Pii: S0022-1694(00)00336-x

Trends in ̄ood and low ̄ows in the US were evaluated using a regional average Kendall's S trend test at two spatial scales and over two timeframes. Field signi®cance was assessed using a bootstrap methodology to account for the observed regional cross-correlation of stream ̄ows. Using a 5% signi®cance level, we found no evidence of trends in ̄ood ̄ows but did ®nd evidence of upward trends in low ̄ows at the larger scale in the Midwest and at the smaller scale in the Ohio, the north central and the upper Midwest regions. A dramatically different interpretation would have been achieved if regional cross-correlation had been ignored. In that case, statistically signi®cant trends would have been found in all but two of the low ̄ow analyses and in two-thirds of the ̄ood ̄ow analyses. We show that the cross-correlation of ̄ow records dramatically reduces the effective number of samples available for trend assessment. We also found that low ̄ow time series exhibit signi®cant temporal persistence. Even when the serial correlation was removed from the time series, signi®cant trends in low ̄ow series were apparent, though the number of signi®cant trends decreased. q 2000 Elsevier Science B.V. All rights reserved.