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

Bair, Edward H.; Calfa, Andre Abreu; Rittger, Karl; Dozier, Jeff

doi:10.5194/tc-12-1579-2018

Cited by 82 publications

(78 citation statements)

References 71 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their analysis used images with MODIS sensor zenith angles within 30°of nadir, which leads to less skew and other viewing geometry problems (Tan et al, 2006;Xiaoxiong et al, 2005). Our previous work , Bair, Calfa, et al, 2018Rittger et al, 2016) used these products for SWE reconstructions.…”

Section: Water Resources Researchmentioning

confidence: 99%

“…Note that these errors represent the best match to the in situ measurements from a 9-pixel neighborhood centered on each of the sites. This best-of-9-pixel neighborhood approach is often used with MODIS measurements (Bair, Calfa, et al, 2018;Rittger et al, 2016) because of geolocational uncertainty (section 2.2) and spatial variability of the snow surface. Figures 4d-4f clearly show that CUES is less dusty than SASP or SBSP with no ΔVIS values larger than 25% and many in situ observations indicating a value of zero.…”

Section: Age-based Albedo Decay Modelmentioning

confidence: 99%

See 1 more Smart Citation

An Examination of Snow Albedo Estimates From MODIS and Their Impact on Snow Water Equivalent Reconstruction

Bair

Rittger

Skiles

et al. 2019

Water Resources Research

Self Cite

View full text Add to dashboard Cite

Snow albedo is a dominant control on snowmelt in many parts of the world. An empirical albedo decay equation, developed over 60 years ago, is still used in snowmelt models. Several empirical snow albedo models developed since show wide spread in results. Remotely sensed snow albedos have been used in a few studies, but validations are scarce because of the difficulty in making accurate in situ measurements. Reconstruction of snow water equivalent (SWE), where the snowpack is built in reverse, is especially sensitive to albedo. We present two new contributions: (1) an updated albedo model where grain size and light absorbing particle content are solved for simultaneously and (2) multiyear comparisons of remotely sensed and in situ albedo measurements from three high‐altitude sites in the western United States. Our remotely sensed albedos show 4 to 6% RMSE and negligible bias. In comparison, empirical albedo decay models, which require extensive in situ measurements, show RMSE values of 7 to 17% with biases of −6 to −14%. We examine the sensitivity of SWE reconstructions to albedo error at two sites. With no simulated error in albedo, reconstructed SWE had MAE values of 7 to 13% and 5–6% bias. The accuracy actually improved with some simulated added error, likely because of a fundamental bias in the reconstruction approach. Conversely, the best age‐based decay model showed an 18–20% MAE and bias in reconstructed SWE. We conclude that remotely sensed albedos where available are superior to age‐based approaches in all aspects except simplicity.

show abstract

Section: Water Resources Researchmentioning

confidence: 99%

Section: Age-based Albedo Decay Modelmentioning

confidence: 99%

An Examination of Snow Albedo Estimates From MODIS and Their Impact on Snow Water Equivalent Reconstruction

Bair

Rittger

Skiles

et al. 2019

Water Resources Research

Self Cite

View full text Add to dashboard Cite

show abstract

“…These statistical models, which include a variety of approaches including regression models (such as multiple linear regression-MLR), binary regression trees, and lookup tables, have been applied using observations that span both larger (e.g., continental) scales (Bormann et al, 2013;Sturm et al, 1995Sturm et al, , 2010 and smaller (e.g., watershed) scales (Jonas et al, 2009;Wetlaufer et al, 2016). Other machine learning approaches such as Random Forests and Artificial Neural Networks have also become popular for estimating snow quantities (particularly SWE and snow cover) using a variety of input data including data from satellite sensors (e.g., Bair et al, 2018;Dobreva & Klein, 2011;Tedesco et al, 2004), land surface models (e.g., Snauffer et al, 2018), and ground observations (e.g., Tabari et al, 2010;Buckingham et al, 2015;Gharaei-Manesh et al, 2016). These approaches have been shown to be highly adaptable to capture nonlinear relationships involved in snow measurement (Czyzowska-Wisniewski et al, 2015), allowing them to outperform linear approaches such as MLR.…”

Section: Introductionmentioning

confidence: 99%

Improving Snow Water Equivalent Maps With Machine Learning of Snow Survey and Lidar Measurements

Broxton

Leeuwen

Biederman

2019

Water Resources Research

View full text Add to dashboard Cite

In the semiarid interior western USA, where a majority of surface water supply comes from mountain forests, high‐resolution aerial lidar‐based surveys are commonly used to study snow. These surveys provide rich information about snow depth, but they are usually not accompanied with spatially explicit measurements of snow density, which leads to uncertainty in the estimation of snow water equivalent (SWE). In this study, we use a novel approach to distribute ~300 field measurements of snow density with artificial neural networks. We combine the resulting density maps with aerial lidar snow depth measurements, bias corrected with a very large and precisely geolocated array of field‐measured snow depths (~4,000 observations), to create and validate maps of snow depth, snow density, and SWE over two sites along Arizona's Mogollon Rim in February and March 2017. These maps show differences between midwinter and late‐winter snow conditions. In particular, compared to that of snow depth, the spatial variability of snow density is smaller for the later snow survey than the earlier snow survey. These gridded data also show that the representativeness of Snow Telemetry and other point measurements is different for the midwinter and late‐winter snow surveys. Overall, the lidar artificial neural network SWE estimates can be as much as 30% different than if Snow Telemetry density were used with lidar snow depths to estimate SWE.

show abstract

“…From recent work (Bair et al, 2018b), we have shown that the SNOWPACK (Lehning et al, 2002a, b;Bartelt and Lehning, 2002) model is capable of accurate SWE prediction when supplied only with snow depth for precipitation as well as the other requisite forcings (i.e., radiation, snow albedo, temperatures, and wind speed). Over a 5-year period using hourly in situ measured energy balance forcings and a snow pillow for validation at a high-elevation site in the western US, the SWE modeled by the numerical snow cover model SNOWPACK showed a bias of −17 mm, or 1 % (Bair et al, 2018b). Likewise, the success of the Airborne Snow Observatory has demonstrated that given accurate snow depth measurements, SWE can be well modeled.…”

Section: Previous Work With Akah Snow Measurementsmentioning

confidence: 99%

Comparison of modeled snow properties in Afghanistan, Pakistan, and Tajikistan

et al. 2020

Self Cite

View full text Add to dashboard Cite

Ice and snowmelt feed the Indus River and Amu Darya in western High Mountain Asia, yet there are limited in situ measurements of these resources. Previous work in the region has shown promise using snow water equivalent (SWE) reconstruction, which requires no in situ measurements, but validation has been a problem. However, recently we were provided with daily manual snow depth measurements from Afghanistan, Tajikistan, and Pakistan by the Aga Khan Agency for Habitat (AKAH). To validate SWE reconstruction, at each station, accumulated precipitation and SWE were derived from snow depth using the numerical snow cover model SNOWPACK. High-resolution (500 m) reconstructed SWE estimates from the Parallel Energy Balance Model (ParBal) were then compared to the modeled SWE at the stations. The Alpine3D model was then used to create spatial estimates at 25 km resolution to compare with estimates from other snow models. Additionally, the coupled SNOWPACK and Alpine3D system has the advantage of simulating snow profiles, which provides stability information. The median number of critical layers and percentage of faceted layers across all of the pixels containing the AKAH stations were computed. For SWE at the point scale, the reconstructed estimates showed a bias of − 42 mm (−19 %) at peak SWE. For the coarser spatial SWE estimates, the various models showed a wide range, with reconstruction being on the lower end. A heavily faceted snowpack was observed in both years, but 2018, a dry year, according to most of the models, showed more critical layers that persisted for a longer period.

show abstract

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

Cited by 82 publications

References 71 publications

An Examination of Snow Albedo Estimates From MODIS and Their Impact on Snow Water Equivalent Reconstruction

An Examination of Snow Albedo Estimates From MODIS and Their Impact on Snow Water Equivalent Reconstruction

Improving Snow Water Equivalent Maps With Machine Learning of Snow Survey and Lidar Measurements

Comparison of modeled snow properties in Afghanistan, Pakistan, and Tajikistan

Contact Info

Product

Resources

About