Partitioning the causes of spatiotemporal variation in the sunny day sap flux density of a larch plantation on a hillslope in northwest China

Abstract Aims Tree sap flux density, as the basis for estimating forest transpiration, exhibits high spatiotemporal variation on hillslopes. This is largely due to the variation in the canopy leaf area index (LAI), potential evapotranspiration (PET), and volumetric soil moisture (VSM) that are induced by the slope position and shaded terrain effects. To improve the estimation of the daily sap flux density ( J s ) of sloped forests, a new approach is required that will extrapolate the J s measured at given slope position to other slope positions by considering both effects on the influencing factors. Method This study was carried out on a southeast-facing hillslope of a Larix principis-rupprechtti plantation in NW China. The J s was monitored during the growing season of 2016 on three plots located at the foot-slope (FS), mid-slope (MS) and upper-slope (US). To exclude interferences from fluctuating weather conditions, only data from 16 sunny days were used. The dynamic optimal J s was applied to quantify the J s response to changes in the PET, LAI and VSM induced by the slope position and shaded terrain effects. Result The mean daily J s (ml⋅cm −2 ⋅min −1 ) of 16 sunny days differed among slope positions as MS (0.059) > FS (0.050) > US (0.041). The strong shaded terrain effect at FS was verified by the markedly delayed starting time of sap flow at the FS when compared to that at the US and MS under a weak shaded terrain effect. The strong shaded terrain effect at the FS induced a PET reduction which decreased with the rising DOY until the end of July and then increased, this effect at FS also induced an LAI reduction that varied with the DOY inversely to PET. The mean J s (ml⋅cm −2 ⋅min −1 ) of 16 sunny days was reduced by 0.063 at US and 0.040 at MS mainly due to the slope position effect, and was reduced by 0.051 at the FS due to the joint impacts of a strong shaded terrain effect (0.0505) and a slope position effect (0.0005). When extrapolating the J s measured at the MS to estimate the J s at the US, the consideration of the slope position effect lowered the estimation error to 7.8% markedly decreased from 50.6% when not considering any effect. However, the J s estimation error at the FS increased from 25% to 87% when considering only the slope position effect, but decreased to 11.2% when considering both the shaded terrain effect and the slope position effect. Conclusion The substantial spatiotemporal variation in J s exists on hillslopes due to the varying effects of the slope position and the shaded terrain. The rising slope position leads to a decrease of PET, LAI, VSM and consequently, J s , while the strong shaded terrain effect at the lower slope (e.g., SF in this study) leads to a decrease in LAI, PET and J s . Thus, it is necessary to consider the slope position and shaded terrain effects, as these variables can greatly improve the J s estimation accuracy when scaling up from measured values at one slope position to other slope positions and even the entire hillslope.