Search

Accurate forecasting of runoff as an important hydrological variable is a key task for water resources planning and management. Given the importance of this variable, in the current study, a multivariate linear stochastic model (MLSM) is combined with a multilayer nonlinear machine learning model (MNMLM) to generate a hybrid model for the spatial and temporal simulation of runoff in the Quebec basin, Canada. Monthly hydrological data from 2001 to 2013, including precipitation and runoff data from nine stations and Normalized Difference Vegetation Index (NDVI) extraction of MODIS data, are applied as input to the proposed hybrid model. At the first step of the hybrid modeling, data normality and stationary were examined by performing various tests. In the second step, MLSM was developed by defining four different scenarios and as a result 15 sub-scenarios. The first and second scenarios were developed based on one exogenous variable (precipitation or NDVI). In contrast, the second and third scenarios were developed based on two additional variables. In the first and third scenarios, the data are modeled without preprocessing. In the second and fourth scenarios, a preprocessing step is performed on the data. Then, in the third step, various combinations based on different time delays from runoff data were applied for developing nonlinear model. The comparisons are made between observed and simulated time series at various stations and based on the root mean squared error (RMSE), mean absolute error (MAE), correlation coefficient (R) and Akaike information criterion (AIC). The efficiency of the proposed hybrid model is compared with a novel machine learning model that was introduced in 2021 by Sultani et al., and it was also compared with the results obtained from the linear and nonlinear models. In most stations, delays (t-1) and (t-24) are identified as the most effective delays in hybrid and nonlinear modeling of runoff. Also, in most stations, the use of climatic parameters and physiographic factors as exogenous variables along with runoff data improves the results compared to the use of one variable. Results showed that at all stations, proposed hybrid model generally leads to more accurate estimates of runoff compared with various linear and nonlinear models. More accurate estimates of peak runoff values at all stations were another excellence of proposed hybrid model than other models. [ABSTRACT FROM AUTHOR]

The active layer thickness (ALT) is a key parameter for permafrost studies. Changes in the ALT are affected mainly by air and ground temperatures, physical and thermal properties of the surface and subsurface materials, soil moisture, vegetation, and the duration and thickness of snow cover. Ground penetrating radar (GPR) and electrical resistivity tomography (ERT) were employed across a snow fence during the thawing season to delineate and monitor the active layer of permafrost in Cambridge Bay, Nunavut, Canada. The variation of the ALT is well captured by the high-resolution time-lapse radargram. At the position of the fence, the active layer thickens over the thawing period from 0.5 m depth at the beginning to 1.0 m depth at the end. The active layer is thicker in the pre-fence area (C zone) than in the post-fence area (H zone). As the air temperature increases with time, the difference in thickness between the two zones decreases, eventually becoming almost equal. Changes in the ALT are represented in the ERT by low resistivities (< 200 Ωm), which decrease gradually with time. This occurs most significantly in the H zone due to the rapidly increasing temperature in the absence of snow cover. The electrical resistivity structure of the active layer is well correlated with the vegetation activity, as measured by the normalized difference vegetation index, air/ground temperatures, soil moisture, snow cover, and snow accumulation controlled by the fence. Geophysical data interpretation and correlation schemes with vegetation and meteorological data explored in this paper can be applied to monitor the active layer, which is expected to thin during the freezing season. [ABSTRACT FROM AUTHOR]

The analysis of a series of five normalized difference vegetation index (NDVI) images produced information about a Labrador (Canada) portion of the tundra-taiga interface. The twenty-five year observation period ranges from 1983 to 2008. The series composed of Landsat, SPOT and ASTER images, provided insight into regional scale characteristics of the tundra-taiga interface that is usually monitored from coarse resolution images. The image set was analyzed by considering an ordinal classification of the NDVI to account for the cumulative effect of differences of near-infrared spectral resolutions, the temperature anomalies, and atmospheric conditions. An increasing trend of the median values in the low, intermediate and high NDVI classes is clearly marked while accounting for variations attributed to cross-sensor radiometry, phenology and atmospheric disturbances. An encroachment of the forest on the tundra for the whole study area was estimated at 0 to 60 m, depending on the period of observation, as calculated by the difference between the median retreat and advance of an estimated location of the tree line. In small sections, advances and retreats of up to 320 m are reported for the most recent four- and seven-year periods of observations. [ABSTRACT FROM AUTHOR]

There is an ongoing need for a long-term Grizzly Bear Management Plan in Yoho National Park as a result of human-bear conflicts and periodic human mortality. Avalanche slopes offer preferred feeding sites to the grizzly bear population, and hence a more detailed knowledge of their location and extent is needed as a management tool. Landsat Thematic Mapper (TM) satellite image bands, a normalized difference vegetation index (NDVI), and the principal component associated with "greenness" (derived from the TM image data), along with digital elevation model variables provided the best classification with an avalanche slope accuracy of 82 %. The classified data were imported into a geographic information system (GIS) and the area of the avalanche slopes determined to be 66 km2 or approximately 5% of the total park area. The resulting GIS database is being used by the park for management purposes. [ABSTRACT FROM AUTHOR]

Standardized principal components analysis (SPCA) is performed on normalized difference vegetation index (NDVI) time series of a 100 latitude by 100 longitude area of western Canada including grassland, parkland, and forest ecosystems. When used with input from a single growing season (April-October), early components correspond closely to ecoregional mapping based on long-term vegetation composition. Later components isolate areas showing agricultural practice or climatic stress particular to the year. When three years' growing seasons are input together into a multiyear SPCA, similar patterns occur in the first component. However, components occur early in the series that discriminate areas having different seasonality patterns associated with plant phenology. Deciduous-dominated areas are well distinguished from grasslands. Multiyear SPCA includes an early component apparently related to latitudinal variation in day length, which does not appear in single-year component series. An early component unrelated to any known geographical or climatological pattern or event appears, which may relate to sensor degradation. Both single-year and multiyear SPCA isolate NDVI variability in higher numbered components that is limited in space and/or in time. This allows interpretation of transient or localized events using detailed local data, separating them from regional trends that occur in earlier (lower numbered) components of the series. These results demonstrate the possibility of refining ecoregion mapping based on selected early components to incorporate actual interannual variability for selected time periods as well as the long-term stable elements of biogcoclimatic regions. Longer time series could potentially quantify observed unidirectional trends resulting from climate change. [ABSTRACT FROM AUTHOR]

Field studies have shown that dense tree canopies and regular tree arrangements reduce noise from a point source. In urban areas, noise sources are multiple and tree arrangements are rarely dense. There is a lack of data on the association between the urban tree canopy characteristics and noise in complex urban settings. Our aim was to investigate the spatial variation of urban tree canopy characteristics, indices of vegetation abundance, and environmental noise levels. Using Light Detection and Ranging point cloud data for 2015, we extracted the characteristics of 1,272,069 public and private trees across the island of Montreal, Canada. We distinguished needle-leaf from broadleaf trees, and calculated the percentage of broadleaf trees, the total area of the crown footprint, the mean crown centroid height, and the mean volume of crowns of trees that were located within 100m, 250m, 500m, and 1000m buffers around 87 in situ noise measurement sites. A random forest model incorporating tree characteristics, the normalized difference vegetation index (NDVI) values, and the distances to major urban noise sources (highways, railways and roads) was employed to estimate variation in noise among measurement locations. We found decreasing trends in noise levels with increases in total area of the crown footprint and mean crown centroid height. The percentages of increased mean squared error of the regression models indicated that in 500m buffers the total area of the crown footprint (29.2%) and the mean crown centroid height (12.6%) had a stronger influence than NDVI (3.2%) in modeling noise levels; similar patterns of influence were observed using other buffers. Our findings suggest that municipal initiatives designed to reduce urban noise should account for tree features, and not just the number of trees or the overall amount of vegetation., (Copyright © 2021 Elsevier Inc. All rights reserved.)