Water, Volume 12, Issue 12

Anthology ID:
Bib Export formats:

pdf bib
CliGAN: A Structurally Sensitive Convolutional Neural Network Model for Statistical Downscaling of Precipitation from Multi-Model Ensembles
Chiranjib Chaudhuri | Colin Robertson

Despite numerous studies in statistical downscaling methodologies, there remains a lack of methods that can downscale from precipitation modeled in global climate models to regional level high resolution gridded precipitation. This paper reports a novel downscaling method using a Generative Adversarial Network (GAN), CliGAN, which can downscale large-scale annual maximum precipitation given by simulation of multiple atmosphere-ocean global climate models (AOGCM) from Coupled Model Inter-comparison Project 6 (CMIP6) to regional-level gridded annual maximum precipitation data. This framework utilizes a convolution encoder-dense decoder network to create a generative network and a similar network to create a critic network. The model is trained using an adversarial training approach. The critic uses the Wasserstein distance loss function and the generator is trained using a combination of adversarial loss Wasserstein distance, structural loss with the multi-scale structural similarity index (MSSIM), and content loss with the Nash-Sutcliff Model Efficiency (NS). The MSSIM index allowed us to gain insight into the model’s regional characteristics and shows that relying exclusively on point-based error functions, widely used in statistical downscaling, may not be enough to reliably simulate regional precipitation characteristics. Further use of structural loss functions within CNN-based downscaling methods may lead to higher quality downscaled climate model products.

pdf bib
Historical and Projected Changes to the Stages and Other Characteristics of Severe Canadian Prairie Droughts
Barrie Bonsal | Lu Zhuo | Elaine Wheaton | Ronald E. Stewart

Large-area, long-duration droughts are among Canada’s costliest natural disasters. A particularly vulnerable region includes the Canadian Prairies where droughts have, and are projected to continue to have, major impacts. However, individual droughts often differ in their stages such as onset, growth, persistence, retreat, and duration. Using the Standardized Precipitation Evapotranspiration Index, this study assesses historical and projected future changes to the stages and other characteristics of severe drought occurrence across the agricultural region of the Canadian Prairies. Ten severe droughts occurred during the 1900–2014 period with each having unique temporal and spatial characteristics. Projected changes from 29 global climate models (GCMs) with three representative concentration pathways reveal an increase in severe drought occurrence, particularly toward the end of this century with a high emissions scenario. For the most part, the overall duration and intensity of future severe drought conditions is projected to increase mainly due to longer persistence stages, while growth and retreat stages are generally shorter. Considerable variability exists among individual GCM projections, including their ability to simulate observed severe drought characteristics. This study has increased understanding in potential future changes to a little studied aspect of droughts, namely, their stages and associated characteristics. This knowledge can aid in developing future adaptation strategies.