The University of British Columbia

The primary focus of my research is on the processes on climate-glacier interface: quantitative assessment of glacier mass balance (balance between snow accumulation and snow and ice melt) by numerical modeling. My approach includes modeling of recent and future glacier mass balance on local, regional and global scales. Specifically, my research covers:

• Modeling of ice and snow melt (using self-developed mass balance model)

• Investigating the response of glacier melt to future climate change on global scale, and glacier contribution to sea level rise

• Evaluating the performance of Global Climate Models (GCMs) and statistically downscaling their output to be used in regional impact studies and glacier modeling

• Investigating and evaluating statistical scaling methods to assess glacier volume changes

Brief description of my PhD research

The aim of my PhD project was to model the future sea level rise from the melt of mountain glaciers and ice caps. The rise of global sea level is expected to continue over the coming decades having a major impact on coastal cities, deltaic lowlands, small islands, and coastal ecosystems. Melting and disintegrating mountain glaciers have been identified as the second largest contributor to rising sea level after thermal expansion of the oceans.

I have modeled the changes in glacier volume on global scale, based on temperature and precipitation scenarios as defined by various GCMs and translated these changes into sea level changes. The research consisted of the following steps: First step was the development and calibration of glacier mass balance model that simulates glacier accumulation, ablation and refreezing. Several variants of mass balance models were evaluated in sensitivity studies at one glacier (Radić and Hock, 2006; Hock et al., 2007). For the global assessment the mass balance model was calibrated with available mass balance observations world-wide and forced with gridded climate data (reanalysis products and climatologies). Multiple regressions between the calibrated model parameters and variables from the gridded climate data were then applied. The result was a set of transfer functions that allowed me to assign parameter values to any glacier in the world based on their climatic setting. The model was then applied to each individual mountain glacier and ice cap available in the recently updated and extended World Glacier Inventory (WGI-XF). The total volume of all mountain glaciers and ice caps has been assessed in Radić and Hock (2010) using statistical upscaling methods and volume-area scaling. To quantify future volume changes I have run the calibrated mass balance model for all glaciers in WGI-XF with downscaled monthly 21st century temperature and precipitation from ten GCMs, based on the mid-range greenhouse emission scenario A1B. As glaciers lose mass due to climate warming they retreat and hence their hypsometry changes. I used volume-area-length scaling (previously compared with an ice-flow model in Radić and Hock, 2007; 2008) to account for these changes and their feedbacks to glacier mass balance, allowing receding glaciers to approach a new equilibrium in a warming climate. The results, i.e. spatially distributed contribution of glaciers to future sea level rise are presented in Radić and Hock (2011).

Brief description of my Postdoctoral research

My current research is part of a broader collaborative research project of Western Canadian Cryospheric Network (WC2N). Our glaciology group at UBC is modeling the past, recent and future behavior of all glaciers in South-western Canada and Northern Washington State. The future of these glaciers may have major implications for the regional hydrology. We also collaborated with a research unit of BC Hydro in projecting the future of glaciers in Columbia Basin. We developed regional glaciation model consisting of the following integral parts: terrain data preparation (deriving regional glacier bed topography from digital elevation data and ice masks using the inverse modeling), climate representation (use of reanalysis products to represent the past and recent climate and GCMs for the future scenarios), downscaling the gridded climate data to glacier scale using a variety of statistical and downscaling methods, and development of the glacier mass balance model coupled with glacier ice-flow model for the whole domain.

My part of the research in our glaciology team is focused on:

  (1) evaluation of GCMs to be applied in future projections of glacier mass balance,

  (2) developing the methods of downscaling of GCMs for the glacier modeling, and

  (3) modeling future glacier volume changes.  

The point (1) is addressed in Radić and Clarke (2011) by evaluating the performance of 22 GCMs with the range of different performance measures/metrics, comparing modeled versus observed climatologies, and ranking the models according to their demonstrated ability to reproduce observed features of present-day climate in the region of interest. Since our application is in glacier modeling, the evaluation of models is focused on climate variables important for the processes on the climate-glacier interface. I used two different approaches for this analysis: (1) the statistical metrics, such as root-mean-square errors computed for several climate variables and (2) Self-Organizing Maps (SOMs; type of unsupervised Artificial Neural Network) for climate classification and model validation.

After running the regional glaciation model with the reanalysis product (here we use North American Regional Reanalysis, NARR) we create the mass balance fields on high resolution (200m) for the region of interest. To project future mass balance we run the model with downscaled GCMs. As an addition to the simple downsaling method ('delta approach') I developed an alternative method based on the variant of ‘weather typing’, i.e. linking the synoptic patters over the region to the local climate variables of interest (in our case: surface air temperature and precipitation). 

For the point (3) we investigate how projections of glacier changes in western Canada (ice covered portions of Yukon, Alberta and British Columbia) vary when using models of different complexities. The complex model is a radiation-indexed degree-day melt model coupled with a 2-D ice flow model and applied region-wide with a resolution of 200 m (Anslow et al., in preparation). The bed topography is provided by an inversion algorithm (Clarke et al., in preparation). The model with lower complexity combines a degree-day melt model with the volume-area scaling, and is applied to each individual glacier in the domain. The models are forced with a range of 21st century climate scenarios from six GCMs, whose climate fields are statistically downscaled with two different methods: 'delta-approach' and 'weather-typing'.