9:00 AM - 9:15 AM
[AHW18-01] Development of a model for estimating cypress plantation transpiration using satellite remote sensing - through sap flow observation data and forest inventory data -
★Invited Papers
Keywords:Satellite remote sensing, Evapotranspiration, Transpiration, Sap flow, Forest inventory
Evapotranspiration is one of the major components of the global hydrological cycle, of which canopy transpiration (Et) accounts for a large proportion of total evapotranspiration; wide-area observations are essential to obtain a complete picture of Et, but are difficult to achieve due to the huge amount of time and labour required for actual wide-area measurements. Satellite remote sensing is a very effective method for estimating Et because it provides information over a wide area and at low cost.
In previous studies, numerous models have been developed for Et estimation. The main challenges of these are that 1. it is influenced by the type and condition of vegetation in areas of vegetation cover; 2. it is difficult to accurately distinguish between transpiration and evaporation in total evapotranspiration models; unlike other evaporation factors, Et is not only influenced by meteorological factors such as temperature and solar radiation, but also by vegetation condition, It is also influenced by the amount of photosynthesis. Developing an Et estimation model that takes these factors account is expected to make it possible to estimate Et with high accuracy and over a wide area.
The aim of this study was to develop an Et estimation model for Japanese cypress plantations, considering 1. meteorological and 2. plant physiological effects, using satellite remote sensing. Sap flow data from cypress plantations observed in Tochigi, Kanagawa and Fukuoka prefectures were used as the measured values for Et. Satellite images were taken by Landsat 5, 7 and 8 and data from the same period as the sap flow observations were used. To assess meteorological effects, air density was calculated from surface temperatures derived from satellite images and the relationship with sap flow was investigated. The results showed a high correlation between air density - sap flow rate from January to March. This relationship was used to define the meteorologically influenced Et rate. Seasonal variations in visible and near-infrared reflectance in cypress forests were investigated using forest inventory data (around the Kiryu River dam, Gunma Prefecture) showing the distribution of tree species, and Landsat5, 7 and 8 images from 2011 to 2018. The results suggest that near-infrared is sensitive to leaf area and normalized indices using visible red and visible green are sensitive to photosynthesis. The estimated leaf area and photosynthetic rate were used to define the amount of Et that is affected by plant physiology. These two factors may vary in the proportion they affect depending on the season, i.e. the degree of their influence may vary depending on the estimated daily temperature. In order to establish the proportion of each element affected, an index of surface temperature on a given day between 0 and 1 was developed, with the minimum and maximum surface temperatures at the sap flow measurement sites set to 0 and 1 respectively. This indicator established the percentage affected by meteorological and the percentage affected by plant physiology. Additionally, sapwood area, which has a significant influence on Et, was estimated using DBH or tree age and incorporated into the model.
The results showed a correlation of r = 0.74-0.88 between the measured sap flow and the model estimates. The accuracy was improved to r = 0.76-0.89 by correcting the model using rainfall data immediately before the satellite image was taken.
The model developed in this study has the following advantages: it allows Et estimation using only satellite imagery and forest inventory, is less sensitive to regional climatic differences, considers vegetation activity, and allows estimation of the contribution of Et to total evapotranspiration.
In previous studies, numerous models have been developed for Et estimation. The main challenges of these are that 1. it is influenced by the type and condition of vegetation in areas of vegetation cover; 2. it is difficult to accurately distinguish between transpiration and evaporation in total evapotranspiration models; unlike other evaporation factors, Et is not only influenced by meteorological factors such as temperature and solar radiation, but also by vegetation condition, It is also influenced by the amount of photosynthesis. Developing an Et estimation model that takes these factors account is expected to make it possible to estimate Et with high accuracy and over a wide area.
The aim of this study was to develop an Et estimation model for Japanese cypress plantations, considering 1. meteorological and 2. plant physiological effects, using satellite remote sensing. Sap flow data from cypress plantations observed in Tochigi, Kanagawa and Fukuoka prefectures were used as the measured values for Et. Satellite images were taken by Landsat 5, 7 and 8 and data from the same period as the sap flow observations were used. To assess meteorological effects, air density was calculated from surface temperatures derived from satellite images and the relationship with sap flow was investigated. The results showed a high correlation between air density - sap flow rate from January to March. This relationship was used to define the meteorologically influenced Et rate. Seasonal variations in visible and near-infrared reflectance in cypress forests were investigated using forest inventory data (around the Kiryu River dam, Gunma Prefecture) showing the distribution of tree species, and Landsat5, 7 and 8 images from 2011 to 2018. The results suggest that near-infrared is sensitive to leaf area and normalized indices using visible red and visible green are sensitive to photosynthesis. The estimated leaf area and photosynthetic rate were used to define the amount of Et that is affected by plant physiology. These two factors may vary in the proportion they affect depending on the season, i.e. the degree of their influence may vary depending on the estimated daily temperature. In order to establish the proportion of each element affected, an index of surface temperature on a given day between 0 and 1 was developed, with the minimum and maximum surface temperatures at the sap flow measurement sites set to 0 and 1 respectively. This indicator established the percentage affected by meteorological and the percentage affected by plant physiology. Additionally, sapwood area, which has a significant influence on Et, was estimated using DBH or tree age and incorporated into the model.
The results showed a correlation of r = 0.74-0.88 between the measured sap flow and the model estimates. The accuracy was improved to r = 0.76-0.89 by correcting the model using rainfall data immediately before the satellite image was taken.
The model developed in this study has the following advantages: it allows Et estimation using only satellite imagery and forest inventory, is less sensitive to regional climatic differences, considers vegetation activity, and allows estimation of the contribution of Et to total evapotranspiration.