Sensitivity Analysis of Net Primary Production Estimation using a Semi-empirical BRDF Model and Reflectance Observed by RC Helicopter for Japanese Cedar Forest

摘要

全地球広域的に観測を行う衛星に搭載されるセンサーでは視野角の幅が広く，観測された分光反射率データには二方向性反射率（BRDF：Bi-directional Reflectance Distribution Function）影響があることが知られている。全球植生純一次生産量（NPP：Net Primary Production）を求め，その精度を改善するためには，BRDFモデルを利用して，NPPの推定におけるBRDFの影響を評価する必要がある。本研究ではスギ林に着目し，無人ヘリコプターによりBRF（Bidirectional Reflectance Distribution Factors）観測を行った。観測で得られたBRDFデータより，BRDFモデルの適用性を検証した上で，パラメータを決定し，モデルから計算した反射率より杉スギ林のNPPを推定した。その結果は0.36kgCO_2/m~2/monthとなり，森林調査の結果0.32kgCO_2/m~2/monthと一致した。%Among the existing global observing sensors, such as Terra/MODIS, NOAA/AVHRR, ADEOS-Ⅱ/ GLI, with a large field of view, there is a need to take Bidirectional Reflectance Distribution Function (BRDF) effect into account when using data of these sensors to analyze. For estimating the global NPP (Net Primary Production) from the spectral data of global observing sensors, and improving the accuracy of this estimation method simultaneously, it is necessary to know how this method was affected by BRDF effect, especially in the sensor's observation conditions. Using a Japanese cedar forest as the objective and a semi-empirical kernel-driven BRDF model (the RossThick-LiSparse model), this study analyzed BRDF effects on NPP estimation. The BRDF data of the forest used in this study were measured by a sensor onboard a radio-controlled helicopter with bidirectional reflectance factors (BRF) observations in July 2002. After validating the application of Ross-Li BRDF model to Japanese cedar forest, parameters of this model for the Japanese cedar forest were obtained, and used to retrieve reflectance for the nadir view and nadir illumination. With the retrieved reflectance, for the cedar forest, NPP estimations were affected by BRDF effects of approximately 11% under the GLI simulated observation conditions. This study also sought to validate the NPP estimation algorithm based on the pattern-decomposition-based vegetation index (VIPD) and photosynthetically active radiation (PAR) for multi-spectral sensor data. From the retrieved reflectance, the NPP of the study forest was estimated to be 0.36 kgCO_2/m~2/month. For comparison, forest surveys at the same study site have been conducted since the BRF observation. Using the ground-measured data, NPP was estimated to be 0.32 kgCO_2/m~2/month, which is in agreement with the preliminary result.