[1] The science of forests and floods is embroiled in conflict and is in urgent need of reevaluation in light of changing climates, insect epidemics, logging, and deforestation worldwide. Here we show how an inappropriate pairing of floods by meteorological input in analysis of covariance (ANCOVA) and analysis of variance (ANOVA), statistical tests used extensively for evaluating the effects of forest harvesting on floods smaller and larger than an average event, leads to incorrect estimates of changes in flood magnitude because neither the tests nor the pairing account for changes in flood frequency. We also illustrate how ANCOVA and ANOVA, originally designed for detecting changes in means, do not account for any forest harvesting induced change in variance and its critical effects on the frequency and magnitude of larger floods. The outcomes of numerous studies, which applied ANCOVA and ANOVA inappropriately, are based on logical fallacies and have contributed to an ever widening disparity between science, public perception, and often land-management policies for decades. We demonstrate how only an approach that pairs floods by similar frequency, well established in other disciplines, can evaluate the effects of forest harvesting on the inextricably linked magnitude and frequency of floods. We call for a reevaluation of past studies and the century-old, preconceived, and indefensible paradigm that shaped our scientific perception of the relation between forests, floods, and the biophysical environment.

No abstract

Abstract. The L moments are used in the three stages of regional frequency analysis: the delineation of homogeneous regions, the identification of a regional parent distribution, and the estimation of distribution parameters. Numerical analysis is conducted on 5 min to 24 hours annual rainfall extremes from 375 precipitation gaging stations in Canada. The numerical analysis concluded that Canada could be considered as a single homogeneous region in which the L skewness and L kurtosis display no significant spatial variability. Also, on the basis of mean annual precipitation (MAP), Canada can be divided into climatologically homogeneous subregions, in which the L coefficient of variation is virtually constant. The parent distribution was identified as the general extreme value (GEV), the parameters of which depend on the MAP and storm duration. A hierarchical regional approach is proposed for fitting the identified GEV distribution, where the L skewness, L coefficient of variation, and mean are estimated on a regional, subregional, and single-site basis, respectively. Monte Carlo simulations indicate that design storms estimated by the proposed hierarchical approach are substantially more accurate than those estimated by the single-site method. The simulations also demonstrate that the proposed hierarchical approach makes the estimation of design storms at ungaged sites less dependent on the availability of precipitation data. While current procedures for estimating design storms in Canada are exclusively based on single-site frequency analysis, it has long been recognized that regional analysis techniques have the ability to significantly reduce uncertainties in quantile estimates relative to that inherent in the single-site approach [Lettenmaier et al., 1987;Pilon and Adamowski, 1992]. Regionalization procedures can further be considered equivalent to extending of the gaging network and provide planners and designers with a better alternative for the estimation of design storms at ungaged sites than currently used rainfall frequency maps. The advantages of regional frequency models and the large uncertainties involved in single-site approaches justify the need for a new regional methodology leading to more reliable design storm estimates at both short-term record sites and ungaged locations.The purpose of this paper is to develop a regional rainfall frequency approach for estimating design storms in Canada and to evaluate the accuracy of such approach relative to that of the current single-site frequency method. Linear Moment StatisticsThis study draws heavily on the linear moments or L moments [Hosking, 1990]. The L moments suffer less from the effects of sampling variability than the conventional moments, they are more robust to outliers in the data, and hence they enable more secure inferences to be made from samples about the underlying probability distribution [Royston, 1991]. The rth 31,645

No abstract

[1] A well-established precept in forest hydrology is that any reduction of forest cover will always have a progressively smaller effect on floods with increasing return period. The underlying logic in snow environments is that during the largest snowmelt events the soils and vegetation canopy have little additional storage capacity and under these conditions much of the snowmelt will be converted to runoff regardless of the amount or type of vegetation cover. Here we show how this preconceived physical understanding, reinforced by the outcomes of numerous paired watershed studies, is indefensible because it is rationalized outside the flood frequency distribution framework. We conduct a meta-analysis of postharvest data at four catchments (3-37 km 2 ) with moderate level of harvesting (33%-40%) to demonstrate how harvesting increases the magnitude and frequency of all floods on record (19-99 years) and how such effects can increase unchecked with increasing return period as a consequence of changes to both the mean (þ11% to þ35%) and standard deviation (À12% to þ19%) of the flood frequency distribution. We illustrate how forest harvesting has substantially increased the frequency of the largest floods in all study sites regardless of record length and this also runs counter to the prevailing wisdom in hydrological science. The dominant process responsible for these newly emerging insights is the increase in net radiation associated with the conversion from longwave-dominated snowmelt beneath the canopy to shortwave-dominated snowmelt in harvested areas, further amplified or mitigated by basin characteristics such as aspect distribution, elevation range, slope gradient, amount of alpine area, canopy closure, and drainage density. Investigating first order environmental controls on flood frequency distributions, a standard research method in stochastic hydrology, represents a paradigm shift in the way harvesting effects are physically explained and quantified in forest hydrology literature.Citation: Green, K. C., and Y. Alila (2012), A paradigm shift in understanding and quantifying the effects of forest harvesting on floods in snow environments, Water Resour. Res., 48, W10503,