1A  

Conference: 24–27 September 2012www.spie.org/rs

DefenceSecurity+2012

Conference: 24–27 September 2012Exhibition: 25–26 September 2012 www.spie.org/sd

Location Edinburgh International Conference CentreEdinburgh, United Kingdom

Technical Abstracts

RemoteSensing

2012

  2A  

Karin Stein Fraunhofer-IOSB Institute of Optronics, System Technologies and Image Exploitation, Germany2012 Symposium Chair

Charles R. Bostater  Marine-Environmental Optics Lab &  Remote Sensing Center, Florida  Institute of Technology, United States2012 Symposium Co-Chair

David H. Titterton Defence Science and Technology Lab., United Kingdom2012 Symposium Chair

Reinhard Ebert Fraunhofer IOSB, Germany2012 Symposium Co-Chair

SPIE would like to express its deepest appreciation to the symposium chairs, conference chairs, Programme committees, and session chairs who have so generously given of their time and advice to make this symposium possible. The symposium, like our other conferences and activities, would not be possible without the dedicated contribution of our participants and members.

This Programme is based on commitments received up to the time of publication and is subject to change without notice.

Co-Sponsoring Organisations Delivered with the support of Scottish Enterprise

Cooperating Organisations

  3A  

ContentsSPIE Remote Sensing8531:   Remote Sensing for Agriculture, Ecosystems, and Hydrology . . . . . . . . . . .4

8532:   Remote Sensing of the Ocean, Sea Ice, Coastal Waters, and  Large Water Regions 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33

8533:  Sensors, Systems, and Next-Generation Satellites . . . . . . . . . . . . . . . . . .48

8534A:   Remote Sensing of Clouds and the Atmosphere . . . . . . . . . . . . . . . . . . . .70

8534B:   Lidar Technologies, Techniques, and Measurements for  Atmospheric Remote Sensing  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .79

8535:   Optics in Atmospheric Propagation and Adaptive Systems . . . . . . . . . . . .85

8536:   SAR Image Analysis, Modeling, and Techniques . . . . . . . . . . . . . . . . . . . .92

8537:   Image and Signal Processing for Remote Sensing. . . . . . . . . . . . . . . . . .107

8538A:   Earth Resources and Environmental Remote Sensing/GIS  Applications  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .127

8538B:   Special Joint Session on Remote Sensing and Natural Disasters:  Remote Sensing 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .145

8539:   High-Performance Computing in Remote Sensing  . . . . . . . . . . . . . . . . .150

SPIE Security+Defence8540:   Unmanned/Unattended Sensors and Sensor Networks . . . . . . . . . . . . . .156

8541:   Electro-Optical and Infrared Systems: Technology and Applications . . . .165

8542A:   Electro-Optical Remote Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .181

8542B:   Emerging Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .191

8542C:  Quantum-Physics-Based Information Security . . . . . . . . . . . . . . . . . . . . .197

8542D:  Military Applications in Hyperspectral Imaging and  High Spatial Resolution Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .203

8543:   Technologies for Optical Countermeasures  . . . . . . . . . . . . . . . . . . . . . . .207

8544:   Millimetre Wave and Terahertz Sensors and Technology . . . . . . . . . . . . .215

8545:   Optical Materials and Biomaterials in Security and  Defence Systems Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .222

8546:   Optics and Photonics for Counterterrorism,  Crime Fighting and Defence  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230

8547:   High-Power Lasers: Technology and Systems . . . . . . . . . . . . . . . . . . . . .241

  4A  

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and HydrologyMonday - Wednesday 24–26 September 2012 • Part of Proceedings of SPIE Vol. 8531  Remote Sensing for Agriculture, Ecosystems, and Hydrology XIV

8531-1, Session 1

Comparing results of a remote sensing driven interception-infiltration

model for regional to global applications with ECMWF dataMarkus Tum, Erik Borg, Deutsches Zentrum für Luft- und Raumfahrt e.V. (Germany)

We present a remote sensing based modelling approach to simulate the one dimensional water transport in the vadose zone of unsaturated soils on a daily basis, which can be used for regional to global applications. To calculate the hydraulic conductivity our model is driven by van Genuchten parameters, which we estimated using the ISRIC-WISE Harmonized Global Soil Profile Dataset Ver. 3.1 and the Rosetta programme. All needed parameters were calculated for 26 global main soil types and 102 soils of second order. All soil types are based on the original, global FAO 1974 soil classification. Soil depth and the layering of one to six layers were independently defined for each soil. Interception by vegetation is also considered by using Leaf Area Index (LAI) time series from SPOT-VEGETATION. Precipitation is based on daily time series from the European Centre for Medium-Range Weather Forecasts (ECMWF). For our area of interest - Germany we compared our model output with soil moisture data from the ECMWF, because it is based on the same precipitation dataset. We found a good agreement for the general characteristics of our modelled plant available soil water with this dataset, especially for soils which are close to the standard characteristics of the ECMWF. Disagreements were found for shallow soils and soils under stagnant moisture, which are not considered in the ECMWF model scheme, but can be distinguished with our approach. Our proposed approach to combine established models to describe interception and the one-dimensional vertical water transport with time-series of remote sensing data intends to contribute to the realistic parameterization of the soil water budged. This is especially needed for the global and regional assessment of e.g. net primary productivity which can be calculated with vegetation models.

8531-2, Session 1

Comparison of leaf area index derived by statistical relationships and inverse radiation transport modeling using RapidEye data in the European alpine uplandSarah Asam, Julius-Maximilians-Univ. Würzburg (Germany); Doris Klein, Deutsches Zentrum für Luft- und Raumfahrt e.V. (Germany); Stefan Dech, Deutsches Zentrum für Luft- und Raumfahrt e.V. (Germany) and Julius-Maximilians-Univ. Würzburg (Germany)

The Leaf Area Index (LAI) is a key parameter of vegetation structure and a relevant input parameter for flux modeling of energy and matter in the biosphere. It is provided by global operational products in a coarse spatial resolution (1 km) such as the MODIS LAI product. However, in a heterogeneous landscape with a small-scale land use pattern such as the alpine upland in Bavaria, global products are less suited. Additionally, the highly dynamic and owner dependent mowing and pasturing practices in grasslands result in a high temporal and spatial heterogeneity. Thus, the newly available high spatial resolution (5 m) RapidEye data are tested for their applicability for deriving LAI in grassland. Thereby, also the potential of RapidEye’s new red edge channel is investigated. As LAI derivation methods, the empirical-statistical approach based on regression functions with vegetation indices as well as radiation transfer modeling are used. First, established indices (Normalized Difference Vegetation Index, Renormalized Difference Vegetation Index, Soil-Adjusted Vegetation Index) as well as new indices substituting the red band information with RapidEye’s red edge band (Normalized Difference Vegetation 

Index NDVIrededge, Rededge Ratio Index 1, Rededge Ratio Index 2) are calculated for four scenes from beginning of May, end of May, June and September 2011. The correlations of these indices with in situ LAI data, which are collected during the contemporaneous weeks in the River Ammer catchment, are analyzed. The in situ LAI was measured with the LI-COR LAI2000 Plant Canopy Analyzer. From the resulting statistical relationships individual regression equations are derived. The coefficients of determination (R2) of the regressions vary between indices, with results suggesting that the red edge indices outperform the traditional indices. In the physical modeling approach, the RapidEye reflectances are used as input data to an inverted radiation transfer model. The coupled PROSPECT+SAILh model (PROSAIL) is parameterized with leaf optical and canopy properties collected in the field over the vegetation periods of 2011 and 2012. Regarding the inversion, look up tables are applied. While the radiation transfer model inversion has been reported to be the superior method in other studies, specific calibration and underdetermination issues are expected. Thus, the comparison of the results obtained by the radiation modeling for all time steps to those achieved with the empirical-statistical method reveals valuable information about the quality and constraints of these algorithms in order to give an estimate on the uncertainties associated with modeling.

8531-3, Session 1

Contribution of radar images for grassland management identificationPauline Dusseux, Xing Gong, Univ. Rennes 2 (France) and LIAMA (China); Thomas Corpetti, LIAMA (China); Laurence Hubert-Moy, Samuel Corgne, Univ. Rennes 2 (France)

This paper is concerned with the identification of grassland management using both optical and radar data. In that context, grazing, mowing and a mix of these two managements are commonly used by the farmers on grassland fields. These practices and their intensity of use have different environmental impact. Thus, the objectives of this study are, firstly, to identify grassland management practices using a time series of optical and radar imagery at high spatial resolution and, secondly, to evaluate the contribution of radar data to improve identification of farming practices on grasslands. Because of cloud coverage and revisit frequency of satellite, the number of available optical data is limited during the vegetation period. Thus, radar data can be considered as an ideal complement. The present study is based on the use of SPOT, Landsat and RADARSAT-2 data, acquired in 2010 during the growing period. After pre-processing computation, several vegetation indices, bio- physical variables, backscattering coefficients and polarimetric discriminators were computed on the data set. Furthermore, based on a statistic test taking into account the separability between variables, only some variables, identified as the most discriminating, were used to classify grassland fields. To take into account the temporal variation of variables, temporal indexes as first and second order derivatives were used. Classification process was based on training samples resulting from field campaigns and computed according six methods: Decision Trees, K-Nearest Neighbor, Neural Networks, Support Vector Machines, the Naive Bayes Classifier and Linear Discriminant Analysis. Results show that combined use of optical and radar remote sensing data is more efficient for grassland management identification.

8531-4, Session 2

Overview of USAID-World Bank-NASA collaboration to address water management issues in the MENA region (Invited Paper)Shahid Habib, NASA Goddard Space Flight Ctr. (United States)

The World Bank, USAID and NASA have recently established a joint project to study multiple issues pertaining to water related applications in the Middle East North Africa (MENA) region. The main concentration of the project is on utilization of remote sensing data and hydrological 

  5A  

models to address crop irrigation and mapping, flood mapping and forecasting, evapotranspiration and drought problems prevalent in this large geographic area. Additional emphases are placed on understanding the climate impact on these areas as well. Per IPCC 2007 report, by the end of this century MENA region is projected to experience an increase of 3°C to 5°C rise in mean temperatures and a 20% decline in precipitation. This poses a serious problem for this geographic zone especially when majority of the hydrological consumption is for the agriculture sector and the remaining amount is for domestic consumption. The remote sensing data from space is one of the best ways to study such complex issues and further feed into the decision support systems. NASA’s fleet of Earth Observing satellites offer a great vantage point from space to look at the globe and provide vital signs necessary to maintain healthy and sustainable ecosystem. These observations generate multiple products such as soil moisture, global precipitation, aerosols, cloud cover, normalized difference vegetation index, land cover/use, ocean altimetry, ocean salinity, sea surface winds, sea surface temperature, ozone and atmospheric gasses, ice and snow measurements, and many more. All of the data products, models and research results are distributed via the Internet freely through out the world. This project will utilize several NASA models such as global Land Data Assimilation System (LDAS) to generate hydrological states and fluxes in near real time. These LDAS products will then be further compared with other NASA satellite observations (MODIS, VIIRS, TRMM, etc.) and other discrete models to compare and optimize evapotranspiration, soil moisture and crop irrigation, drought assessment and water balance. The floods being a critical disaster in many of the MENA countries, NASA’s global flood mapping and modeling framework (CREST) will be customized for country specific needs and delivered to the remote sensing organizations for their future use. Finally, capacity building is a critical part of this project and NASA will assist in this effort as well.

8531-5, Session 2

Evaluating several satellite precipitation estimates and global ground-based dataset on Sicily (Italy)Francesco Lo Conti, Univ. degli Studi di Palermo (Italy); Kuo-Lin Hsu, University of California, Irvine (United States); Leonardo V. Noto, Univ. degli Studi di Palermo (Italy); Soroosh Sorooshian, University of California, Irvine (United States)

The developing of satellite-based precipitation retrieval systems, presents great potentialities for several applications ranging from weather and meteorological applications to hydrological modelling. Evaluating performances for these estimates is essential in order to understand their real capabilities and suitability related to each application.

In this study an evaluation analysis of satellite precipitation retrieval systems has been carried out for the area of Sicily (Italy). The high density rain gauges has been used to evaluate selected satellite precipitation products. Sicily has an area of 26,000 km2 and the gauge density of the network considered in this study is about 250 km2/gauge. It is an island in the Mediterranean sea with a particular climatology and morphology, which is considered as an interesting test site for satellite precipitation products on the European mid-latitude area. Three satellite products (CMORPH, PERSIANN, TMPA-RT) along with two adjusted products (TMPA and PERSIANN Adjusted) have been selected for the evaluation. Evaluation and comparisons between selected products is performed with reference to the data provided by the gauge network of Sicily and using statistical and visualization tools.

Results show that bias is considerable for all satellite products and climatic considerations are reported to address this issue along with an overall analysis of the PMW retrieval algorithm performance. Moreover bias issues are observed for the adjusted products even though it is reduced respect to only-satellite products. In order to understand this result, the ground-based precipitation dataset used by adjusted products (GPCC dataset), has been examined and weaknesses arising from spatial sampling of precipitation process have been identified for the study area. Therefore possible issues deriving from using global ground-based datasets for local scales are pointed out from this application.

8531-6, Session 2

An integrated information system for the acquisition, management and sharing of environmental data aimed to decision makingGoffredo La Loggia, Elisa Arnone, Giuseppe Ciraolo, Antonino Maltese, Leonardo Valerio Noto, Univ. degli Studi di Palermo (Italy); Umberto Pernice, TeRN - Technological District of the Basilicata (Italy)

This paper reports the first results of the Project SESAMO - SistEma informativo integrato per l’acquisizione, geStione e condivisione di dati AMbientali per il supportO alle decisioni (Integrated Information System for the acquisition, management and sharing of environmental data aimed to decision making), funded by the Regional Sicilian Government within the “Linea di intervento 4.1.1.1 - POR FESR Sicilia 2007-2013”.

The main aim of the project is to provide monitoring services for decision support, integrating data from different environmental monitoring systems (including WSN). From a technological viewpoint an ICT platform based on a service-oriented architecture (SOA) will be developed: in this way it will be possible to coordinate a wide variety of data acquisition systems, based on heterogeneous technologies and communication protocols, providing different monitoring services.

The implementation and validation of the SESAMO platform involves three specific domains: 1) Urban water losses; 2) Early warning system for rainfall-induced landslides; 3) Precision irrigation planning. 

Services in the first domain are enabled by a low cost sensors network collecting and transmitting data, in order to allow the pipeline network managers to analyse pressure, velocity and discharge data for reducing water losses in an urban contest. 

Services in the second domain are enabled by a prototypal early warning system able to identify in near-real time high-risk zones of rainfall-induced landslides. These services include three macro-components: the first allows deriving susceptibility maps of landslides vocation, which identify the hot-spots areas, on the basis of static variables characterizing the territory; the second will integrate rainfall data obtained by rainfall radar and forecast models to monitor and take into account actual rainfall; the third will incorporate hydrological and slope stability models to forecast landslides probability of hot-spot area using rainfall data from second component and measurements from the WSN monitoring system (humidity, temperature, etc..). Some of the variables monitored in real time (e.g. terrain displacements, terrain acceleration) will be also used as indicators of a near oncoming risk.

Services in the third domain are aimed to optimize irrigation planning of vineyards depending on plant water stress. Irrigation planning is nowadays based on field measurements of pre-dawn leaf water potential, that give a fragmentary framework of the vineyard water stress, both in time and in space. Moreover measurements of quality is related to the ability of the agronomist to interpret the plant response also in uncomfortable situations (such as in night time). Data interpretation should take into account environmental forcing due to solar radiation, air temperature and humidity, wind speed and direction, air carbon dioxide concentration, etc.. Irrigation planning could be based on evapotranspiration maps derived using remote or proximity sensing techniques. Within this framework a relationship between leaf water potential and actual evapotranspiration has been implemented and calibrated on some cultivars. Maps of both leaf water potential and actual evapotranspiration will be released in near-real time to the farmers through the SESAMO platform to allow the irrigation planning as function of the actual plants water stress.

The paper reports the first results of the latter two domains of services detailing the remote sensing applications for rainfall monitoring and precision irrigation planning.

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and Hydrology

  6A  

8531-7, Session 2

Possibility of use of surface water resource in an arsenic contaminated region, Prubasthali I and II Block, Burdwan, West Bengal: a GIS approachBiplab Biswas, The Univ. of Burdwan (India)

Arsenic values in groundwater above the maximum permissible limit of 0.05 mg/l (Indian standards) have been reported from Bhagirathi-Hooghly flood plain regions of West Bengal. The present study at Purbasthali I and II block in Burdwan district, West Bengal, is located in active the flood plain. Many flood plain features are mapped using conventional topographical maps and updated by satellite images. Lots of shallow depth (0.50mg) are located around the major flood plain features like ox-bow lakes. Using Buffer analysis in GIS, it has been studied the majority of the villages are located around these features. These features are capable of holding lots of water and can supply water for the villagers at least for drinking purpose. Arsenic of surface water is harmless and it can be suggested that the villagers can use surface water, with proper purification for their daily uses. If the remaining less than 50 per cent population are interested to walk for an average distance of 250 meters, will get a big pond or lake for the much needed arsenic free water. The villagers must be provided purified alternative safe drinking water from surface water bodies.

8531-94, Session 2

Combined X- and L-band PSI analyses for assessment of land subsidence in JakartaFifamè N. Koudogbo, Javier Duro, Alain Arnaud, Altamira Information (Spain); Philippe Bally, ESA European Space Research Institute (Italy); Hasanuddin Z. Abidin, Institut Teknologi Bandung (Indonesia); Heri Andreas, Institute of Technology Bandung (Indonesia)

Jakarta is the capital city of Indonesia with a population of about 9 million people, inhabiting an area of about 660 Km2 along the coast of the Java Sea. The subsidence due to groundwater extraction, increased development, natural consolidation of soil and tectonics in Jakarta has been known since the early part of the 20th century [1]. Evidence of land subsidence exists through monitoring with GPS, level surveys, preliminary InSAR investigations and cracking of buildings. 

Studies conservatively estimate land subsidence in Jakarta occurring at an average rate of 5 cm per year, and in some areas, over 1 meter was already observed. If land subsidence continues at this rate, by 2010, critical areas particularly in north and west Jakarta could have subsided by 500 cm. Recent studies of land subsidence found that while typical subsidence rates were 7.5-10cm a year, in localized areas of north Jakarta subsidence in the range 15-25 cm a year was occurring, which if sustained, would result in them sinking to 4 to 5 meters below sea level by 2025. Land subsidence will require major interventions, including increased pumping, dikes and most likely introducing major infrastructure investment for sea defence [1].

With the increasing prevalence of Earth Observation (EO), the World Bank and the European Space Agency (ESA) have set up a partnership that aims at highlighting the potential of EO information to support the monitoring and management of World Bank projects. It in this framework that was defined the EOWorld projects [2]. Altamira Information, company specialized in ground motion monitoring, was in charge of one of those projects, focusing on the assessment of land subsidence in Jakarta.

The area of interest (AOI) considered extends over about 1300 km2; it covers the Agglomeration of Jakarta but also the suburban district of Cikarang, located in the Benkasi district. 

The technical solution proposed by Altamira is based on the combination of both VHR X-band data for high density of measurement points and L-band to detect strong motion, which is optimal for a densely constructed area with strong ground motion. The processing was performed with the PSI processing chain developed by Altamira and qualified in the framework of the GSE ESA project Terrafirma called Stable Points Network interferometric Process [3]. 

In order to cover the extended AOI, two adjacent frames of COSMO-SkyMed data have been required. The frames have been processed separately, and techniques of mosaicking has been developed in order to derive uniform and calibrated information on terrain deformation affecting the AOI.

ALOS and COSMO-SkyMed measurement maps show a good agreement. Common subsidence patterns have been highlighted. The better temporal distribution of measurements achieved with the high revisit rate of the COSMO-SkyMed mission (acquisition each 16 days) allows to better assess changes of trend of the subsidence (acceleration and slowdown). This fast revisit time also allows fast motion to be monitored. 

In this paper the technical solution proposed by Altamira is presented and the results discussed.

[1] Z. H. Abidin, H. Andreas, M. Gamal, I. Gumilar, M. Napitupulu, Y. Fukuda, T. Deguchi, Y. Maruyama and E. Riawan. Land subsidence characteristics of the Jakarta basin (Indonesia) and its relation with groundwater extraction and sea level rise. in IAH selected papers 16, Groundwater Response to Changing Climate, eds. M. Taniguchi and I.P. Holman, CRC Press, 113-130, 2010

[2] http://siteresources.worldbank.org/INTURBANDEVELOPMENT/Resources/336387-1278006228953/EOworld_Progress_Report.pdf

[3] N. Adam, A. Parizziet M. Crosetto. Practical Persistent Scatterer Processing Validation in the Course of the Terrafirma Project. Journal of Applied Geophysics, vol. 69, pp.59-65, 2009

8531-8, Session 3

RapidEye water quality support service for the environmental agency in Brandenburg, GermanySandra Reigber, RapidEye AG (Germany)

In 2000, the EU parliament made the framework decision that all member countries need to analyse and evaluate the water quality of lakes with a surface area greater than 50 ha. With more than 10.000 lakes Brandenburg has more water bodies than any other state in Germany. Less than 250 of them have to be analysed regarding to the EU water directive. An entire overview of the water quality of all lakes in Brandenburg is therefore not given. At the same time, such an overview is not achievable with conventional sampling techniques due to logistic and financial reasons. Alternatively, remote sensing based analysis methods can be effectively used for this task.

The RapidEye constellation of five identical earth observation satellites is capable of collecting over 4 million km? of 5 m resolution imagery every day. This, combined with the existence of the RedEdge band, makes RapidEye well suited for a large-scale lake water monitoring. This paper presents the results of a feasibility study with the objective to monitor the trophic state of all Brandenburgian lakes with a surface area greater than 10 ha (828 lakes).

In order to determine the trophic state of the lakes, the chlorophyll-a content and the Secchi depth were estimated from RapidEye remote sensing data. Regarding to the EU water framework directive, chlorophyll-a (chl-a) is an indicator of the biological quality of a lake. The Secchi depth (SD) is a parameter for the transparency of a lake. According to the European and national water directives both parameters have to be determined at least four times a year. 

In 2009, RapidEye was able to generate six satellite image coverages of Brandenburg throughout the bio-productive season of the lakes. Coverages were done: once in spring (March-April), four times during the summer season (May-September) and once in autumn (October-November). Because of the enormous amount of satellite images (more than 1000 RapidEye tiles), all data were automatically atmospherically corrected using the atmospheric correction tool ATCOR. Simultaneous to the RapidEye data recording, in-situ data were acquired at several reference lakes. Using these data, chlorophyll-a and Secchi depth algorithms were developed and applied to all RapidEye satellite images. As a result of this study, seasonal maps of chlorophyll-a concentration and Secchi depth were generated for every analysed lake. Additionally, lake minimum, maximum, mean and median values of chlorophyll-a and Secchi depth were estimated for further analyses. 

In a last step all results were evaluated based on a set of validation data. These data were acquired in the course of the regular water sampling activities of the Ministry of Environment, Health and Consumer Protection in Brandenburg. The provided validation data were taken within a minimum of five days before or after an image acquisition.

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and Hydrology

  7A  

First validation results show that the user requirements of the Ministry can be achieved for most of the lakes in this study. Additionally, a detailed error analysis was carried out and will be also described in this paper.

8531-9, Session 3

Monitoring a river channel network at Salar de Uyuni using Landsat ETM+ imagesSeyed Enayat Hosseini Aria, Rick Donselaar, Roderik Lindenbergh, Roderik Koenders, Jiaguang Li, Anneleen Oyen, Technische Univ. Delft (Netherlands)

In this paper the potential of Landsat ETM+ images to detect the temporal and spatial changes of river channels at the terminus of a fluvial system in the Uyuni closed drainage basin in the south west of Bolivia has been investigated. Rivers in semi-arid areas experience downstream reduction of water volume and consequently decrease of fluvial channel size. There is increasing evidence that these dryland rivers terminate as a stream breaks its natural or artificial bank and deposit sediment on the floodplain (crevasse splay). Abandonment of old river channels and creation of the new ones results in a complex pattern over time.

Salar de Uyuni in the basin center is the world’s largest salt pan with an area of ca. 12,500 km2, with drainage area about 5000 km2 (Fig.1). Daily precipitation records between 1995-2010 from the local weather stations were used to find out highly precipitation events. Rainfall is concentrated in the austral summer months December-March, and characterized by short, 24-48 hour periods of torrential rain. We have also collected field work information in terms of sedimentology and terrestrial measurements which were used to interpret the images more precisely, and producing quantitative results.

The useful available Landsat ETM+ images for this region (WRS-2, Path 233, Row 074) are from July 1999 till April 2003, as after May 2003, the Landsat-7 scan-line corrector failed, so the ETM+ is losing approximately 22% of the data due to the increased scan gap. On average, there are about five complete images in each year, where still some images are affected by clouds. In this research, the focus is on detecting new crevasses splays and changes in meanders (winding bends of a river) which formed after intensive precipitation. In addition, identification of areas affecting salinization and degradation for the identified 4-year period were also studied. Image processing techniques such as best band selection for semi-arid areas, stretching, band ratioing, and data fusion were applied to enhance change detection and interpretation. 

The first ETM+ image analysis results show changes in river morphology, and allow to identify new crevasses splays and changes in channel locations. It was also observed that the reflectance of abandoned channels increased after several consecutive weeks of high precipitation(Fig. 2.). It seems that those channels became reactivated, while their water transported sediments in suspension which caused a rapid growth in the reflectance in visible bands. Salt identification using Landsat spectral-band analysis is easiest at end of the dry season (Fig. 3.), as salt dissolve during the rainy season. Furthermore, salt detection is hampered in multispectral images by increasing moisture, because of the lower spectral response. Outcrop observations confirm that a salt crust formed in the upper part of the soil in the dry period. The study shows that Landsat ETM+ images in combination field work data have good potential to identify temporal and spatial changes in river morphology, more specifically the variation in channel locations and crevasse splay growth.

8531-10, Session 3

An object-based method for mapping ephemeral river areas from WorldView-2 satellite dataBenedetto Figorito, CNR-ISSIA (Italy); Eufemia Tarantino, Gabriella Balacco, Umberto Fratino, Politecnico di Bari (Italy)

Karst landscapes are generally not dominated by surface runoff. The “Murge” karstic area in Apulia (Southern Italy) shows a well-developed drainage-network, formed by a dense dendritic pattern in the headwater zone (“Murge Alte”) which evolves into regularly spaced, incised valleys moving towards the coastal area (“Murge Basse”). These valleys are locally named “lame”, and show subvertical rocky flanks and a flat bottom. Valleys cutting the Murge area act as water 

channels only during and immediately after heavy rainfall, and can be classified as episodic (ephemeral) rivers. 

In the last years, the important hydraulic function of the lame has been heavily altered, so that these sites, which were originally drainage lines only, became pasture or agricultural land and later areas of intensive quarrying and urbanisation. These alterations have increasingly transformed, disturbed, and partially or totally destroyed the karst landscape causing modification of the surficial and underground drainage, and deterioration in the quality of groundwater. Moreover, floods, once a rare phenomenon, are becoming frequent in Apulia, especially along its Ionian side. As consequence, a high level of attention must always be given to the protection of the territory in order to avoid high risk situations. Once hydraulic operations are carried out, they must be maintained and monitored in terms of land use and vegetation dynamics in order to continue their efficiency which cannot be put to risk by poor territorial planning.

Classification is a widely studied issue in remote sensing image processing. The common application ranges from land use analysis to change detection. Among the classes of interest, urban areas, farmland, forest, and river/lake areas are traditionally selected. The observation of water body from remote sensing images, is of particular importance during these recent years for two main reasons: there is an important need to assess existing water resource and, because of the increasing water scarcity and related problems, timely information of water increase may help to develop some strategy to restrict flood calamities.

Recently available Worldview-2 high-resolution imagery (WV-2) with nine spectral bands and very high resolutions (spectral and radiometric) affords the opportunity to bring forth new knowledge regarding the on-going debate of whether object-oriented or spectral-based classification approaches are more accurate. 

In this work object-oriented and spectral-based methods for land cover mapping and water-body delineation of an ephemeral river area along the Ionian side of Apulia were implemented. Results were compared with analogous spatial resolution data, i. e. based on historical true color orthophoto, to determine temporal transformations along the investigated study area. The object-oriented and spectral-based approaches were evaluated to estimate impact based upon classification accuracy.

8531-11, Session 4

Small-scale albedo-temperature relationship contrast with large-scale relations in Alaskan acidic tussock tundraHella E. Ahrends, Univ. of Cologne (Germany); Steven Oberbauer, Florida International Univ. (United States); Werner Eugster, ETH Zürich (Switzerland)

Arctic tundra vegetation is characterized by an extreme heterogeneity at a small spatial scale. Small differences in microtopography and moisture conditions cause different types of ecosystems. Different ecosystems are expected to respond differently to climate change. These differences are therefore critical for extrapolating plot measurements to larger spatical scales (e.g. the resolution of aircraft and satellite imagery) and for correctly representing arctic ecosystems in climate models. A detailed knowledge and understanding of soil-vegetation-atmosphere feedback mechanisms at different spatial scales is needed. However, the short growing season and harsh environmental conditions strongly limit the frequency and spatial cover of spectral and thermal measurements. Here we aim at presenting our results from the analyses of data simultaneously measured by a mobile multi-sensor platform. We show the need for such observations for an improved understanding of complex feedback-mechanisms in the tundra.

In the framework of the International Tundra Experiment (ITEX) Arctic Observation Network (AON) we aim to establish the infrastructure for automated multi-sensor observations of tundra vegetation across a topographic transect in the Arctic. During growing seasons 2010 and 2011 first measurements were performed at two sites located in the acidic tussock tundra in northern Alaska. The measurement set-up is moved along a transect that can be spanned by a cable and covers an area of approximately 50 m length and 2 m width. The sensor trolley was equipped with state-of-the-art instruments for recording the distance to vegetation canopy (SR50a Sonic Distance, Campbell Scientific), up- and downwelling short- and longwave radiation (CNR4 net radiometer, Kipp & Zonen), air temperature and surface temperature (SI-111 IR radiometer, Apogee Instruments Inc.) and spectral reflection (Jaz Combo-2, Ocean Optics; GreenSeeker 

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and Hydrology

  8A  

RT100 (505), NTech). The spectral and thermal properties of the surface were measured at a height of approximately 0.8-1.2 m over local ground level. The transects reflect the typical small-scale transitions from dry, shrub and lichen-dominated tundra surfaces to wet, hummocky and tussock-dominated tundra over which traditional large-scale approaches need to integrate. The data were aggregated to distance increments of 45 cm along the transects and standardized (mean-centered) to account for observation date-specific offsets in measurements that were related to specific light and weather conditions but not to the local vegetation surface. 

A relative increase in the albedo of 0.01 (1%) was related to an increase in radiometric surface temperatures of 0.1 to 1 K, which is the inverse of the generally accepted surface temperature-albedo relationship observed at larger spatial scales. We explain this finding with cooling effects of the albedo-influencing surface wetness which primarily results from moss and soil evaporation. This cooling effect dominates over other more general heating effects that can be expected over surfaces with lower albedo under absence or near-absence of evaporation. Our findings are also supported by NDVI measurements. These locally inverted temperature-albedo feedbacks need to be considered in climate models that resolve Arctic environments with a high abundance of moss covers. Our results show that frequent observations of different tundra ecosystems from multi-sensor platforms can provide data critical for the interpretation of large scale data from aircraft or satellite platforms and for understanding the land-atmosphere-interactions for the Arctic and the global system.

8531-12, Session 4

A sequential Bayesian procedure for integrating heterogeneous remotely sensed data for irrigation managementPaolo Addesso, Roberto Conte, Maurizio Longo, Rocco Restaino, Gemine Vivone, Univ. degli Studi di Salerno (Italy)

Continuous and detailed tracking of physical properties variations is crucial in many agricultural applications of remote sensing. In irrigation management the crop water requirements are evaluated through a balance equation that quantifies the water exchange between land surface and atmosphere. In this case a crucial parameter is constituted by the radiometric surface temperature [Rivas04] that can be estimated by using near and thermal infrared data.

However the desirable conjunction of high spatial and temporal resolutions results in conflicting requirements for any real sensor. Accordingly a research effort is nowadays devoted to integrate data at high acquisition rate with data provided by sensors with small pixel sizes. Hence the objective of the present work is to design a procedure for sequential estimation of relevant physical quantities by taking advantage of such heterogeneous information sources.

The exploitation of heterogeneous data is commonly approached from two different points of view. The first aims to construct a fictitious dataset that satisfies the application specifications (data-level fusion). Valuable results have been achieved with this methodology in the field of multispectral image enhancement through panchromatic images, the so-called pansharpening process [Alparone07]. The complementary approach focuses on the combination of higher level characteristics (feature-level fusion) and now represents the most effective method for segmentation of remotely sensed images [Tarabalka10]. 

However the optimality of the existing methods is surely debatable, being they often derived from pragmatic approaches. On the other side the Bayesian framework offers a powerful statistical tool for dealing with random quantities and, very interestingly, with uncertain data. In this paper we exploit this opportunity by formalizing the fusion of data collected from different sensors within the FInite Set Statistics (FISST) framework [Goodman97]. In particular we model the available images as Ambiguously Generated Ambiguous (AGA) measurements [Mahler07] in which the typical randomness due to statistical variations of the observations is combined with the vagueness determined by the pixel size.

The described approach is here applied to derive a Bayesian sequential method that profits from multitemporal multisource data in order to improve estimation of radiometric surface temperature that is essential for agricultural resource management. The performances are evaluated by comparing the predicted values with a high spatial resolution ground truth.

References

[Alparone07] L. Alparone, L. Wald, J. Chanussot, C. Thomas, P. Gamba, and L.M. Bruce, “Comparison of pansharpening algorithms: Outcome of the 2006 GRS-S data-fusion contest,” IEEE Trans. Geosci. Remote Sens., vol. 45, no. 10, pp. 3012-3021, Oct 2007.

[Goodman97] I. Goodman, R. Mahler and H. Nguyen, Mathematics of Data Fusion, Kluwer Academic Publishers, 1997.

[Mahler07] R.P.S. Mahler, Statistical Multisource-Multitarget Information Fusion, Artech House, 2007.

[Rivas04] R. Rivas V. Caselles, “A simplified equation to estimate spatial reference evaporation from remote sensing-based surface temperature and local meteorological data, Remote Sens. Environ., vol. 93, pp. 68-76, Oct 2004.

[Tarabalka10] Y. Tarabalka, M. Fauvel, J. Chanussot, and J. Benediktsson, “SVM- and MRF-based method for accurate classification of hyperspectral images,” IEEE Geosci. and Remote Sens. Letters, vol. 7, no. 4, pp. 736-740, Oct 2010.

8531-13, Session 4

Frost monitoring of fruit tree with satellite dataJinlong Fan, Mingwei Zhang, China Meteorological Administration (China)

The orchards are developing very fast in the northern China in recent years with the increasing demands on fruits in China. In most part of the northern China, the risk of frost damage to the fruit tree in early spring is very high under the background of the global warming. The main reason is that the tolerance to low temperature is decreasing when the leaf is extending, the fruit tree is blooming, and the small green fruit is forming. In some year, the grow season comes earlier than it does in the normal year due to the warm weather in earlier spring and the risk will be higher in this case. The frost damage happens when the cold atmosphere around zero degree flows over the area. According to the reports, frost event in spring happens almost every year in some area in northern China. In bad cases, late frosts in spring can be devastating to all fruit.

In the past 10 year, more than 5 severe frost events have happened in Ninxia Region, northwest China, in April and May and caused no harvest. So frost damage to fruit trees is a significant concern. Lots of attentions have been given to monitoring, evaluating, and protecting the frost. Two orchards in Ninxing, Taole and Jiaozishan orchards were selected as the study areas. The apple, pear are major trees in the orchards. The frost events have been recorded in the orchards since 2000. 5 severe events in April 8 and May 3, 2004, May 13, 2006, May 12, 2007, April 12, 2010 are chosen as the case study. The MODIS data and FY data are used to monitoring the events in combination with minimum air temperature recorded at the weather station. FY is the short name of the Chinese meteorological satellite- FengYun, meaning wind and cloud in English. FY-3 is the second general polar orbiting Chinese meteorological satellite. 11 sensors are onboard FY-3A, launched in May 2008, and FY-3B, launched in November 2010. MERSI onboard FY-3 is similar with the MODIS. The surface temperature is retrieved from the MODIS data and FY data during the events. A remote sensing model for the identification of the frost freeze is developed referring to the meteorological frost model. Finally the paper presents the methodology of monitoring frost with satellite data. The monitoring information will be expected to help the fruit farmers to cope with the damage and loss.

8531-14, Session 5

A comparison of two coupling methods for improving a sugarcane model’s yield estimationJulien Morel, Jean-François Martiné, Agnès Bégué, Pierre Todoroff, CIRAD (France); Michel Petit, Institut de Recherche pour le Développement (France)

Coupling remote sensing data with crop model has been shown improving accuracy of model’s yield estimation.

MOSICAS model simulates with a great accuracy sugarcane yield in controlled conditions plot, based on different variables, including interception efficiency index (i). In this study, we compared two different coupling approaches between this model and remote 

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and Hydrology

  9A  

sensing acquired interception efficiency index data.

Study area was located in Reunion Island (Indian Ocean). Six agricultural plots with surfaces ranging from 8.1 to 25.6 hectares were studied from October 2011 to April 2012 in southern part of the island. Nine repetitions were made for each of these plots, in order to overcome spatial heterogeneities. Three pluviometers were dispatched over our study area, gathering daily data. Weekly ground measurements were made using an Accupar LP80 coupled with external sensor, collecting a total of 39 values of interception efficiency index. Normalized Difference Vegetation Index values were computed using four Spot 4 and three Spot 5 “Top of Canopy” images acquired over the same period. A linear relationship between i and NDVI has then be computed, showing significant accuracy (R? = 0.8934). New values of i have then been calculated with this relationship using Spot 4 and Spot 5 images acquired from January 2011 to October 2011 (NDVI2011). A logistic regression has then been applied on NDVI2011 to compute NDVI values at daily time step. Acquired data have been used to compute i at daily time step for each plots’ growth period. Growth simulations for every studied plot have been made using MOSICAS sugarcane model. Three types of simulations have been made. The first one consisted in a standard simulation where the only input data are daily precipitations, daily temperatures and daily global radiations. The second type has been made using the forcing coupling method, where MOSICAS computed values of i have been replaced by NDVI computed i for each available satellite image. The third type finally consisted in using the assimilation coupling method, where all MOSICAS simulated i have been replaced by NDVI computed i. 

Preliminary results showed interesting prospects. For a field measured yield of 96 tons of sugarcane per hectare, MOSICAS computed a yield of 143 tons per hectare, while MOSICAS coupling with i using assimilation method computed a yield of 128 ton per hectare. Forcing method did not show notable improvement in yield prediction, as we had a lack of satellite images for the last half part of the simulation, preventing us from computing i values for this period.

8531-16, Session 5

Land cover change estimation for protected areas in Sub-Saharan AfricaZoltan Szantoi, Dario Simonetti, Andreas Brink, European Commission Joint Research Ctr. (Italy)

The European Union supports conservation efforts as well as Protected Area’s management in the African continent, especially in Sub-Saharan Africa. However, access to up to date information regarding the status, threats and value of these areas are lacking. Thus, a semi-automated processing chain to detect land cover change for protected areas and their surroundings is being developed by the Joint Research Centre of the European Commission. The Global Forest Resource Monitoring (TREES3) project based on systematic sampling of medium resolution imagery showed that many of the protected areas are disturbed and their surrounding zones are changing rapidly. Thus, the aim of this study was to detect, map and quantify these changes in selected protected areas across Sub-Saharan Africa and to estimate the tree cover loss and degradation of the natural landscape, using medium resolution imagery (Landsat and UK-DMC 2) and object based classification with spectral library. The imagery was collected for the years of 1990 - 2000 - 2010. The pre-processing steps of the imagery involved radiometric calibration, cloud and cloud shadow masking, topographic correction using Shuttle Radar Topographic Mission’s digital elevation data (90m), de-hazing, mosaicing and radiometric normalization. Six different land cover classes were mapped for the protected areas as well as for their 20 km buffer zone using the object based classification method. The classes were based on the recent TREES3 project of the European Commission’s Joint Research Centre’s Land Cover Classification System, which included (1) tree cover, (2) tree mosaic, (3) other wooded land, (4) other vegetation cover, (5) bare or artificial and (6) water. In this paper we are focusing mainly on the explanation of the pre-processing steps of the semi-automated method, which was developed to work in a semi-automatic way over different type of ecoregions. However, the detected land cover dynamics are also presented as change detection maps for the study areas, while quantitative results reveal information on tree loss, deforestation, fragmentation or in some cases revegetation during the investigated period. The changes are then discussed within the framework of tropical deforestation, agricultural intensification and urbanization.

8531-17, Session 5

Water productivity assessment by using MODIS images and agrometeorological data in the Petrolina municipality, BrazilAntônio Heriberto C. Teixeira, Embrapa Semiárido (Brazil); Morris Sherer-Warren, Agência Nacional de Águas (Brazil); Fernando B. T. Hernandez, Univ. Estadual Paulista (Brazil); Hélio L. Lopes, Univ. Federal do Vale do São Francisco (Brazil)

The municipality of Petrolina, located in the semi-arid region of Brazil, is highlighted as an important growing region, with the commercial agriculture being the main activity. The areas with fruit crops are expanding creating a boost for the rural economy; however the irrigated areas have cleared natural vegetation inducing a loss of biodiversity. The contrast between these two systems becomes apparent when analysing the regional biomass production (BIO). Two models were coupled to assess BIO. Monteith´s equation was applied for estimating the absorbed photosynthetically active radiation (APAR), while the Teixeira’s algorithm was used to retrieve the latent heat flux (lE) that together with net radiation (Rn) acquired with the Slob equation, included the effect of land moisture on BIO by the energy partitioning. As the agricultural drainage can adversely affect the water quality, the water productivity (WP) was analysed by the ratio of BIO by ET along the years, considering the irrigated and rain fed areas, being this classification done by a threshold value for the surface resistance to the water fluxes (rs). The bands 1, 2, 31 and 32 from 25 MODIS images together with 10 agro-meteorological stations were used, covering the period of 2010 to 2011, with field calibrations and interpolations. Energy balance measurements in irrigated crops and natural vegetation, during 2002 and 2004, were used to develop equations for surface albedo (a0), using the bands 1 and 2, and for surface temperature (T0), using the split windows technique with the bands 31 and 32. The basic input remote sensing parameters were NDVI, a0 and T0 for the regional scale ET and rs acquirements. The highest BIO values occur after the rainy period in April, around 2100 kg ha-1 month-1 for irrigated crops and 1550 kg ha-1 month-1 for natural vegetation. Maximum mean values for ET and WP, 67 mm month-1 and 2.9 kg m-3, respectively, happened also during this month. As the rains keep the soil moisture uniform this time of the year, the differences between irrigated crops are the lowest ones. The highest incremental values of BIO, ET and WP, as a result of the irrigated crops introduction, are from August to October, outside the rainy period, when the sun is around the zenith position, average totals of precipitation around 5 mm month-1 and high atmosphere demand, with reference evapotranspiration (ET0) rates from 170 to 195 mm month-1. More uniformity on the water variables studied along the year occurs in natural vegetation, evidenced by the lower standard deviation when comparing to irrigated crops, where due to the different cultural and irrigation managements the differences in WP values are more than the double of those for natural vegetation during 

 the driest period of the year. The models applied with MODIS images at the municipality level are considered to be suitable for water productivity analyses and for quantifying the effects of increasing irrigated areas over natural vegetation on regional water consumption in situations of quick changing land use pattern, as in the case of Petrolina municipality.

8531-18, Session 6

Hyperspectral imaging: do information content, land cover classification, sensitivity analysis and inverse modeling of spectral reflectance lead to the same set of optimal spectral bands? (Invited Paper)Massimo Menenti, A. Mousivand, Seyed Enayat Hosseini Aria, Ben Gorte, Technische Univ. Delft (Netherlands)

Terrestrial targets are characterized by heterogeneity at all scales and the observed spectral radiance across the 0.4 m - 2.4 m spectral region is determined by a complex combination of target geometry and bio-geophysical and chemical composition of target elements. The assumption underlying the quest for the continuous and reliable provision of hyper-spectral data from space is that exhaustive coverage and high spectral resolution sampling of reflected radiance is necessary to characterize such terrestrial targets. Literature provides conflicting evidence to this regard and we argue 

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and Hydrology

  10A  

that to a significant extent this is due to the fundamentally different requirements of four different applications of hyper - spectral image data:

- identification of independent spectral features;

- classification of land cover;

- understand the relation of spectral radiance with the properties of observed targets;

- retrieval of target properties by inversion of spectro - directional radiometric data

Theoretically the information content of a spectral sample is measurable by entropy which shows the amount of disorder, or unpredictability in the sample. If a spectral band is assumed as an independent sample, we can measure the information content of it, which is affected by several factors. Basically, the more heterogeneous the spectral radiance in a spectral band is, the higher the information content.. Therefore, radiometric resolution, width of bands (spectral resolution), etc. can influence on heterogeneity of a band which leads to changes in its information content. We have determined the optimal set of bands for very different scenes by using two different approaches: a) minimizing the correlation of selected bands; b) maximizing the amount of mutual information. In a spectral point of view, the spectral bands which can reconstruct full reflectance spectrum precisely can immensely useful to improve pattern recognitions issues. 

The classification of land cover builds upon the separability, reliability and accuracy of class characterization using spectral attributes. Different band combinations are needed to assign various classes correctly, i.e. a unique band set may not be ideal for all applications. We identified different optimal sets for water, vegetation and ‘mineral’ sites, but these sets are not necessarily the same needed to distinguish *between* water, vegetation and ‘mineral’ objects.

The relation of spectral radiance with target has been investigated by a detailed analysis of the sensitivity of spectral radiance to the properties of observed targets. This has been done by using a model of radiative transfer in the soil-vegetation system and developing a new method to perform sensitivity analysis.. With this approach, in spite of looking for high variance among various spectral bands, we have identified the spectral bands most sensitive to specific target properties. 

Retrieval by inverse modeling of spectro-radiometric image data has been widely employed to extract bio-geophysical information (such as LAI) from vegetated areas. When the objective is to estimate a certain vegetation property, the optimal band selection with focus on the content of the data is not applicable and efficient anymore. In this sense, a precise knowledge about which spectral bands are most sensitive to which parameter is an asset.

8531-20, Session 6

Assessing irrigated cropland dynamics in central Asia between 2001 and 2010 based on MODIS time seriesChristopher Conrad, Fabian Loew, Moritz Rudloff, Gunther Schorcht, Julius-Maximilians-Univ. Würzburg (Germany)

Despite the transformation processes initiated after independence from the Soviet Union in 1991, most Central Asian countries heavily rely on high productivity in the agricultural sector. But the decline of the 8 Million ha of irrigated land caused by water uncertainties, declining irrigation and drainage infrastructure, and several other factors is recurrently reported to be continuing in the entire Aral Sea Basin. Spatially explicit information, for instance received from remote sensing based analyses, can be beneficial for regional assessments of trends in cropland development or for identifying options for improvements of land and water management in this region. 

The proposed presentation focuses on assessment of irrigated cropland dynamics in Central Asia during the past decade by analyzing MODIS time series. Extend of cropland, cropping intensity (number of cropping seasons per year), and spatio-temporal cropping patters (monoculture and rotation systems) are assessed based on phenological profiles extracted from 8day MODIS products at a spatial resolution of 250m. Four MODIS tiles were processed to cover entire Central Asia. The quality assessment science data sets of the selected MODIS products indicated a high data quality throughout the year, which obviously occurs due to excellent atmospheric conditions and the dry continental climate in the lowlands of Central Asia. Decision trees generated from cropping calendars (expert knowledge) and automated algorithms (random forest) have been applied and 

compared. Time series of vegetation indices were calculated and temporally segmented for feature generation. Descriptive statistics of temporal segmentation accounts for variability of management, water availability, temperature development influencing sowing dates and other crop phenological steps. For training (only random forest) and validation, existing high resolution (Landsat and RapidEye) crop maps of Khorezm, the Fergana Valley (both in Uzbekistan), the Kashka Darya region (Tajikistan and Uzbekistan) and the Kyzyl-Orda region in Kazakhstan were utilized. Multiple seasons (e.g. winter wheat followed by a summer crop) were extracted by counting the peaks of each time series at pixel level and crop rotations were assessed by statistical comparison of two subsequent years (auto-correlation and cross-correlation).

Accuracy assessment returned acceptable results, even though in some regions degraded but vegetated land can hardly be distinguished from crops such as cotton under drought conditions - irrespectively if knowledge-based decision trees or automated classifiers were applied. The results show high dynamics of cropland extends in both dimensions, spatial and temporal. Especially water scarce years such as 2000, 2001, or 2008, show low to zero productivity at the fringes of the irrigation systems in the river catchments of Amu Darya and Syr Darya. Altogether, MODIS time series of croplands allow for regional assessments of cropping conditions in irrigation systems and the situation of water deficit indicated by unused downstream locations of the irrigation systems. Based on MODIS like data long-term observations at regional scale can be established not only in Central Asia, but also in other arid environments, where irrigation agriculture is essential for rural income generation and food security.

8531-21, Session 6

Caveats in calculating crop specific pixel purity for agricultural monitoring using MODIS time seriesGregory Duveiller, European Commission Joint Research Ctr. (Italy)

Monitoring agriculture at regional to global scales with remote sensing requires the use of sensors that can provide information over large geographic extends with a high revisit frequency. Current sensors satisfying these criteria have, at best, a spatial resolution of the same order of magnitude as the field sizes in most agricultural landscapes. Landscape fragmentation combined with crop rotation practices further complicates the task of obtaining adequate crop specific time series of remotely-sensed observations. As a consequence, a coarser signal describing the general cropland is used, limiting the potential of remote sensing technology.

Research has demonstrated that crop specific monitoring is possible with coarse spatial resolution such as MODIS (approximately 250 m at nadir) if a selection purer time series is used. To do so, a mask of the target crop is necessary at fine spatial resolution in order to calculate the crop specific pixel purity at the coarser spatial resolution. This pixel purity represents the relative contribution of the surface of interest, in this case the surface covered by the target crop, to the signal detected by the remote sensing instrument. A straightforward way to compute pixel purity is to calculate the area of the target crop that falls in the coarse spatial resolution grid. However, the observation footprint is generally much larger than the squared projection of the pixel. Furthermore, the relative contribution within this footprint is not homogeneous and depends on the spatial response of the sensor. These effects are particularly important for MODIS which has a triangular point spread function and which scans the Earth with high view zenith angles, heavily distorting the observation footprint.

This study analyses the consequences of calculating crop specific pixel purity using a model of the MODIS spatial response with respect to basing it only on the percentage of surface falling in the squared grid in which the product is delivered. The MODIS spatial response model is constructed taking into account different components of its point spread function (PSF): the detector PSD, the image motion PSD and the optical PSF. Since the result is anisotropic, the model must further be adjusted according to the angle between the north-south direction and the ground track of the satellite, the latter varying with latitude. Using this model on a given case study, differences in pixel purity can range between -10 to 15%, depending on the correspondence between the grid and the fields and on the absolute value of pixel purity that is considered. The spatial response is also different depending on whether the MODIS sensor is on-board of the Terra (descending) or Aqua (ascending) platform, and the repercussion on the calculation is also explored. The effect of high view zenith 

Conference 8531: Remote Sensing for Agriculture, Ecosystems, and Hydrology

  11A  

angles on the purity calculation is also quantified with respect to the spatial response at nadir. Finally, the consequences of underestimating the spatial response when calculating pixel purity is illustrated by analysing the effect on the quality of daily MODIS time series.

8531-22, Session 6

Plant optical properties for chlorophyll assessmentRumiana Kancheva, Georgi Georgiev, Denitsa Borisova, Space Research and Technology Institute (Bulgaria)

Remote sensing techniques acquire increasing importance in vegetation phytodiagnostics. Visible and near infrared multispectral data have proved abilities in vegetation monitoring. This wavelength region reveals significant sensitivity to plant pigment content. The optical signatures of leaves are mostly defined by the composition of photosynthetic pigments and their stress-induced changes, and as such provide valuable information about the physiological status of plants. The information is carried by the specific spectral behaviour of healthy plants and plants subjected to short-term or long-term stress impacts. Chlorophyll content is a prime bioindicator of plant condition being responsible for light absorption and the photosynthetic processes. In our study, optical multispectral data have been used to reveal the performance of different spectral signatures in chlorophyll estimation. Reflectance factors, vegetation indices, red edge shift, fluorescence parameters, and chromaticity features, have been related in a statistical manner to plant chlorophyll in order to examine the statistical significance of plant spectral response to chlorophyll variations. High correlation have been found permitting quantitative dependences to be established and used for plant diagnosis. Empirical relationships between plant spectral properties and chlorophyll concentration have been established that allow chlorophyll estimation and plant condition assessment in terms of chlorophyll variation to be performed by using diverse spectral indicators.

8531-23, Session 7

Sources of uncertainty for eddy covariance measurements over heterogeneous surfaces in a semi-arid region: impact to remote sensing (Invited Paper)John H. Prueger, Agricultural Research Service (United States); Lawrence E. Hipps, Utah State Univ. (United States); Joe G. Alfieri, Agricultural Research Service (United States); Christopher M. U. Neale, Utah State Univ. (United States); William P. Kustas, U.S. Dept. of Agriculture (United States); Jerry L. Hatfield, Agricultural Research Service (United States)

Surface measurements of land-atmosphere exchange processes, such as the turbulent transport of heat and moisture, provide critical “ground truth” data for the evaluation of remote sensing-based products and models. Eddy covariance measurements of turbulent fluxes of water heat and momentum are routinely used in complex remote sensing experiments and are considered the most physically based approach for such measurements. Regardless, there is uncertainty associated with eddy covariance measurements of surface fluxes due to both the limitations of the sensors and the theoretical assumptions underlying the various measurement techniques. Additional sources of uncertainty can be found in the spatiotemporal variations of both the atmospheric conditions and the surface properties within the source area of the measurements. This is particularly true for remote sensing scenes captured over complex landscapes where limited fetch and patchwork-like surface characteristics are problematic for many surface measurement techniques. However, all of these potential sources of uncertainty can significantly impact comparisons of remotely sensed and surface data depending on the complexity of the landscape and the resolution of the data. We present and discuss potential sources of eddy covariance measurements and present results from a remote sensing study in Bushland, Texas under complex or extreme conditions such as heterogeneous terrain and strong adv