资源环境科技发展态势分析平台

A number of studies using the so-called space-for-time substitution (STS) approach have predicted that climate warming of snow-influenced watersheds would result in alarming increases in long-term ET and concurrent decreases in runoff. This empirical approach implicitly assumes that any spatial covariation of factors across the landscape would remain fixed in time, that is, the "covariance-stationarity" assumption. A better understanding of this assumption, in terms of what it means and how it affects predictions, is needed to assess the implications of STS predictions. To address this, we examined the influence of the covariance-stationarity assumption on STS predictions of long-term ET in response to warming. We did this by simulating changes in ET resulting from elevational transfers of factors in ways that emulate STS. This approach was applied to the upper San Joaquin River, a snow-influenced watershed in the Sierra Nevada, USA. Predicted increases in ET from STS including the covariance-stationarity assumption were 3-7 times greater than STS treating air temperature as an independent factor, over most of the higher elevations. Based on results from a factorial analysis, these differences were attributed primarily to the pedologic gradient across elevation, which transitions from loamy Alfisol soils at lower elevations to sandy Entisol/Inceptisol soils and exposed bedrock at upper elevations. The covariance-stationarity assumption does not seem realistic for soil and air temperature over timescales of projected climate. Therefore, the confounding of these factors in the STS approach can substantially overestimate the sensitivity of long-term ET to air temperature.