Satellite based analysis of northern ET trends and associated changes in the regional water balance from 1983 to 2005

Summary

We developed an evapotranspiration (ET) algorithm driven by satellite remote sensing inputs, including AVHRR GIMMS NDVI, MODIS land cover and NASA/GEWEX solar radiation and albedo, and regionally corrected NCEP/NCAR Reanalysis daily surface meteorology. The algorithm was used to assess spatial patterns and temporal trends in ET over the pan-Arctic basin and Alaska from 1983 to 2005. We then analyzed associated changes in the regional water balance defined as precipitation (P) minus ET, where monthly P was defined from Global Precipitation Climatology Project (GPCP) and Global Precipitation Climatology Center (GPCC) sources. Monthly ET results derived from both in situ meteorological measurements and coarse resolution model reanalysis inputs agreed well (RMSE = 5.1–6.3 mm month⁻¹; R² = 0.91–0.92) with measurements from eight independent flux towers representing regionally dominant land cover types. ET showed generally positive trends over most of the pan-Arctic domain, though negative ET trends occurred over 32% of the region, primarily in boreal forests of southern and central Canada. Generally positive trends in ET, P and available long-term river discharge measurements imply that the pan-Arctic terrestrial water cycle is intensifying despite uncertainty in regional P and associated water balance estimates. Increasing water deficits in eastern Alaska, Canadian Yukon and western Prairie Provinces, and Northern Mongolia agree with regional drought records and recent satellite observations of vegetation browning and productivity decreases. Our results indicate that the pan-Arctic water balance is responding to a warming climate in complex ways with direct links to terrestrial carbon and energy cycles.

Introduction

Hydrological and ecological processes are strongly coupled in northern high latitude boreal forest and Arctic tundra regions. Precipitation and snowmelt influence plant-available moisture during the growing season, which impacts water, energy and carbon cycles through vegetation canopy controls on transpiration, plant-atmosphere exchanges of water vapor and carbon dioxide, and the partitioning of net radiant energy into sensible and latent heat fluxes. Substantial changes in hydrological and ecological processes have been observed in the northern high latitudes with persistent climatic warming over the past century (Trenberth et al., 2007), while further warming is projected to continue (Meehl et al., 2007). Recent hydrological changes observed for boreal forest and tundra regions include changes in seasonality and magnitude of river discharge, thawing and degradation of permafrost and associated rapid gains/losses in lake area, increasing soil active layer depths, and receding glaciers (e.g. McClelland et al., 2006, Oelke et al., 2004, White et al., 2007, Zhang et al., 2005). Ecological responses to these changes include earlier onset and lengthening growing seasons, increased vegetation structure and growth, and alteration of land–atmosphere CO₂ exchange (McDonald et al., 2004, Oechel et al., 2000, Sturm et al., 2001, Zhang et al., 2007a). A growing body of evidence indicates that changes in the regional water cycle are impacting ecological processes in this domain. Recent, post-2000 occurrences of widespread and persistent drought have occurred across the boreal high latitudes (Schindler and Donahue, 2006, Zhang et al., 2008), resulting in a general decrease in vegetation productivity for boreal forest indicated by atmospheric CO₂ anomalies (Angert et al., 2005), satellite NDVI and vegetation productivity records (Goetz et al., 2005, Zhang et al., 2008), and stand inventory data (Hogg et al., 2008, Schindler and Donahue, 2006).

According to the well-known Clausius–Clapeyron relationship between air temperature and water vapor capacity of the atmosphere, warming is expected to promote increases in evapotranspiration and precipitation leading to a general intensification (i.e., acceleration) of the global water cycle (Held and Soden, 2000, Meehl et al., 2007). Huntington (2006) reviewed the literature regarding historical trends in hydrologic variables and suggested an ongoing intensification of the global water cycle. Because the northern high-latitudes have experienced substantial warming during the last decades, the water cycle in these regions may also be intensifying. Serreze et al. (2006) and White et al. (2007) showed that the major components of the Arctic freshwater cycle including precipitation, runoff and precipitable water have generally experienced positive trends consistent with an intensifying water cycle. However, the relative magnitudes and spatial heterogeneity of these trends are highly uncertain due to sparse measurement networks and large natural variability in seasonal to annual weather patterns for the region.

Better understanding of recent changes in the northern high latitude terrestrial water cycle requires improved documentation of spatial patterns and changes of its primary components including precipitation, evapotranspiration and river discharge. Precipitation and river discharge measurements are currently available from pan-Arctic observation networks (e.g. Yang et al., 2005; McClelland et al., 2006). Evapotranspiration (ET) is highly heterogeneous both spatially and temporally due to strong vegetation canopy control on transpiration. However, there are very few direct measurements of ET over global land areas. The ET fields from GCM-based reanalyses such as ERA-40 and NCEP/NCAR Reanalysis (NNR) are not considered reliable because they are not constrained by precipitation and radiation but by circulation and temperature observations (Betts et al., 2003, Ruiz-Barradas and Nigam, 2005). ET plays an important role in linking the water, energy and carbon cycles and represents over 60% of precipitation over the global land area (L’vovich and White, 1990). Soil evaporation and plant transpiration are determined by surface meteorology and plant biophysics. Relatively sparse measurements of these variables for the northern high latitudes make accurate assessment of ET a challenge. Remotely sensed data, especially from polar-orbiting satellites, provide relatively frequent and spatially contiguous monitoring of surface biophysical variables affecting ET, including albedo, biome type and vegetation density. Satellite-based ET products have been produced at regional and global scales with varying accuracy (e.g. Bastiaanssen et al., 1998, Cleugh et al., 2007, Mu et al., 2007, Fisher et al., 2008). However, currently there is no continuous, long-term satellite based ET record for the northern high latitudes. The Normalized Difference Vegetation Index (NDVI) from the NOAA Advanced Very High Resolution Radiometer (AVHRR) record extends from 1981 to present and can be used to estimate regional patterns and trends in ET for the pan-Arctic domain since NDVI is sensitive to vegetation structure and photosynthetic canopy cover.

We developed an ET algorithm mainly driven by satellite remote sensing inputs to estimate regional patterns, monthly to annual rates and recent (23-year) trends in ET for the pan-Arctic basin and Alaska based on lessons learned from previous studies and driven by long-term AVHRR Global Inventory Modelling and Mapping Studies (GIMMS) NDVI records (Pinzon et al., 2005, Tucker et al., 2005) and bias-corrected surface meteorology reanalysis inputs. We then applied these results with available precipitation records, and river discharge measurements to assess recent changes in the pan-Arctic terrestrial water balance. We also compared these results to satellite based vegetation productivity records to determine whether observed water balance changes are consistent with recent trends in vegetation productivity for the region. The primary objectives of this paper are to (1) assess spatial patterns and temporal trends in ET and precipitation over the northern high latitudes from 1983 to 2005; (2) determine whether these results and associated changes in the regional water balance are consistent with an intensification (acceleration) of the terrestrial water cycle, and (3) assess whether the regional water balance changes are consistent with satellite based records of a greening arctic and recent drought-induced declines in boreal forest productivity.