Towards ground-truthing of spaceborne estimates of above-ground life biomass and leaf area index in tropical rain forests

Abstract

Abstract. The canopy height h of forests is a key variable which can be obtained using air-or spaceborne remote sensing techniques such as radar interferometry or LIDAR. If new allometric relationships between canopy height and the biomass stored in the vegetation can be established this would offer the possibility for a global monitoring of the above-ground carbon content on land. In the absence of adequate field data we use simulation results of a tropical rain forest growth model to propose what degree of information might be generated from canopy height and thus to enable ground-truthing of potential future satellite observations. We here analyse the correlation between canopy height in a tropical rain forest with other structural characteristics, such as above-ground life biomass (AGB) (and thus carbon content of vegetation) and leaf area index (LAI) and identify how correlation and uncertainty vary for two different spatial scales. The process-based forest growth model FORMIND2.0 was applied to simulate (a) undisturbed forest growth and (b) a wide range of possible disturbance regimes typically for local tree logging conditions for a tropical rain forest site on Borneo (Sabah, Malaysia) in South-East Asia. In both undisturbed and disturbed forests AGB can be expressed as a power-law function of canopy height h (AGB = a ·h b ) with an r 2 ∼ 60% if data are analysed in a spatial resolution of 20 m × 20 m (0.04 ha, also called plot size). The correlation coefficient of the regression is becoming significant better in the disturbed forest sites (r 2 = 91%) if data are analysed hectare wide. There seems to exist no functional dependency between LAI and canopy height, but there is also a linear correlation (r 2 ∼ 60%) between AGB and the area fraction ofCorrespondence to: P. Köhler (peter.koehler@awi.de) gaps in which the canopy is highly disturbed. A reasonable agreement of our results with observations is obtained from a comparison of the simulations with permanent sampling plot (PSP) data from the same region and with the large-scale forest inventory in Lambir. We conclude that the spaceborne remote sensing techniques such as LIDAR and radar interferometry have the potential to quantify the carbon contained in the vegetation, although this calculation contains due to the heterogeneity of the forest landscape structural uncertainties which restrict future applications to spatial averages of about one hectare in size. The uncertainties in AGB for a given canopy height are here 20−40% (95% confidence level) corresponding to a standard deviation of less than ± 10%. This uncertainty on the 1 ha-scale is much smaller than in the analysis of 0.04 ha-scale data. At this small scale (0.04 ha) AGB can only be calculated out of canopy height with an uncertainty which is at least of the magnitude of the signal itself due to the natural spatial heterogeneity of these forests.

Mapping forest stand age in China using remotely sensed forest height and observation data

“…Statistical analysis has been a common method of fitting the biomass-height relationship. Widely used regression models include power [Chave et al, 2005;Köhler and Huth, 2010;Mitchard et al, 2011;Saatchi et al, 2011;Wang et al, 2013], linear [Fang et al, 2006;Skowronski et al, 2007], the quadratic polynomial [Lefsky et al, 2005b], and exponential functions [Yu et al, 2010]. Previous studies have shown that biomass-height relationships differ greatly across environmental gradients and among different forest types [Drake et al, 2003;Pan et al, 2004;Wang et al, 2013].…”

The carbon fluxes in different successional stages: modelling the dynamics of tropical montane forests in South Ecuador

“…Normally, these methods can only be used at small spatial scales. Even when a greater number of trees is monitored, the risk of over-or underestimating the regional carbon balance due to heterogeneity in tropical forests (Köhler and Huth 2010) or other up scaling effects remains.…”

Characterizing 3D vegetation structure from space: Mission requirements

“…The remote sensing component of the analysis 326 framework relies on lidar samplers and radar and passive optical imagers to sample and 327 map landscape vegetation spectral and spatial "metrics" at high spatial resolution (~25m). 328To relate the lidar and radar measures or "metrics" to in situ measures, lidar and radar 329 measures are then regressed against in situ timber height and volume measures in sample 330 plots (Kohler and Huth, 2010). The resulting regression equations are used to convert 331 landscape level lidar and radar metrics into regional, contiguous biomass and 3D 332 vegetation structure products.…”

“…To relate the lidar and radar measures or "metrics" to in situ measures, lidar and radar 329 measures are then regressed against in situ timber height and volume measures in sample 330 plots (Kohler and Huth, 2010). The resulting regression equations are used to convert 331 landscape level lidar and radar metrics into regional, contiguous biomass and 3D 332 vegetation structure products.…”