Feasibility of ERS-1/2 Interferometry for Forest Inventory
Traditional forest inventory is both expensive and time-consuming. In theory, remote sensing methods offer a good alternative and/or a supporting method for traditional forest inventory and, therefore, the utilization of remote sensing techniques has been subject of intensive investigations during the past few years. Most of the developed satellite methods are based on the use of data obtained at visible, near-infrared or infrared bands. Poor availability of such data - due to the limited penetration capability (e.g. through clouds) at these wavelengths - hinders utilization of these methods. This has stimulated research towards the use of radar-based methods, especially since the launch of remote sensing satellites equipped with a synthetic aperture radar (ERS-1, ERS-2, JERS-1, RADARSAT). However, the reported accuracy of satellite-based methods for standwise stem volume estimates has been worse than 40 % irrespective of the spectral band used, Table 1. Hence, airborne measurements have been recognized as a potential tool for small-area inventories, while the aim of satellite-aided studies is mainly concentrated on large-area monitoring. For operational standwise inventory, even single-source airborne data may not be accurate enough. In order to meet the accuracy requirements (typically 15 %) of standwise forest inventory, data fusion, combining several remote sensing data sources, is suggested.
SAR interferometry and forests
Recent advances in SAR interferometry include detecting subtle changes in the Earth's land and ice surfaces over periods of days to years with a global scale, millimeters accuracy and all-weather capability that are unprecedented. Recent examples illustrate how SAR interferometry can be applied to study glaciers (Goldstein et al. 1993), earthquakes (Massonnet et al. 1993), and volcanoes (Massonnet et al. 1995). SAR interferometry can be used to generate very-high-resolution topographic maps.
Gray and Farris-Manning (1993) reported a loss of coherence at both C- and X-band for forested areas under light to moderate winds (using airborne interferometric SAR) implying a degradation of performance with ERS-1 3-day orbit. Hagberg et al. (1995) found out that coherence was found to be sensitive to temperature changes around 0oC but surprisingly insensitive to wind speed. Hagberg et al. (1995) suggested also that tree height and density of forests can be estimated with interferometric phase information. There are three major phenomena that determine the effective volume scattering distribution of the forests (Hagberg et al., 1995): 1) attenuation of the canopy, which is assumed to be high for boreal forest at C-band, 2) the movements of the scatterers, which are assumed to be largest in the upper part of the trees, and 3) proportion of the area that are filled with trees.
The ranges of backscatter intensities over forests and agricultural fields overlap strongly. Therefore, it is difficult to distinguish between and within these two classes based exclusively on the backscattering intensity at C-band. The interferometric correlation, together with the backscatter intensity and the backscatter intensity change, has proved to a useful tool for the classification of the land-surface classes (Wegmüller et al. 1995).
A laser altimeter is a possible complement to any space-based SAR system. Combined laser or radar altimetry and SAR interferometry can show areas of clear-cutting as well as estimates of the rates of regrowth where extensive logging has occurred. This was the recommendation of 39 scientists gathered in Boulder, Colorado, on February 3-4, 1994, for the SAR Interferometry and Surface Change Detection (RSMAS Technical Report).
The main objective of the on-going project is to evaluate the feasibility and usefulness of ERS-1/2 SAR repeat-pass interferometry for estimating forest resources. The strategy is as follows:
a) interferometric SAR data (backscatter amplitude and its change, coherence and interferograms) are compared with field inventory data and HUTSCAT-derived stand profiles (tree height profiles) to evaluate what forest parameters can be estimated using SAR interferometry alone. The combination of ranging radar together with interferometric SAR images, as suggested in Boulder workshop, are studied for the first time. With this manner the idea proposed by Hagberg et al. (1995) can be verified.
b) interferometric SAR data (including interferometric correlation, backscatter intensity and backscatter intensity change) is combined with other potential multi-temporal satellite data (Landsat, SPOT, RADARSAT and JERS-1) and airborne data (aerial photographs, image produced by a imaging spectrometer AISA and a 94-GHz imaging radiometer and forest stand profiles produced by a ranging scatterometer HUTSCAT) to test whether interferometric SAR data includes additional explanatory power in developing remote sensing-based forest inventory methods.
Seasonal effects, optimum data combinations and interferometric SAR parameters are studied.
Special emphasis is in the evaluation of accuracy and cost-benefit analysis of interferometric SAR techniques compared to the present methods.
Three test sites locate in southern Finland, Teijo (130 km west of Helsinki), Porvoo (30 km east of Helsinki) and Kalkkinen (130 km north of Helsinki) representing a variety of different forest types and covering about 10 000 hectares of forest land. Kalkkinen is the main area of activities with a large multi-source, multi-temporal remote sensing data set.
From the 5000-ha test site Kalkkinen the following information is collected:
Field inventory data
Field inventory data is collected by Uudenmaa-Häme Forestry Center in summer 1996. About 100 parameters describing stand characteristics such as stem volume per hectare, basal area per hectare, mean tree height and tree species are measured for each stand (homogeneous forest areas of about one hectare in size). In order to evaluate the accuracy of field inventory, 40 stands were extremely carefully checked by sample plot measurements. The average value of the stem volume per hectare of the Kalkkinen test site is 141 m3/ha.
Remote sensing data
The remote sensing data set includes satellite data from SPOT,
Landsat, ERS-1/2 (Table 1), JERS-1 and Radarsat (SAR) and airborne
data from imaging spectrometer AISA, airborne ranging radar (HUTSCAT),
94 GHz airborne imaging radiometer and digitized aerial photographs.
The area was successfully measured by SPOT and Landsat satellites
in the late August. Color-infrared photographs in a scale of 1:5000,
1:10000, and 1:20000 and imaging spectrometer measurements were
conducted from Kalkkinen at the beginning of June. The remote
sensing data will be collected by the end of October, with an
exception of 94 GHz radiometer measurement, which is scheduled
for early spring 1997 under wet snow conditions. The ranging radar
HUTSCAT is capable to probe the canopy from the top to the bottom
with range resolution of 65 cm. The HUTSCAT profiles are available
in all three test sites. The tree-height-determining capability
of HUTSCAT is used as a ground truth information for interferograms.
Additional information includes digital elevation model (DEM), digital land-use map and base map 1:20000. Air/soil/vegetation temperature and precipitation monitoring statistics in selected areas is also gathered.
After preprocessing (e.g. radiometric correction, geometric correction, geocorrection and orthorectification), the multi-source, multi-temporal remote sensing data is combined with GIS and ground truth data in ARC/INFO system. Standwise predictor variables are calculated using intensity, texture, band ratio transformations, and image processing techniques. Multivariate data analysis techniques are applied to develop models to estimate stand characteristics and biodiversity information.
The project schedule is the following:
Preparatory work (January-May 1996)
Data acquisition (June - October 1996)
Preprocessing (September- December 1996)
Data analysis (January -October 1997)
Reporting (September - December 1997)
The following results are anticipated from this project:
a) The HUTSCAT feature to produce tree height maps with an accuracy of 1.5 metres is used to evaluate the capability and accuracy of ERS-1/2 Tandem interferometric fringes to estimate tree height and density of forests, the idea proposed by Hagberg et al. (1995). As ground truth information, over 300 km of high-accuracy tree height maps are available.
b) Feasibility of a combined set of radar profilometry (HUTSCAT) and SAR interferometry to monitor areas of clear-cuttings, deforestation and defoliation as well as estimate rates of regrowth.
c) Feasibility of combined set of SAR interferometry, optical satellite images (SPOT and Landsat), radar satellite images (JERS-1 and Radarsat) and airborne images to estimates forest stand characteristics?
d) Estimation corcerning the smallest area for which the estimates can be computed reliably; examples are 1) the whole country or a part of the country (order of magnitude 10 million hectares), b) a forestry board district (0.5-1 million hectares), c) a municipality (50 000 hectares), d) forest holding.
e) Optimum interferometric parameters for forest inventory.
f) Suggestion of instruments needed for a forest inventory satellite mission.
g) Costs and benefits of the proposed methods.
Mäkisara, K., and Tomppo, E., 1996:
Airborne imaging spectrometry in national forest inventory. Proceedings of IGARSS'96 Conference, 27-31 May, 1996, Lincoln, Nebraska.
Keywords: ESA European Space Agency - Agence spatiale europeenne, observation de la terre, earth observation, satellite remote sensing, teledetection, geophysique, altimetrie, radar, chimique atmospherique, geophysics, altimetry, radar, atmospheric chemistry
|Copyright 2000 - European Space Agency. All rights reserved.|