Optimization of Snow-Related Parameters in Noah Land Surface Model (v3.4.1) Using Micro-Genetic Algorithm (v1.7a)

The snowfall prediction is important in winter and early spring because snowy conditions generate enormous economic damages. However, there is a lack of previous studies dealing with snow prediction, especially using land surface models (LSMs). Numerical weather prediction models directly interpret the snowfall events, whereas the LSMs evaluate the snow cover fraction, snow albedo, and snow depth through interaction with atmospheric conditions. When the initially-developed empirical parameters are local or inadequate, we need to optimize the parameter sets for a certain region. In this study, we seek for the optimal parameter values in the snow-related processes – snow cover fraction, snow albedo, and snow depth – of the Noah LSM, for South Korea, using the micro-genetic algorithm and the in-situ surface observations and remotely-sensed satellite data. Snow data from surface observation stations representing five land cover types – deciduous broadleaf forest, mixed forest, woody savanna, cropland, and urban and built-up lands – are used to optimize five snow-related parameters that calculate the snow cover fraction, maximum snow albedo of fresh snow, and the fresh snow density associated with the snow depth. Another parameter, reflecting the dependence of snow cover fraction on the land cover types, is also optimized. Optimization of these six snow-related parameters has led to improvement in the root-mean squared errors by 17.0 %, 8.2 %, and 5.6 % on snow depth, snow cover fraction, and snow albedo, respectively. In terms of the mean bias, the underestimation problems of snow depth and overestimation problems of snow albedo have been alleviated through optimization of parameters calculating the fresh snow by about 45.1 % and 32.6 %, respectively.

A novel Machine Learning-based gap-filling of fine-resolution remotely sensed snow cover fraction data by combining downscaling and regression

Satellite-based remotely-sensed observations of snow cover fraction (SCF) can have data gaps in spatially distributed coverage from sensor and orbital limitations. We mitigate these limitations in the example fine-resolution Moderate Resolution Imaging Spectroradiometer (MODIS) data by gap-filling using auxiliary 1-km datasets that either aid in downscaling from coarser-resolution (5 km) MODIS SCF wherever not fully covered by clouds, or else by themselves via regression wherever fully cloud-covered. This study’s prototype predicts a 1-km version of the 500 m MOD10A1 SCF target. Due to non-collocatedness of spatial gaps even across input and auxiliary datasets, we consider a recent gap-agnostic advancement of partial convolution in computer vision for both training and predictive gap-filling. Partial convolution accommodates spatially consistent gaps across the input images, effectively implementing a 2-dimensional masking. To overcome reduced usable data from non-collocated spatial gaps across inputs, we innovate a fully generalized 3-dimensional masking in this partial convolution. This enables a valid output value at a pixel even if only a single valid input variable and its value exist in the neighborhood covered by the convolutional filter zone centered around that pixel. Thus our gap-agnostic technique can use significantly more examples for training (~67%) and prediction (~100%), instead of only less than 10% for the previous partial convolution. We train an example simple 3-layer legacy Super-Resolution Convolutional Neural Network (SRCNN) to obtain downscaling and regression component performances that are better than baseline values of either climatology or MOD10C1 SCF as relevant. Our generalized partial convolution can enable multiple earth science applications like downscaling, regression, classification and segmentation that were hindered by data gaps.

Bias correction and downscaling of snow cover fraction projections from regional climate models using remote sensing for the European Alps

Mountain seasonal snow cover is undergoing major changes due to global climate change. Assessments of future snow cover usually rely on physical based models, and often include post-processed meteorology. Alternatively, here we propose a direct statistical adjustment of snow cover fraction from regional climate models by using long-term remote sensing observations. We compared different bias correction routines (delta change, quantile mapping, and quantile delta mapping) and explore a downscaling based on historical observations for the Greater Alpine Region in Europe. All bias correction methods adjust for systematic biases, for example due to topographic smoothing, and reduce model spread in future projections. Averaged over the study region and whole year, snow cover fraction decreases from 12.5 % in 2000–2020 to 10.4 (8.9, 11.5; model spread) % in 2071–2100 under RCP2.6, and 6.4 (4.1, 7.8) % under RCP8.5. In addition, changes strongly depended on season and altitude. The comparison of the statistical downscaling to a high-resolution physical based model yields similar results for the altitude range covered by the climate models, but different altitudinal gradients of change above and below. We found trend-preserving bias correction methods (delta change, quantile delta mapping) more plausible for snow cover fraction than quantile mapping. Downscaling showed potential but requires further research. Since climate model and remote sensing observations are available globally, the proposed methods are potentially widely applicable, but are limited to snow cover fraction only.

A Compilation of Snow Cover Datasets for Svalbard: A Multi-Sensor, Multi-Model Study

Reliable and accurate mapping of snow cover are essential in applications such as water resource management, hazard forecasting, calibration and validation of hydrological models and climate impact assessments. Optical remote sensing has been utilized as a tool for snow cover monitoring over the last several decades. However, consistent long-term monitoring of snow cover can be challenging due to differences in spatial resolution and retrieval algorithms of the different generations of satellite-based sensors. Snow models represent a complementary tool to remote sensing for snow cover monitoring, being able to fill in temporal and spatial data gaps where a lack of observations exist. This study utilized three optical remote sensing datasets and two snow models with overlapping periods of data coverage to investigate the similarities and discrepancies in snow cover estimates over Nordenskiöld Land in central Svalbard. High-resolution Sentinel-2 observations were utilized to calibrate a 20-year MODIS snow cover dataset that was subsequently used to correct snow cover fraction estimates made by the lower resolution AVHRR instrument and snow model datasets. A consistent overestimation of snow cover fraction by the lower resolution datasets was found, as well as estimates of the first snow-free day (FSFD) that were, on average, 10–15 days later when compared with the baseline MODIS estimates. Correction of the AVHRR time series produced a significantly slower decadal change in the land-averaged FSFD, indicating that caution should be exercised when interpreting climate-related trends from earlier lower resolution observations. Substantial differences in the dynamic characteristics of snow cover in early autumn were also present between the remote sensing and snow model datasets, which need to be investigated separately. This work demonstrates that the consistency of earlier low spatial resolution snow cover datasets can be improved by using current-day higher resolution datasets.

Evaluation of a Convection-Permitting Modeling of Precipitation over the Tibetan Plateau and Its Influences on the Simulation of Snow-Cover Fraction

Precipitation is a critical input to land surface and hydrology modeling and prediction. Dynamical downscale modeling has added value to representing precipitation, when compared with the performance of coarse-resolution reanalysis and global climate models, over the Tibetan Plateau (TP). Convection-permitting modeling (CPM) may even outperform dynamical downscale models (DDMs). In this study, 4-km CPM results were compared to 28-km DDM results for a snow season (1 October 2013–31 May 2014) over the TP. The CPM- and DDM-simulated precipitation, as well as three merged gridded precipitation datasets, were evaluated against in situ observations below 4800 m. The five precipitation datasets (CPM, DDM, CMFD, COPRPH, and TRMM) showed large differences over the TP with underestimation of TRMM and overestimation of CPM and DDM compared to observations. The most significant difference occurred in the Brahmaputra Grand Canyon. Given the substantial uncertainty in observed precipitation at high mountains, snow cover simulated by a high-resolution land data assimilation system was used to indirectly evaluate the above precipitation data using MODIS observations. Simulated snow-cover fraction was greatly underestimated using all the merged precipitation datasets. However, simulations using the DDM- and CPM-generated precipitation as input outperformed those using other gridded precipitation data, showing lower biases, higher pattern correlations, and closer probability distribution functions than runs driven by the merged precipitation. The findings of this study generally support the assumption that high-resolution CPM-produced precipitation has an added value for use in land surface and hydrology simulations in high-mountain regions without reliable in situ precipitation observations.

Biascorrected projections of snow cover fraction from EURO-CORDEX regional climate models with MODIS remote sensing observations for the European Alps

<p>Snow is a key environmental parameter in mountains, and in this changing climate reductions in snow are expected. Traditionally, future estimates of snow are based on dedicated snow/hydrological models forced by climate projections, which, however, are computationally intensive and which decouple hydrology from climate forcing. Recently, regional climate models (RCM) have been used as an alternative, although snow is only an auxiliary parameter in RCMs and not as accurately represented as compared to dedicated snow models. Nonetheless, RCMs encompass the climate-hydrology feedbacks, cover large areas, and have recently become available in moderate horizontal resolutions.</p><p>Here, we skip the need to biascorrect the input variables to the snow/hydrological models (i.e. temperature, precipitation, …) and use observations to directly biascorrect the target variable, i.e. snow cover. Quantile delta mapping (QDM), a trend preserving bias correction method, is used to correct biases in EURO-CORDEX RCMs that provide snow cover fraction as output (CCLM4-8-17, ALADIN63, WRF331F, WRF381P, RACMO22E, RCA4) using remote sensing observations of snow cover from MODIS for the European Alps. As such, snow cover biases were accounted for, which originated mostly from orographic mismatches as well as temperature and precipitation biases. Model ensemble means were calculated for two emission scenarios (rcp26 and rcp85; with 6 and 21 GCM-RCM combinations available). The biascorrected projections can be used to put the climate model projections into context of current observations thus facilitating interpretations.</p><p>These are results from CliRSnow, a project that aims at providing bias corrected and downscaled projections of snow cover for the whole Alpine region until 2100. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 795310.</p>

Unraveling driving forces explaining significant reduction in satellite-inferred Arctic surface albedo since the 1980s

The Arctic has warmed significantly since the early 1980s and much of this warming can be attributed to the surface albedo feedback. In this study, satellite observations reveal a 1.25 to 1.51% per decade absolute reduction in the Arctic mean surface albedo in spring and summer during 1982 to 2014. Results from a global model and reanalysis data are used to unravel the causes of this albedo reduction. We find that reductions of terrestrial snow cover, snow cover fraction over sea ice, and sea ice extent appear to contribute equally to the Arctic albedo decline. We show that the decrease in snow cover fraction is primarily driven by the increase in surface air temperature, followed by declining snowfall. Although the total precipitation has increased as the Arctic warms, Arctic snowfall is reduced substantially in all analyzed data sets. Light-absorbing soot in snow has been decreasing in past decades over the Arctic, indicating that soot heating has not been the driver of changes in the Arctic snow cover, ice cover, and surface albedo since the 1980s.