Methodology & Results

Results

Two types of melt models are being tested and compared: temperature index and energy balance. Results show strong agreement between the two models. Work continues within specific “calibration basins” (Figure 1). These are sub-basins for which we have sufficient river discharge data that can be used to calibrate and validate the melt models. Model runs for the calibration basins (Figure 2) are being used to tune the model for application across the primary basin containing the calibration basin. Daily melt volume is partitioned across three sources: snow on land, snow on glacier ice and exposed glacier ice. Finally, the contribution to river discharge from snow and ice for each of the five primary CHARIS basins is being determined by model runs that have been tuned using that basin’s calibration basin. Partitioning results currently are available for the Indus basin, example in Figure 3, and anticipated for Syr Darya and Amu Darya by the end of November 2016. Results for the Ganges and Brahmaputra are anticipated by December 2016. Model results for all basins will cover the time period 2001-2016 at a daily time interval.  Results to date will be presented at the 2016 American Geophysical Union (AGU) fall meeting in San Francisco.

calibration-basins

Figure 1. Before running a melt model for major basins the model is tuned to get the right discharge for calibration basins for which we have adequate runoff, temperature and rainfall data. Credit: NSIDC

Example of daily and monthly partitioning results from the Hunza calibration basin

Figure 2. Example of daily and monthly partitioning results from the Hunza calibration basin. Credit: NSIDC

indus-partition-results-copy

Figure 3. Partition results for the Indus. Credit: NSIDC

Methodology

To determine the contribution of glacier ice and seasonal snow to runoff, we are applying a suite of satellite remote sensing, reanalysis and ground-based data as input to specific snow and ice melt models. Gridded maps of snow and glacier area/elevation are used as input to a temperature-index melt model. A separate energy balance model is also being tested. The melt models estimate runoff from snow covered grid cells, based on cell area and melt depth. Glacier melt is estimated in the same way, once snow has disappeared from glacierized grid cells. The temperature-index melt model is driven by daily mean temperature from reanalysis data. In order to improve the CHARIS melt models we continue to analyze optimal sources for air temperature, rainfall and evapo-transpiration (ET) data to be used for model input. This includes the comparison of available weather station temperature data with air temperature data obtained by downscaling from a selection of reanalysis data sets.

We are also evaluating the accuracy of the melt model results using innovative isotopic and geochemical tracers to identify and quantify the sources of water (ice melt, snow melt, rainfall and ground water) flowing into selected rivers that represent the major hydro-climates of the study area. In collaboration with our Asian partners, we are assessing the performance of the various melt models. Results of this study will be applied to future efforts to assess the social-economic impacts of water uses and their vulnerability to changes in flow magnitude and timing.

Satellite Products and Processing

We use data from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor to map glacierized areas and to map seasonal snow cover.

The MODIS Persistent Ice (MODICE) algorithm makes use of the time series of fractional snow and ice cover from the MODIS Snow covered Area and Grain size (MODSCAG) algorithm. MODICE produces a consistently-derived map of annual minimum exposed snow and ice. For the CHARIS project, MODICE annual maps of glacierized areas from 2000 to the present are used as input to a temperature-index melt model.

Figure 4. MODICE minimum annual ice and snow extent, 2010, Upper Indus basin.
Credit: NSIDC

A temporally and spatially continuous daily time series of snow cover extent is being produced using MODIS MOD10A1 daily snow cover maps. Temporal and spatial persistence filters are used to fill missing and cloud covered cells in daily snow maps. Maps of snow cover extent are used as input to a temperature-index melt model.

Elevation Data

Elevation data (Figure 5) are essential for modeling seasonal snow and glacier ice melt. Elevation defines the basins and water courses of a landscape and the exposure of the surface to direct and reflected solar radiation and other meteorological conditions. Accurate elevation data are difficult to obtain in the mountainous, largely inaccessible regions of this study. For this reason we must rely on digital elevation models (DEMs) generated from satellite data. The available data sources—interferometric synthetic aperture radar, satellite image stereoscopy and spaceborne laser altimetry—have shortcomings that make it necessary to combine the best features of each to obtain an optimal DEM.

An example of MODIS mean monthly area-elevations for seasonal snow cover in the Upper Indus Basin

Figure 5. An example of MODIS mean monthly area-elevations for seasonal snow cover in the Upper Indus Basin. Credit: NSIDC

The Role of Reanalysis

Near-surface and air temperature data from atmospheric reanalyses are used to drive melt and hydrological models. Rather than selecting a single best reanalysis product, we are using a suite of products, including but not limited to the three current state-of-the-art reanalyses: the NASA Modern Era Retrospective Analysis for Research and Applications (MERRA), the European Center for Medium Range Weather Forecasts (ECMWF) ERA-Interim and the National Centers for Environmental Prediction (NCEP) Climate Forecast System Reanalysis (CFSR).

River Discharge Data

River discharge data from stream gauging stations are being used for model validation and streamflow analysis. Because this study focuses on the contribution to streamflow from melting glacier ice and seasonal snow, we are using streamflow data from gauging stations that define sub-basins that include the maximum extent of seasonal snow cover and all glacier cover for each of the five major drainage basins. Additional gauging stations nested within these larger sub-basins have been selected so that contributions from highly glacierized and snow covered basins can be evaluated.

Water Sample Isotope Analysis

The CHARIS project objectives include quantifying glacier and snow-covered areas from remote sensing data and modeling the melt contribution of these hydrologic components to streamflow (hydrograph separation). In addition we are using innovative isotopic and geochemical tracers to identify and quantify the sources of water (ice melt, snow melt, rainfall and ground water) flowing into selected rivers representing the major hydro-climates of the study area. In glacier, snow and rain-fed basins, these three sources must be differentiated from each other and from the contribution of groundwater to baseflow. We are evaluating the accuracy of melt model results using these isotopic and geochemical tracers that identify and quantify water sources.