Maxim Neumann

Assessing Performance of L- and P-Band Polarimetric Interferometric SAR Data in Estimating Boreal Forest Above-Ground Biomass

Biomass estimation performance using polarimetric interferometric synthetic aperture radar (PolInSAR) data is evaluated at L- and P-band frequencies over boreal forest. PolInSAR data are decomposed into ground and volume contributions, retrieving vertical forest structure and polarimetric layer characteristics. The sensitivity of biomass to the obtained parameters is analyzed, and a set of these parameters is used for biomass estimation, evaluating one parametric and two non-parametric methodologies: multiple linear regression, support vector machine, and random forest. The methodology is applied to airborne SAR data over the Krycklan Catchment, a boreal forest test site in northern Sweden. The average forest biomass is 94 tons/ha and goes up to 183 tons/ha at forest stand level (317 tons/ha at plot level). The results indicate that the intensity at HH-VV is more sensitive to biomass than any other polarization at L-band. At P-band, polarimetric scattering mechanism type indicators are the most correlated with biomass. The combination of polarimetric indicators and estimated structure information, which consists of forest height and ground-volume ratio, improved the root mean square error (rmse) of biomass estimation by 17%-25% at L-band and 5%-27% at P-band, depending on the used parameter set. Together with additional ground and volume polarimetric characteristics, the rmse was improved up to 27% at L-band and 43% at P-band. The cross-validated biomass rmse was reduced to 20 tons/ha in the best case. Non-parametric estimation methods did not improve the cross-validated rmse of biomass estimation, but could provide a more realistic distribution of biomass values.

Review of Machine Learning Approaches for Biomass and Soil Moisture Retrievals from Remote Sensing Data

The enormous increase of remote sensing data from airborne and space-borne platforms, as well as ground measurements has directed the attention of scientists towards new and efficient retrieval methodologies. Of particular importance is the consideration of the large extent and the high dimensionality (spectral, temporal and spatial) of remote sensing data. Moreover, the launch of the Sentinel satellite family will increase the availability of data, especially in the temporal domain, at no cost to the users. To analyze these data and to extract relevant features, such as essential climate variables (ECV), specific methodologies need to be exploited. Among these, greater attention is devoted to machine learning methods due to their flexibility and the capability to process large number of inputs and to handle non-linear problems. The main objective of this paper is to provide a review of research that is being carried out to retrieve two critically important terrestrial biophysical quantities (vegetation biomass and soil moisture) from remote sensing data using machine learning methods.

Impacts of Spatial Variability on Aboveground Biomass Estimation from L-Band Radar in a Temperate Forest

Estimation of forest aboveground biomass (AGB) has become one of the main challenges of remote sensing science for global observation of carbon storage and changes in the past few decades. We examine the impact of plot size at different spatial resolutions, incidence angles, and polarizations on the forest biomass estimation using L-band polarimetric Synthetic Aperture Radar data acquired by NASA’s Unmanned Aerial Vehicle Synthetic Aperture Radar (UAVSAR) airborne system. Field inventory data from 32 1.0 ha plots (AGB 0.5 ha), suggesting a stability of field-estimated biomass at scales of about 1.0 ha. UAVSAR backscatter was linked to the field estimates of aboveground biomass to develop parametric equations based on polarized returns to accurately map biomass over the entire radar image. Radar backscatter values at all three polarizations (HH, VV, HV) were positively correlated with field aboveground biomass at all four spatial scales, with the highest correlation at the 1.0 ha scale. Among polarizations, the cross-polarized HV had the highest sensitivity to field estimated aboveground biomass (R2 = 0.68). Algorithms were developed that combined three radar backscatter polarizations (HH, HV, and VV) to estimate aboveground biomass at the four spatial scales. The predicted aboveground biomass from these algorithms resulted in decreasing estimation error as the pixel size increased, with the best results at the 1 ha scale with an R2 of 0.67 (p < 0.0001), and an overall RMSE of 44 Mg·ha−1. For AGB < 150 Mg·ha−1, the error reduced to 23 Mg·ha−1 (±15%), suggesting an improved AGB prediction below the L-band sensitivity range to biomass. Results also showed larger bias in aboveground biomass estimation from radar at smaller scales that improved at larger spatial scales of 1.0 ha with underestimation of −3.62 Mg·ha−1 over the entire biomass range.

Detection of Durable and Permanent Changes in Urban Areas Using Multitemporal Polarimetric UAVSAR Data

Change detection using synthetic aperture radar (SAR) data is useful in emergency situations and unfavorable weather conditions. In this letter, change detection using multitemporal polarimetric Uninhabited Aerial Vehicle SAR data is investigated in an urban environment. The most robust polarimetric parameters are determined, and change detection techniques using a maximum likelihood ratio and a hyperbolic tangent model function are applied to the selected parameter. The model function was introduced to quantify the change characteristics and to rule out seasonal changes or those related to mobile features, and thus to only detect durable and permanent changes in urban environments. A comparison of results with historical Google Earth images showed a good level of agreement. Fitting of the hyperbolic tangent function to the multitemporal polarimetric parameters significantly reduces the false detection rate and indicates whether a building was constructed or destroyed, as well as when the detected changes occurred.

Parametric and non-parametric forest biomass estimation from PolInSAR data

Biomass estimation performance from model-based polarimetric SAR interferometry (PolInSAR) using generic parametric and non-parametric regression methods is evaluated at L- and P-band frequencies over boreal forest. PolInSAR data is decomposed into ground and volume contributions, estimating vertical forest structure, and using a set of obtained parameters for biomass regression. The considered estimation methods include multiple linear regression, support vector machine and random forest. The biomass estimation performance is evaluated on DLRs airborne SAR data at L- and P-bands over Krycklan Catchment, a boreal forest test site in Northern Sweden. The combination of polarimetric indicators and estimated structure information has improved the root mean square error (RMSE) of biomass estimation up to 28% at L-band and up to 46% at P-band. The cross-validated biomass RMSE was reduced to 20 tons/ha.

Validation of the new SRTM digital elevation model (NASADEM) with ICESAT/GLAS over the United States

A new version of the digital elevation model (DEM) generated from Shuttle Radar Topography Mission (SRTM) data is to begin release in 2016. The so-called NASADEM results from re-processing the raw radar echoes and telemetry, guided by global measurements of topography from the ICESats Geoscience Laser Altimeter System (GLAS). Significant improvements in accuracy were obtained thanks to the removal of large-scale systematic biases due to a variety of arte-facts ranging from residual boom oscillations to the presence of vegetation.

Biomass change in disturbed, secondary, and primary tropical forests from TanDEM-X

The time variation of phase height from interferometric SAR (InSAR) from TanDEM-X is shown for 3 years, in Tapajos National Forest, Brazil. Its RMS, for one secondary stand, about a model linear in time is 0.5 m. This RMS is compared to that for 30 stands at one epoch. The single-epoch RMS for a model linear in mean field height is 2.2 m. It is suggested that the improved performance of the temporal variation may be due to errors in finding the phase height of the ground, which is necessary for single-epoch estimation, but not needed for the “change” measurement. Pending further fieldwork, a tentative conversion of 20 Mg/ha/yr corresponding to 1 m/yr, is proposed. Abrupt discontinuities, as well as phase height rates as a function of stand age/aboveground biomass, are discussed.

Polarimetric Backscatter Optimization for Biophysical Parameter Estimation

In this letter, we introduce a polarization optimization concept to maximize the sensitivity of the synthetic aperture radar (SAR) backscatter measurements to a biophysical parameter. An iterative method based on Lagrangian multipliers is introduced for optimization. Using a priori information, the optimization identifies the polarization most sensitive (or least sensitive) to the quantity of interest, with best predictive characteristics. The methodology is tested for estimating forest aboveground biomass using polarimetric SAR data acquired by DLRs E-SAR airborne sensor at L- and P-band frequencies over a boreal forest test site in Krycklan Catchment, Sweden. The results show an improvement of sensitivity to forest biomass using the optimized polarization compared to canonical polarizations. Via polarization basis transformation, the correlation of biomass to backscatter is shown to improve by up to 0.23 and 0.59 at L- and P-bands, respectively.

Quantifying spatial and temporal dynamics of tropical forest structure using high resolution airborne lidar

This paper analyzes the spatial and temporal dynamics of forest structure in tropical wet forest at the La Selva Biological Station, Costa Rica. Three small-footprint lidar datasets from 1997, 2006, and 2009 are used over old-growth and secondary forests at 1-meter spatial resolution. The results quantify the spatial variability, the vertical forest structure profiles and the temporal variability, demonstrating the determination of the forest succession stages from remote sensing data.

Polarimetric optimization for biophysical parameter estimation

This paper explores polarization optimization of synthetic aperture radar (SAR) data, constrained to maximize some biophysical data conditions, such as the residual sum of squares (RSS). Based on a-priori information, the optimization helps to identify the polarization most sensitive (and least sensitive) to the quantity of interest, with best predictive characteristics. In this presentation, the relationship of SAR polarimetry to forest biomass is considered. The results are presented over boreal forest using DLRs airborne E-SAR sensor data at L- and P-band frequencies.