Impact of Leaf Area Index from Various Sources on Estimating Gross Primary Production in Temperate Forest Using JULES Land Surface Model

Abstract

We analyzed the effect of various types of LAI on simulation of gross primary production (GPP) in evergreen needleleaf and deciduous broadleaf temperate forests by LSM named Joint UK Land Environment Simulator (JULES). Over the period 2015-2017, LAI was observed by LAI-2200 plant canopy analyzer. Sources of estimated LAI included enhanced vegetation index (EVI) of two satellites with different temporal and spatial resolutions each other (Landsat and MODIS), LAI product of MODIS, and phenology model of JULES. These various LAIs were compared with in-situ observed one, and their effect on GPP simulation was compared with simulated GPP using in-situ observed LAI. As a result, JULES phenology model could not adequately express LAI of both deciduous broadleaf and evergreen needleleaf temperate forests. This misrepresentation of LAI was improved by simple modification, but there remained other limitations such as inaccurate representation of interannual variation. Consequently, GPP estimation based on phenology model LAI differed by up to 38% than that base on observed LAI. LAI from remote sensing EVI was relatively corresponded to observed one. Whereas MODIS LAI product with relatively low spatial resolution (500 m) could not perceive complex heterogeneity of temperate forest, and it provided 31% higher maximum value. Simulated GPP based on satellite-based LAI showed little difference (about 3.0%) in the coniferous forest and up to 8.9% difference in the broadleaf forest. Our study illuminates limitations of phenology model and necessity to improve phenology model in both deciduous broadleaf and evergreen needleleaf temperate forests. Spatial resolution of satellite which is used to retrieve LAI input was not influential when simulating GPP using LSM. On the other hand, it is recommended to use LAI data with sufficiently high spatial resolution so that it can reflect forest cover type of the site well.