Abstract: Artificial Neural Networks (ANNs) are classified as a data-driven technique, which implies that their learning improves as more and more training data are presented. This observation is based on the premise that a longer time series of training samples will contain more events of different types, and hence, the generalization ability of the ANN will improve. However, a longer time series need not necessarily contain more information. If there is considerable repetition of the same type of information, the ANN may not become “wiser”, and one may be just wasting computational effort and time. This study assumes that there are segments in a long time series that contain a large quantum of information. The reason behind this assumption is that the information contained in any hydrological series is not uniformly distributed, and it may be cyclic in nature. If an ANN is trained using these segments rather than the whole series, the training would be the same or better based on the information contained in the series. A pre-processing can be used to select information-rich data for training. However, most of the conventional pre-processing methods do not perform well due to large variation in magnitude, scale and many zeros in the data series. Therefore, it is not very easy to identify these information-rich segments in long time series with large variation in magnitude and many zeros. In this study, the data depth function was used as a tool for the identification of critical (information) segments in a time series, which does not depend on large variation in magnitude, scale or the presence of many zeros in data. Data from two gauging sites were used to compare the performance of ANN trained on the whole data set and just the data from critical events. Selection of data for critical events was done by two methods, using the depth function (identification of critical events (ICE) algorithm) and using random selection. Inter-comparison of the performance of the ANNs trained using the complete data sets and the pruned data sets shows that the ANN trained using the data from critical events, i.e., information-rich data (whose length could be one third to half of the series), gave similar results as the ANN trained using the complete data set. However, if the data set is pruned randomly, the performance of the ANN degrades significantly. The concept of this paper may be very useful for training data-driven models where the training time series is incomplete.
Abstract: Selecting the right model to simulate a specific watershed has always been a challenge, and field testing of watersheds could help researchers to use the proper model for their purposes. The performance of three popular Geographic Information System (GIS)-based watershed simulation models (European Hydrological System Model (MIKE SHE), Agricultural Policy/Environmental Extender (APEX) and Soil and Water Assessment Tool (SWAT)) were evaluated for their ability to simulate the hydrology of the 52.6 km2 Canagagigue Watershed located in the Grand River Basin in southern Ontario, Canada. All three models were calibrated for a four-year period and then validated using an independent four-year period by comparing simulated and observed daily, monthly and annual streamflow. The simulated flows generated by the three models are quite similar and closely match the observed flow, particularly for the calibration results. The mean daily/monthly flow at the outlet of the Canagagigue Watershed simulated by MIKE SHE was more accurate than that simulated by either the SWAT or the APEX model, during both the calibration and validation periods. Moreover, for the validation period, MIKE SHE predicted the overall variation of streamflow slightly better than either SWAT or APEX.
Abstract: This study examines the net snow accumulation and ablation characteristics and trends in the Inland Temperate Rainforest (ITR) of the Upper Fraser River Basin, British Columbia (BC), Canada. It intends to establish whether elevation and/or air temperature play(s) a dominant role in hydrological year peak snow water equivalent (SWE) and whether regional patterns emerge in the interannual variability in peak accumulation. To that end, SWE and air temperature data from seven snow pillow sites in the Upper Fraser River Basin at elevations ranging from 1118 to 1847 m above sea level are analyzed to infer snowpack characteristics and trends for hydrological years 1969–2012, with 2005–2012 being the actual period of data overlap. Average peak SWE ranges from 391.3 mm at Barkerville, BC on 16 April to 924.4 mm at Hedrick Lake, BC on 27 April. Snow cover duration lasts 206–258 days, with snow onset dates from mid-October to early November and snow off dates from late May to early July. Statistically-significant (p ≤ 0.05) cross correlations exist between peak SWE at nearly all sites, indicating regional coherence in seasonal synoptic activity across the study area. However, the lack of relationships between peak SWE and elevation as well as air temperature parameters indicate that mesoscale to local processes lead to distinct snow accumulation and ablation patterns at each site. Four sites with the longest records exhibit no trend in peak SWE values between 1990 and 2012. Changes to snowpack regimes may pose a threat to the productivity and immense biodiversity supported by the ancient western red cedar and hemlock stands growing in the wet toe slopes of the ITR. Thus, it is imperative that continued monitoring of snowpack conditions remains a top priority in the Upper Fraser River Basin, allowing for a better understanding of ecosystem changes in a warming climate.