We investigated climate change in the annual snow cover period in mountainous areas of central Japan by downscaling simulations of four climate change projections based on a Coupled Model Intercomparison Project phase 3 (CMIP3) Special Report on Emission Scenarios (SRES) A1B emission scenario using a regional climate model. Our numerical simulation reproduced well the observed snow depths and areas of snow cover. The projected snow disappearance date in all areas occurred earlier in the future climate due to global warming and were substantially earlier in areas of both light and heavy snowpacks in the present climate. The time shift was smaller at areas where the present-day maximum snow depth is around 100 cm and the snow disappearance date is in mid-April. These projected changes in the duration of snow cover were associated with decreasing snowfall and accelerated snowmelt due to increasing surface air temperatures. The effect was interpreted using an idealized model of temporal variation in surface air temperature. Earlier snowmelt causes local enhancement of surface air temperature increases that will have considerable impact on mountain ecosystems.
Spatial interpolation methods can be used to estimate high density air temperature data to drive the temperature index model used to simulate snowmelt processes. Thus, evaluating the impact of different spatial temperature interpolation methods on snowmelt simulations is necessary. This study creates three air temperature datasets based on different methods for a data sparse basin. These datasets include: 1) an inverse distance weighting (IDW) method; 2) an improved IDW method considering the elevation influence on temperature; and 3) combined use of linear regression and MODIS Land Surface Temperature (LST) data. The datasets are verified at observation stations and applied to a snowmelt hydrologic model using the Soil Water Assessment Tool. The simulation results are compared with observed discharge data and uncertainties discussed. Verification at the observation stations indicates that all datasets can reflect station air temperature. Model simulations and uncertainty analysis show that the dataset created by combined use of linear regression and MODIS LST data achieved the best simulation results and smallest uncertainties. The results also indicate that this dataset can accurately and stably reflect the spatial variation of air temperature compared with other data.
Variation in sapwood thickness (TS) around the circumference of a stem and the presence of intermediate wood can cause errors in the estimation of sapwood area (AS) and individual tree-scale transpiration (Q) based on the sap flow technique. We measured bark thickness (TB), TS, and intermediate wood thickness (TI) in 16 orthogonal directions for wood discs from 57 Japanese cedar (Cryptomeria japonica) trees, and evaluated the impact of variation in TS around the circumference of a stem and the validity of assuming a constant TI for AS estimates. The coefficient of variation of TS in the 16 directions was 5.3–36.7%. AS based on four orthogonal directions resulted in only minor errors for all trees, although this was not the case for AS based on one and two directions. The mean TI over the 16 directions was not significantly correlated to the diameter. If TI was assumed to be constant at the median value for our forest (=0.8 cm), the relative absolute errors for six of the 57 trees exceeded 30%. If these errors are unacceptable, we recommend extracting stem discs to measure TI when estimating AS and Q for trees with intermediate wood having a similar color to sapwood.