Evaluation of remote-sensing-based models of gross primary productivity over Indian sal forest using flux tower and MODIS satellite data. Issue 18 (17th September 2017)

• Record Type: Journal Article
• Authors: Watham, T.; Patel, N. R.; Kushwaha, S. P. S.; Dadhwal, V. K.; Kumar, A. Senthil
• Is Part Of: International journal of remote sensing. Volume 38:Issue 18(2017)
• Page Start: 5069
• Page End: 5090
• Publication Date: 2017-09-17
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2017.1333653
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Forest plays a significant role in regulating the carbon budget and mitigating climate change in long term. However, lack of spatially explicit and accurate information on carbon exchange components from diverse forest ecosystem types in India limits carbon budgeting on a regional scale. Remote-sensing-driven ecosystem models are well-established tools for estimating gross primary productivity (GPP) over large areas but they are seldom found erroneous if implemented without proper calibration of biome-specific parameters. The present study evaluates the combined use of eddy covariance (EC) data and satellite-derived variables for estimating GPP over large areas. Four remote-sensing-driven models, (i) temperature–greenness (TG) model, (ii) greenness–radiation (GR) model, (iii) light use efficiency (LUE) model, and (iv) remote-sensing-based LUE (LUERS ) model, were parameterized with EC measurements and compared with 8-day Moderate Resolution Imaging Spectroradiometer (MODIS) GPP products for a moist Shorea robusta forest in northern part of India. EC observed 8-day average GPP varied from 5.38 to 12.42 g C m −2 day −1 . Among the four tested models, TG model had the highest root mean square error (RMSE) of 1.28 g C m −2 day −1, while GR and LUERS models had moderate RMSE of 0.99 g C m −2 day −1 and 0.98 g C m −2 day −1, respectively. The closest GPP estimate was given by LUE model with RMSE of 0.93 g C m −2 day −1. The RMSE for all four models were four times lower than that of MODIS GPP. Lower maximum LUE (ε m a x ) and uncertainty in the environmental scalar used in MODIS GPP algorithm could have contributed to higher RMSE. More accurate modelling of GPP can help in better understanding of forest ecological functions with the changing climate.