Estimating boreal forest species type with airborne polarimetric synthetic aperture radar

We have applied a non-parametric classifier (k nearest neighbour) to two calibrated orthogonal passes of airborne polarimetric synthetic aperture radar (POLSAR) image data over boreal forest for the purpose of discriminating canopy tree species of predefined stands. We found that a single classifier based on a single feature space (i.e. one set of POLSAR variables for all species) was less accurate than a hierarchical two-stage classifier that used different POLSAR variables for each species. We designed a two-stage classifier that first grouped stands into broad classes: pine, spruce and deciduous, and then classified each sample within the broad classes into individual species. We found that the most effective feature spaces had two or three dimensions. The two-stage classifier attained overall accuracies of between 60% and 75%.

We provide a first use of an equivalency test applied to remote-sensing classification. We use Lloyd's test of equivalency to find equivalent classifiers and thus infer informative POLSAR variables. The POLSAR variables that were most informative varied between the two passes and between the various elements of the hierarchical classifier. For the initial three-class classifier the most informative POLSAR variables were the two circular polarization ratios, several of Touzi's Stokes vector variables, HHVV coherence, several texture measures such as the variance of several scattering coefficients and the order parameter of the K-distribution and characteristics of the polarization signature pedestal. These results demonstrate that C-band POLSAR has great potential for mapping boreal forest cover either on its own or in concert with other geospatial data.     

Reflectance characteristics of arctic tundra vegetation from airborne radiometry

Abstract

The objectives of this study were to (1) develop and test an airborne radiometer system for sampling spectral radiances over arctic tundra landscapes, (2) quantify variations in the Normalized Difference Vegetation Index (NDVI) as a function of tundra vegetation type and seasonal date, and (3) quantify variations of spectral radiance and NDVI in response to possible anthropogenic disturbance. Spectral radiometer data analyzed in this study were acquired over the northern foothills of the Brooks Range in Alaska. The radiometer was well suited to data acquisition in the arctic environment. Mean NDVI values for four major vegetation types during each of three seasonal dates were significantly different from each other. Seasonal trajectories of the NDVI for each vegetation type corresponded to general phenological characteristics of the vegetation. The spectral radiance in red and near infrared wavebands appears to respond to disturbance caused by dust deposition from the Dalton Highway.     

Radar response of periodic vegetation canopies

Vector radiative transfer theory is used to model the scattered intensity from a layer of randomly oriented particles over a periodic rough surface. To account for the periodic nature of row-structured vegetation, the number density of particles within the layer is assumed to be varying periodically in the horizontal direction. Using Fourier series expansions and orthogonality properties, the radiative transfer equation is solved for the transformation matrix relating the incident and scattered intensities, from which the backscattering coefficient of the layer can be computed for any incidence direction and polarization configuration. The experimental component of this investigation consisted of radar observations at 1–5,4–75, and 9–5 GHz made by a truck-mounted system for a field of corn under three conditions: (a) full, which means that the corn plants were in their natural state, (b) defoliated, which was accomplished by stripping off the leaves and removing them, thereby leaving behind only bare vertical stalks, and (c) bare soil, which corresponds to the soil surface after having removed the stalks. The soil surface is modelled as a composite consisting of a deterministic periodic component and a random roughness component. A two-scale polarimetric scattering model is formulated and used to compare with the experimental observations. Excellent agreement between theory and measurements is realized as a function of both incidence and azimuth angles at all three microwave frequencies. The canopy model was then applied to the corn canopy under the two other conditions: stalks alone and full canopy. The model results were compared with radar backscatter measurements made for each of three look directions, including perpendicular and parallel to the row direction and at 45° relative to the row direction. For the stalk canopy, it was observed that the quasi-periodic arrangement of the stalks within the row enhances the backscatter at L-band when looking perpendicular to the row direction, which is attributed to a coherent-scattering effect associated with the stalks. A heuristic approach is used to model the quasi-periodic structure of the stalks by deriving a coherency factor which multiplies the first-order radiative solution for randomly located stalks. A similar coherency factor was also introduced for the leaves of the full canopy. The modified model was found to provide good agreement with experimental observations at L-band for all polarizations and at all look directions.     

Comparison of Envisat radar and airborne laser altimeter measurements over Arctic sea ice

Sea ice thickness is a crucial, but very undersampled cryospheric parameter of fundamental importance for climate modeling. Advances in satellite altimetry have enabled the measurement of sea ice freeboard using satellite microwave altimeters. Unfortunately, validation of these new techniques has suffered from a lack of ground truth measurements. Therefore, an airborne campaign was carried out in March 2006 using laser altimetry and photo imagery to validate sea ice elevation measurements derived from the Envisat/RA-2 microwave altimeter.We present a comparative analysis of Envisat/RA-2 sea ice elevation processing with collocated airborne measurements collected north of the Canadian Archipelago. Consistent overall relationships between block-averaged airborne laser and Envisat elevations are found, over both leads and floes, along the full 1300 km aircraft track. The fine resolution of the airborne laser altimeter data is exploited to evaluate elevation variability within the RA-2 ground footprint. Our analysis shows good agreement between RA-2 derived sea ice elevations and those measured by airborne laser altimetry, particularly over refrozen leads where the overall mean difference is about 1 cm. Notwithstanding this small 1 cm mean difference, we identify a larger elevation uncertainty (of order 10 cm) associated with the uncertain location of dominant radar targets within the particular RA-2 footprint. Sources of measurement uncertainty or ambiguity are identified, and include snow accumulation, tracking noise, and the limited coverage of airborne measurements.     

Diurnal patterns of bidirectional vegetation indices for wheat canopies

The perpendicular vegetation index (PVI) and normalized difference vegetation index (NDVI) were calculated from Mark II radiometer RED (0.63-0.69 μm) and NIR (0.76–0.90 μ) bidirectional radiance observations for wheat canopies. Measurements were taken over the plant development interval flag leaf expansion to watery ripeness of the kernels during which the leaf area index (LAI) decreased from 40 to 2-5. Spectral data were taken on four cloudless days five times (09.30, 11.00, 12.30, 14.00 and 15.30 hours (central standard time, C.S.T.) at five view zenith, Zv (0, 15, 30,45 and 60°) and eight view azimuth angles relative to the Sun, A_v (0, 45, 90, 135, 180, 225, 270 and 315°). The PVI was corrected to a common solar irradiance (PVIC) based on simultaneously observed insolation readings.

The PVIC at nadir view (?=0°) increased as (l/cosZ_s) increased on all the measurement days whereas the NDVI changed little as solar zenith angle (Z_s) changed. Thus, the PVIC responded to increasing path length through the canopy, or the number of leaves encountered, as solar zenith angle changed whereas the NDVI, which has saturated by the time an LAI of 2 was achieved, was nonresponsive.

Off-nadir PVIC ratioed to nadir PVIC increased as the view zenith and solar zenith angles increased (reciprocity in Sun and view angles), and as the view azimuth, A angle approached the Sun position (back scattering stronger that forwardscattering). In contrast, the DNVI was very stable for all view and solar angles consistent with saturation in its response. Even though the PVI is subject to bidirectional effects, it contains more useful information about wheat canopies at LAI > 2 than does the NDVI. The NDVI of the plant canopies changed rapidly at low vegetative cover but its bidirectional sensitivity at low LAI was not investigated.     

Investigations of the hydrobiological situation in Lake Onega using joint spaceborne radar,airborne and in situ measurements

The results of the first comprehensive limnological experiment conducted in Lake Onega in the summer of 1989 are presented. The concurrent spacebornc observations (Cosmos-l870 SAR), airborne surveys (visual observations and IR radiometry), and shipborne observations (visual observations, thermal and optical water characteristics), made it possible to establish the presence of vast regionsof phytoplankton in the lake. It is shown that the activity of phytoplankton may change water surface conditions, thus causing strong radar signal contrasts.     

Preprocessing of side-looking airborne radar data†

Abstract

Studies on microwave surface scattering in The Netherlands have indicated the need for accurate radar systems for applications in remote sensing. An SLAR system with digital recording was developed and is now being used for several programmes. This system was designed with special attention to speckle reduction and system accuracy. The digital data are processed in a computer with algorithms for geometric and radiometric corrections. In the future aircraft position and attitude measurements will also be used in these correction algorithms. Examples of the results are shown.     

Estimating structural properties of riparian forests with airborne lidar data

Riparian forest zones adjacent to surface water such as streams, lakes, reservoirs and wetlands maintain significant forest ecosystem functions including nutrient cycling, vegetative communities, water quality, fish and wildlife habitat and landscape aesthetics. In order to support the sustainable management of riparian forests, riparian zones should first be carefully delineated and then structural properties of riparian vegetation, especially forest trees, should be accurately measured. Geographical information system (GIS) techniques have been previously implemented to determine riparian zones quickly and reliably. However, basic measurements of forest structures in riparian areas have relied heavily on field-based surveys, which can be extremely time consuming in large areas. In this study, riparian forest zones were initially located using GIS techniques and then airborne lidar (light detection and ranging) data were used to determine and analyse structural properties (i.e. tree height, crown diameter, canopy closure and vegetation density) of a sample riparian forest. Lidar-derived tree height and crown diameter measurements of sample trees were compared with field-based measurements. Results indicated that 77.92% of the riparian area in the study area was covered by forest. Based on lidar-derived data, the average tree height, total crown width, canopy closure (above 3 m) and vegetation density (3–15 m) were found to be 74.72 m, 16.82 m, 71.15% and 26.05%, respectively. Although we found differences between measurement methods, lidar-derived riparian tree measurements were highly correlated with field measurements for tree height (R ²?=?88%) and crown width (R ²?=?92%). Differences between measurement methods were likely a result of difficulties associated with field measurements in the dense vegetation that is often associated with forested riparian areas.     

非正侧视机载雷达一种改进的多普勒频移算法

非正侧机载雷达的杂波分布随距离变化而变化,各距离单元的杂波分布不再满足独立同分布条件,导致统计型STAP处理器性能下降。多普勒频移（Doppler Warping,DW）算法沿主波束方向对杂波非均匀进行了补偿,但在其余方向上杂波非均匀依然存在,因而性能较差。提出了一种改进的非正侧视机载雷达杂波抑制算法——修正的多普勒频移法（Modified Doppler Warping,MDW）,先通过多普勒频移法使各距离单元的杂波谱在主波束方向重合,再沿多个多普勒通道使参考单元和检测单元的杂波谱保持一致,进一步消除非正侧视机载雷达的杂波非均匀程度。仿真结果表明,与原有方法相比,该方法的杂波抑制性能有明显提高,且运算量增加不多,是一种具有工程应用价值的方法。     

Effects of off-nadir view angles on the detected spectral response of vegetation canopies

The effect of view angle upon the detected spectral response of vegetation canopies is studied, using NERC 1982 airborne multispectral scanner campaign data. An attempt is made to distinguish between the effects of atmosphere and the anisotropic reflectance of vegetation canopies. The influence of atmospheric backscatter is found to be greatest at very short wavelengths (0.42-0.45 μm)

Preliminary results confirm that the detected spectral response of vegetation canopies varies with view angle and that the nature and extent of these variations are wavelength-dependent and cover-type dependent. In general, direction-dependent reflectance is symmetric about the nadir value for the visible wavebands, but is manifestly asymmetric in the far-red to near-infrared wavebands. Off-nadir effects for vegetation canopies are found to be smallest in the middle-infrared wavebands.     

Detection of initial damage in Norway spruce canopies using hyperspectral airborne data

Current broadband sensors are not capable of separating the initial stages of forest damage. The current investigation evaluates the potential of hyperspectral data for detecting the initial stages of forest damage at the canopy level in the Norway spruce (Picea abies (L.) Karst) forests of Czech Republic. Hyperspectral canopy reflectance imagery and foliar samples were acquired contemporaneously for 23 study sites in August 1998. The sites were selected along an air pollution gradient to represent the full range of damage conditions in even-aged spruce forests. The changes in canopy and foliar reflectance, chemistry and pigments associated with forest damage were established. The potential of a large number of spectral indices to identify initial forest damage was determined. Canopy hyperspectral data were able to separate healthy from initially damaged canopies, and therefore provided an improved capability for assessment of forest physiology as compared to broadband systems. The 673-724 nm region exhibited maximum sensitivity to initial damage. The nine spectral indices having the highest potential as indicators of the initial damage included: three simple band ratios, two derivative indices, two modelled red-edge parameters and two normalized bands. The sensitivity of these indices to damage was explained primarily by their relationship to foliar structural chemical compounds, which differed significantly by damage class.     

Estimating Carex quality with laboratory-based hyperspectral measurements

The quality of certain plants is considered to be a key factor affecting the food habitat or migration of some herbivorous species, and, thus, to estimate the spatial and temporal variation of plant quality is crucial for understanding the grazing and migrating behaviours of these herbivores. This study aimed to explore the possibilities of estimating plant protein and phosphorus contents, with the laboratory-based hyperspectral measurements of fresh Carex leaves, which are the main food source of many wintering bird species in Poyang Lake, China. Fifty-four Carex leaf samples were collected, and their hyperspectral reflectance (at 350–2500 nm) and crude protein and phosphorus contents were measured in the laboratory. The successive projections algorithm (SPA) was applied for spectral dimension reduction, and a multiple linear regression model was calibrated to estimate the crude protein and phosphorus contents from the wavelengths selected with the SPA. The model validation results showed that the root mean square errors (RMSEs) of estimation were 2.51% for crude protein and 0.06% for phosphorus. Compared with a multiple linear model with randomly selected inputs and full-spectrum partial least-square regression (PLSR), the multiple linear regression model combined with the SPA method exhibited a significant advantage in terms of accuracy in estimating the crude protein and phosphorus contents of Carex leaves.     

Estimating above-ground biomass in young forests with airborne laser scanning

Total above-ground biomass of spruce, pine and birch was estimated in three different field datasets collected in young forests in south-east Norway. The mean heights ranged from 1.77 to 9.66 m. These field data were regressed against metrics derived from canopy height distributions generated from airborne laser scanner (ALS) data with a point density of 0.9–1.2 m^?2. The field data consisted of 79 plots with size 200–232.9 m² and 20 stands with an average size of 3742 m². Total above-ground biomass ranged from 2.27 to 90.42 Mg ha^?1. The influences of (1) regression model form, (2) canopy threshold value and (3) tree species on the relationships between biomass and ALS-derived metrics were assessed. The analysed model forms were multiple linear models, models with logarithmic transformation of the response and explanatory variables, and models with square root transformation of the response. The different canopy thresholds considered were fixed values of 0.5, 1.3 and 2.0 m defining the limit between laser canopy echoes and below-canopy echoes. The proportion of explained variability of the estimated models ranged from 60% to 83%. Tree species had a significant influence on the models. For given values of the ALS-derived metrics related to canopy height and canopy density, spruce tended to have higher above-ground biomass values than pine and deciduous species. There were no clear effects of model form and canopy threshold on the accuracy of predictions produced by cross validation of the various models, but there is a risk of heteroskedasticity with linear models. Cross validation revealed an accuracy of the root mean square error (RMSE) ranging from 3.85 to 13.9 Mg ha^?1, corresponding to 22.6% to 48.1% of mean field-measured biomass. It was concluded that airborne laser scanning has a potential for predicting biomass in young forest stands (> 0.5 ha) with an accuracy of 20–30% of mean ground value.     

The effects of changes in forest biomass on radar backscatter from tree canopies

We validated a canopy backscatter model for loblolly pine forest stands at the Duke Forest, North Carolina, by comparing the observed and modelled SAR backscatter from the stands. Given the SAR backscatter data calibration uncertainty, the model made good predictions of C-HH, C-HV, L-HH, L-HV, L-VV, P-HH, and P-HV backscatter for most of 25 stands studied. The model overestimated C-VV backscatter for several stands, and largely overestimated P-VV backscatter for most of the stands. Using the collected SAR backscatter and ground data, and the backscatter model, we studied the influences of changes in biomass on SAR backscatter as a function of radar frequency and polarization, and evaluated the feasibility of deriving the biomass from the backscatter. This study showed that C-HH, C-HV, C-VV, L-VV, and P-VV SAR backscatter may be insensitive to the biomass change. L-HH, L-HV, P-HH, and P-HV SAR backscatter changed more than 5dB as the biomass varied. This study also showed that the L-HH and P-HH backscatter or L-HV and P-H V backscatter may be used to develop algorithms to retrieve trunk biomass or canopy biomass of the loblolly pine forests.     

On the need to observe vegetation canopies in the near-infrared to estimate visible light absorption

This paper examines the rationale for and implications of using a near-infrared band to estimate the absorption of visible light by vegetation canopies. The benefits of using near-infrared observations have already been documented extensively in the literature, notably in the context of applications based on vegetation indices. These include, for instance, a degree of normalization with respect to undesirable perturbing factors. Our intent here is twofold: provide the theoretical basis to justify using measurements outside the main absorption band of vegetation for the purpose of retrieving canopy properties, and uncover the implications of doing so. On the basis of simple radiation transfer considerations, we conclude that near-infrared observations are critical to ensure the accurate retrieval of absorption estimates in the visible domain, and that observations within the absorption band help control the perturbing effect of the soil background.The analytical approach implemented here is conceptually similar to a scale analysis which permits us assessing the most significant contributions to the absorption and scattering processes in the vast majority of geophysical situations. Our final conclusions derived from a series of intermediate steps that need to be performed first. To this end, we illustrate in Section 2 the fact that a suitably-defined one-dimensional radiation transfer model can always be setup to represent accurately the reflected, transmitted and absorbed fraction of vertical fluxes in any vegetation volume at medium spatial resolutions (100 m or lower), and this irrespective of the local variability exhibited by the canopy attributes. This finding is exploited throughout the paper to show that 1) measurements performed in the near-infrared band are needed to ensure a large dynamic range in albedo for dense canopy conditions, by contrast to the visible domain, 2) measurements in the visible domain are effective to remove the contribution due to the background below vegetation for low to intermediate LAI conditions. This is made possible thanks to the soil line concept and the spectral invariance of the interception process, and 3) the estimation of visible light absorption in a canopy on the basis of combinations of spectral bands (as implemented in traditional vegetation indices) hinges on spectral correlations between variables, most notably those controlling the absorbing and scattering properties of the soil and leaves. A series of implications and consequences is drawn from our analysis and, in particular, the suggestion to adopt modern interpretation techniques, superseding the commonly used vegetation index approaches. These advances allow us to improve on current approaches, in particular by lifting some of the hypotheses associated with approaches based on combinations of spectral bands.     

机载雷达辅助无源传感器对杂波环境下机动目标跟踪

机载雷达辅助无源传感器对目标协同跟踪具有重要战术作用,而当前相关算法模型较为简单。为了贴近工程实际,提出一种机载雷达辅助无源传感器对杂波环境下机动目标的跟踪算法。该算法考虑了地球曲率和载机时变姿态等因素的影响,基于地心地固(ECEF)坐标系,联合交互多模型(IMM)和概率数据关联(PDAF)方法,以综合预测协方差的迹为控制变量来管理机载雷达的开关机。仿真结果表明,通过选择合适的控制门限,在节约辐射能量、提升生存能力的同时算法的跟踪性能并无明显下降,从而表明了所提出算法的有效性。     

Estimating ground fractional vegetation cover using the double-exposure method

Fractional vegetation cover (FVC) is a key parameter in ecological models. It is important to determine the ground FVC quickly and accurately in studies of soil erosion, surface energy balance, and carbon cycling. As one of the FVC ground measurement methods, the photographic method is easy to operate with relatively high precision. However, its classification result showed poor accuracy when an image of a high-contrast scene contained a shadow region where a low signal-to-noise ratio (SNR) existed, because the single-exposure image in the photographic method did not contain sufficient surface information about both the illuminated and shadowed parts. This article presents application of a double-exposure photographic method to determine vegetation cover in the shadow region of an image. It consists of two measurements used in acquiring images (normal and over-exposure) and one image-processing part to handle the obtained images. Illuminated vegetation and soil, as well as the shadow region, was classified with the normally exposed image in the intensity, hue, and saturation (IHS) colour space, and the shadow region was further classified as shadowed vegetation and shadowed soil using the over-exposed image. The results indicate that the over-exposed image reduced the average bias of the FVC in the shadow region from 15.40% to ?4.14% and the root mean square error (RMSE) from 0.174 to 0.066. The RMSE of the entire scene was 0.055 in the over-exposed image and 0.092 in the single-exposed image. The double-exposure method also showed a better classification result than the high dynamic range method in deep shadow regions. This study shows that this method is capable of distinguishing vegetation and soil in the shadow region and thus it is an effective and accurate method for ground FVC measurement.     

Measuring rangeland vegetation with high resolution airborne videography in the blue-near infrared spectral region

An airborne video system was used to investigate the visible and near-infrared (NIR) spectral properties of soil and vegetation features across a range of common arid landscape types. The four-camera system was equipped with filters of 25mm bandwidth centred on 450nm ('blue'), 550nm ('green'), 650nm ('red') and 770nm ('NIR'). The aim was to determine what vegetation properties could be detected by combining data from the blue part of the spectrum with the green, red and NIR range, thereby utilizing information contained in the first channel of Landsat Thematic Mapper (TM) (450-520nm). Adding information from the blue end of the spectrum did not assist in discriminating between green vegetation and dry vegetation or green vegetation and bare soil. This separation is best done with a red/NIR ratio. Neither was the blue band an improvement over the PD54 red-green perpendicular distance index in distinguishing between soil and vegetation, irrespective of phenological condition. The blue band can help separate soil from dry vegetation when combined with the sum of brightness values in the red and green bands in a perpendicular distance index. These properties of the spectral dataspace lead to a sequential classification procedure by which airborne videography data can be used to measure vegetation components which are much slower to assess with conventional ground-based methods. Videography has great potential for rapidly verifying or calibrating vegetation cover indices derivedfrom satellite data. Vegetation cover derived from classifying high resolution video data acquired from a heterogeneous floodplain area correlated well with vegetation indices computed from contemporary and co-registered TM data. The most effective indices for measuring vegetation cover with TM data are the PD54 index, brightness in the red band and a perpendicular index based on the sum of the red-green bands and the blue band. However, multiple regression indicates that the addition of a red/NIR ratio as an additional predictor of cover does not greatly improve the performance of these indices.     

Estimating crop chlorophyll content with hyperspectral vegetation indices and the hybrid inversion method

A hybrid inversion method was developed to estimate the leaf chlorophyll content (LCC) and canopy chlorophyll content (CCC) of crops. Fifty hyperspectral vegetation indices (VIs), such as the photochemical reflectance index (PRI) and canopy chlorophyll index (CCI), were compared to identify the appropriate VIs for crop LCC and CCC inversion. The hybrid inversion models were then generated from different modelling methods, including the curve-fitting and least squares support vector regression (LS-SVR) and random forest regression (RFR) algorithms, by using simulated Compact High Resolution Imaging Spectrometer (CHRIS) datasets that were generated by a radiative transfer model. Finally, the remote-sensing mapping of a CHRIS image was completed to test the inversion accuracy. The results showed that the remote-sensing mapping of the CHRIS image yielded an accuracy of R² = 0.77 and normalized root mean squared error (NRMSE) = 17.34% for the CCC inversion, and an accuracy of only R² = 0.33 and NRMSE = 26.03% for LCC inversion, which indicates that the remote-sensing technique was more appropriate for obtaining chlorophyll content at the canopy scale (CCC) than at the leaf scale (LCC). The estimated results of various VIs and algorithms suggested that the PRI and CCI were the optimal VIs for LCC and CCC inversion, respectively, and RFR was the optimal method for modelling.