of the watershed. Copyright © 2012 John Wiley & Sons, Ltd.

Using machine learning for real-time estimates of snow water equivalent in the watersheds of Afghanistan

“…In every basin on 1 April, there were many areas with more reconstructed SWE than the passive microwave saturation limit of 150 mm (Hancock et al, 2013;Kelly et al, 2003;Takala et al, 2011;Tedesco and Narvekar, 2010). Basin wide, differences between the results of this study and the results reported in Daly et al (2012) were small, < 8 mm, for all basins except the Upper Kabul (22 mm difference), which had the highest average SWE in both studies.…”

“…By basin, NS efficiencies are higher averaging 0.89 and ranging from 0.87-0.91, showing that most of the forecasting challenge occurs year-to-year. Given the lack of water resources monitoring and management in Afghanistan and the high proportion of runoff from snowmelt (Daly et al, 2012;Vuyovich and Jacobs, 2011), our machine learning predictions could improve runoff predictions, which are likely based on mean conditions. Figure 10 shows the predictor performance, ranked by importance, a relative measure ranging from 0 to 1 explained in Sect.…”

“…1 Daly et al, 2012), the country outline, the h23v05 MODIS tile, and areas with permanent snow and ice. 3000 m with long term (albeit not current) climate records (Table 1).…”

“…Instead, we used passive microwave-based SWE estimates from AMSR-E from a previous study (Daly et al, 2012); Table 3 shows comparisons of SWE on 1 April.…”

Spatiotemporal patterns of High Mountain Asia's snowmelt season identified with an automated snowmelt detection algorithm, 1987–2016

“…Studies have noted that SWE estimates from the Chang equation have high uncertainties (e.g., Kelly et al, 2003;Kelly, 2009;Tedesco and Narvekar, 2010;Daly et al, 2012), particularly in dense forests. However, as much of our study area is non-forested -and we use SWE only as a rough estimate of snow volume -we choose to rely on the simple Chang equation rather than a more complex algorithm for SWE estimation.…”