Evaluation is an essential step of model development. However, there is a missing definition of appropriate validation strategies, needed to guarantee reproducibility and generalizability of modeling results. Also, there is a lack of a generally agreed set of 'optimal' statistical measure(s) to assess model accuracy. The objective of the present study is to provide for remote sensing practitioners (i.e., non-statisticians) guidance for model validation strategies and to propose an optimal set of statistical measures for the quantitative assessment of model performance in the context of vegetation biophysical variable retrieval from Earth observation (EO) data. For these purposes, main terms and concepts were reviewed. Then, validation strategies were tested on a polynomial regression model and discussed. Moreover, a literature review was carried out, summarizing the statistical measures used to evaluate model performances. Supported by some exemplary datasets, these measures were calculated and their meanings discussed in view of several model validation criteria. From the results, we recommend to further exploit cross-validation and bootstrapping strategies to guarantee the development/validation of reliable models. An 'optimal' statistic set is suggested, including root mean square error (RMSE), coefficient of determination (R 2 ), slope and intercept of Theil-Sen regression, relative RMSE, and Nash-Sutcliffe efficiency index. A wide acceptance and use of these statistics should enable a better intercomparison of scientific results, urgently needed in times of increasing model development activities that are carried out with respect to upcoming EO missions.
The capability of models to predict vegetation biophysical variables is usually evaluated by means of one or several
goodness-of-fit measures, ranging from absolute error indices (e.g. the root mean square error, RMSE) over correlation
based measures (e.g. coefficient of determination, R2) to a group of dimensionless evaluation indices (e.g. relative
RMSE). Hence, the greatest difficulty for the readers is the lack of comparability between the different models'
accuracies. Therefore, the objective of our study was to provide an overview about the quantitative assessment of
biophysical variable retrieval performance. Furthermore, we aimed to suggest an optimal set of statistical measures. This
optimum set of statistics should be insensitive to the magnitude of values, range and outliers. For this purpose, a
literature review was carried out, summarizing the statistical measures that have been used to evaluate model
performances. Followed by this literature review and supported by some exemplary datasets, a range of statistical
measures was calculated and their interrelationships analyzed. From the results of the literature review and the test
analyses, we recommend an optimum statistic set, including RMSE, R², the normalized RMSE and some other
indicators. Using at least the recommended statistics, comparability of model prediction accuracies is guaranteed. If
applied, this will enable a better intercomparison of scientific results urgently needed in times of increasing data
availability for current and upcoming EO missions.
Vegetation indices (VI) combine mathematically a few selected spectral bands to minimize undesired effects of soil background, illumination conditions and atmospheric perturbations. In this way, the relation to vegetation biophysical variables is enhanced. Albeit numerous experiments found close relationships between vegetation indices and several important vegetation biophysical variables, well known shortcomings and drawbacks remain. Important limitations of VIs are illustrated and discussed in this paper. As most of the limitations can be overcome using physically-based radiative transfer models (RTM), advantages and limits of RTM are also presented.
In the context of defining a procedure for near real time land use/land cover (LULC) mapping with seasonal updated
products, this research examines the use of time-series and phenological indicators from MODIS NDVI. 16-day NDVI
composites from MODIS (MOD13Q1) covering the period from 2001 to the present were acquired for three test sites
located in different parts of Europe. The newly proposed Whittaker smoother was used for filtering purposes. Metrics of
vegetation dynamics (such as minimum, maximum and amplitude, etc.) were extracted from the filtered time-series.
Subsequently, the capability of three data sets (raw, filtered data and phenological indicators) was evaluated to separate
between different LULC classes by calculating the overall classification accuracy for the years 2002 and 2009. Ground
truth data for model calibration and testing set was derived combining existing land cover products (GLC2000 and
GlobCover 2009). Based on these results, the benefits of using phenological indicators and cleaned data for land cover
classification are discussed.
With the launch of the German hyperspectral satellite mission 'Environmental Mapping and Analysis Program'
(EnMAP), anticipated in 2014, unprecedented opportunities will open up for a wide range of applications. Along with
different areas of application, the agricultural sector will particularly benefit from the availability of such observation
capability. Information about state and dynamics of the (non-)vegetated land surface, expressed by biophysical variables,
is required for instance in irrigation water determination, stress detection or in advanced crop production modeling.
In the context of the mission, a toolbox will be provided to determine these variables from hyperspectral imagery.
Algorithms to be implemented will range from empirical methods, such as hyperspectral vegetation indices, to physically
based approaches, involving the inversion of canopy reflectance models.
In this study, potential techniques for the EnMAP toolbox are selected and tested using data from two field campaigns
conducted in two different geographic regions. One of the campaigns was carried out in summer 2009 at the German
agricultural 'Landau test site' as a first step towards the scientific preparation of the EnMAP mission. During the
campaign, data of the airborne hyperspectral scanner HyMap were acquired concurrently with ground measurements of
canopy water content and other variables. The second campaign was conducted in the Cuga river basin in Sardinia (Italy)
during summer 2007.
First results of data analyses will be presented and discussed, emphasizing in particular the benefits of multi-temporal
and multi-seasonal hyperspectral data availability over current operational systems.
Spatial and temporal information of soil water content is of essential importance for modelling of land surface processes
in hydrological studies and applications for operative systems of irrigation management. In the last decades, several
remote sensing domains have been considered in the context of soil water content monitoring, ranging from active and
passive microwave to optical and thermal spectral bands.
In the framework of an experimental campaign in Southern Italy in 2007, two innovative methodologies to retrieve soil
water content information from airborne earth observation (E.O.) data were exploited: a) analyses of the dependence of
surface temperature of vegetation with soil water content using thermal infrared radiometer (TIR), and b) estimation of
superficial soil moisture content using reflectance in the visible and near infrared regions acquired from optical sensors.
The first method (a) is applicable especially at surfaces completely covered with vegetation, whereas the second method
is preferably applicable at surfaces without or with sparse vegetation. The synergy of both methods allows the
establishment of maps of spatially distributed soil water content.
Results of the analyses are presented and discussed, in particular in view of an operative context in irrigation studies.
In the context of a sustainable agriculture, a controlled and efficient irrigation management is required to avoid negative
effects of the increasing water scarcity, especially in arid and semi-arid regions.
Within this background, the project 'Participatory multi-Level EO-assisted tools for Irrigation water management and
Agricultural Decision-Support' (PLEIADeS: http://www.pleiades.es) addressed the efficient and sustainable use of water
for food production in water-scarce environments. Economical, environmental, technical, social and political dimensions
are considered by means of a synergy of leading-edge technologies and participatory approaches. Project partners,
represented by a set of nine pilot case studies, include a broad range of conditions characteristic for the European,
Southern Mediterranean and American regions.
PLEIADeS aimed at improving the performance of irrigation schemes by means of a range of measures, made possible
through wide space-time coverage of Earth observation (E.O.) data and interactive networking capabilities of
Information and Communication Technologies (ICT).
Algorithms for a number of basic products to estimate Irrigation Water Requirements (IWR) in an operational context
are defined. In this study, the pilot zone at the Nurra site in Sardinia, Italy, is chosen to test, validate and apply these
methodologies.
KEYWORDS: Soil science, Reflectivity, Data modeling, Vegetation, Agriculture, Solar radiation models, Sensors, Near infrared, Short wave infrared radiation, Sun
In the context of agricultural applications, the knowledge of soil moisture availability is an essential aspect for irrigation
management. The microwave waveband region (SAR) has been primarily used to estimate soil moisture from Earth
Observation (E.O.) data. However, the optical domain (0.4 - 2.5 μm) may as well offer the possibility to get information
about soil moisture since an overall decrease of soil reflectance corresponds to increasing surface soil water content. Data
from two different experiments (ESA SPARC and AgriSAR) have been exploited aiming at estimating soil moisture
from optical E.O. data by using the radiative transfer model PROSAILH. A soil scale factor (α) was introduced into the
model and estimated using a LUT inversion technique. Relatively high negative relationships between the α-factor and
the measured soil water content (up to R2 = 0.73) could be found for several crop types with low vegetation cover. The
results of this study indicate the potential to retrieve surface soil moisture information from optical E.O. data for similar
soil types. The method gives the advantage of retrieving simultaneously soil and canopy characteristics from the same
E.O. data sources by using a physical method of parameter estimation.
Earth Observation (E.O.) technologies provide a valuable data base for the monitoring of crop and soil characteristics on
a large scale, in a rapid, accurate and cost-effective way. The present work aims at evaluating different methods and
models for the estimation of the Leaf Area Index (LAI) by means of hyperspectral data acquired by the optical airborne
instrument CASI during the ESA AgriSAR 2006 campaign. Inversion of a physical model using an iterative optimization
technique (SQP) and a fast look-up-table (LUT) approach is performed and results are compared with an empirical
model based on the relationship between LAI and WDVI. Furthermore, the analyses carried out on the inversion of the
physical models provide the opportunity to test the spectral bands proposed for the upcoming E.O. satellite Sentinel-2
developed by ESA in the framework of GMES (Global Monitoring for Environment and Security). The Sentinel-2
spectral sampling is compared with the one proposed by an independent study determining the wavebands best
characterizing vegetation and crops. Accuracy of LAI estimation, evaluated with the AgriSAR 2006 field measurements,
is discussed in the context of operational agricultural monitoring.
The objective of this study, which is part of the project "crop drought stress monitoring by remote sensing" (DROSMON), is to assess the potential of hyperspectral imagery to determine drought stress of crops due to heterogeneous soil composition by estimating the leaf area index (LAI). LAI, which characterizes the actual status of the crops and therefore the potential yield, may be seen as the most important parameter indicating medium term drought stress. As a result of former river meanders, the soils in the Marchfeld region are interrupted by bands of lighter soil. The higher content of sand in the bands leads to a lower water storage capacity and consequently to a decrease in plant growth. An airborne HyMap image was acquired in June 2005 during anthesis stage of wheat. Inversion of a radiative transfer model by means of a look-up-table (LUT) approach was performed to retrieve LAI and other canopy parameters from wheat canopy reflectance. Additionally, the LAI was estimated by establishing empirical relationships between LAI and spectral indices (MSAVI, TVI and MTVI2). Both ways of LAI estimation showed a reasonable correlation to final yield measurements obtained one month after the image data acquisition. However, there was a slightly better agreement of model inversion results. The results suggest the applicability of hyperspectral imagery to map potential drought risk of (wheat) fields.
Remote sensing at optical wavelengths provides information on agricultural crop status, therefore being a useful tool for the detection and monitoring of drought stress in crop production. In the project "crop drought stress monitoring by remote sensing" (DROSMON) led by the University of Natural Resources and Applied Life Sciences in Vienna, which started in January 2005, remote sensing methods for drought stress classification were based on physical models of canopy reflectance using a combination of SAILH and PROSPECT.
Spectral reflectance of maize and wheat were measured in situ using a field spectroradiometer FieldSpec Pro FR for different crop development stages and drought stress levels at a test site in Vienna, Austria.
An extensive validation program was carried out measuring various physiological properties of the crops. A significant difference in reflectance was observed between the canopies experiencing distinct drought stress levels. The observed differences could be confirmed by model simulations based on the measured biophysical variables. These suggest that there will be a change in spectral reflectance in drought stressed crops, varying according to the different growth stages. This is most marked in the near (NIR) and mid (MIR) infrared wavelength region, probably due to modifications of leaf internal structure, variations in leaf inclination (e.g. due to wilting) and leaf area index.
We present initial results from this research, which partly support these ideas. Further investigations are necessary.
The determination of UV- and light doses received by people as a function of their activities and their environment, for present and future conditions, is the aim of the presented study. In this paper we present first preliminary results. Measurements of the total daily UV dose received by horizontal and vertical parts of the human body were performed on three chosen days in the region of Vienna, Austria. The measurements were performed in the UV and in the visible spectral range using ultraviolet selective sensors and sensors adapted to human eye sensitivity. Data acquisition was performed by using dataloggers. In this way it was also possible to determine the UV intensity and dose as a function of time and location. The UV intensity was determined for typical outdoor and indoor activities such as walking in a street, in a forest or in flat unobstructed areas. Indoors the determination of UV doses is more straightforward, the determination of the visible dose is however much more complex. A software was developed to determine the total daily dose received by the human body as a function of day and occupation.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.