B33D-01 13:40h
A long-term Land data record from AVHRR, MODIS and VIIRS
As a first step to creating a land surface long-term data record from AVHRR, MODIS and VIIRS, we plan to reprocess 23 years worth of AVHRR data (1981-now) using algorithms based on improvements identified in the AVHRR Pathfinder 2 project and on the knowledge gained from our MODIS surface reflectance work. In this presentation, we will describe the different steps in the processing chain leading to a consistent data record. In particular, we will discuss the vicarious calibration approach using accurate Rayleigh computations and observations over ocean and clouds to determine the calibration decay in the different AVHRR sensors. This approach was applied consistently to the AVHRR instruments in orbit since 1981 and was evaluated using coincident AVHRR and MODIS data acquired over invariant desert targets. Results of the evaluation and an estimate of the AVHRR calibration accuracy will be presented. In addition to evaluating the calibration approach, the coincident AVHRR and MODIS period is used to establish the AVHRR reflectance product accuracy and to study the impact of spectral differences between AVHRR and MODIS on Land products.
B33D-02 13:55h
Atmospheric corrections for land long term data records
The gases and particles in the atmosphere affect the signal measured by Earth Observations instruments. Careful selection of spectral bands, where this effect can be minimized, has been achieved for the present and future missions (EOS or NPOES). Nevertheless, atmospheric correction should be conducted in order for surface properties to be successfully monitored by those sensors. The present work describes the atmospheric effects in the visible to middle infrared, the modeling in the radiative code 6S and the operational atmospheric corrections, as it can be conducted on the AVHRR, MODIS and VIIRS sensors. Validation of the corrections is performed on the MODIS data by using independent globally representative ground observation of aerosols (AERONET) and by comparison to surface measurements at selected sites. Comparison between AVHRR and MODIS surface reflectance product enable the transfer of validation to the AVHRR surface reflectance data record. The emphasis is put on the accuracy estimation of the derived surface reflectance for the three sensors and its impact on the interpretation of the long term trend.
B33D-03 14:10h
Evaluation of Operational Albedo Algorithms For AVHRR, MODIS and VIIRS: Case Studies in Southern Africa
Shortwave broadband albedo is the fundamental surface variable that partitions solar irradiance into energy available to the land biophysical system and energy reflected back into the atmosphere. Albedo varies with land cover, vegetation phenological stage, surface wetness, solar angle, and atmospheric condition, among other variables. For these reasons, a consistent and normalized albedo time series is needed to accurately model weather, climate and ecological trends. Although an empirically-derived coarse-scale albedo from the 20-year NOAA AVHRR record (Sellers et al., 1996) is available, an operational moderate resolution global product first became available from NASA's MODIS sensor. The validated MODIS product now provides the benchmark upon which to compare albedo generated through 1) reprocessing of the historic AVHRR record and 2) operational processing of data from the future National Polar-Orbiting Environmental Satellite System's (NPOESS) Visible/Infrared Imager Radiometer Suite (VIIRS). Unfortunately, different instrument characteristics (e.g., spectral bands, spatial resolution), processing approaches (e.g., latency requirements, ancillary data availability) and even product definitions (black sky albedo, white sky albedo, actual or blue sky albedo) complicate the development of the desired multi-mission (AVHRR to MODIS to VIIRS) albedo time series -- a so-called Climate Data Record. This presentation will describe the different albedo algorithms used with AVHRR, MODIS and VIIRS, and compare their results against field measurements collected over two semi-arid sites in southern Africa. We also describe the MODIS-derived VIIRS proxy data we developed to predict NPOESS albedo characteristics. We conclude with a strategy to develop a seamless Climate Data Record from 1982- to 2020.
http://modarch.gsfc.nasa.gov/MODIS/LAND/VAL/terra/privette/
B33D-04 14:25h
Analysis of the VIIRS Land Surface Temperature Algorithm Using MODIS Data
The Visible Infrared Imaging Radiometer Suite (VIIRS) is scheduled to replace AVHRR as the nation's polar orbiting wide swath sensor beginning in late 2006 with the launch of the NPOESS Preparatory Project (NPP). Following in the footsteps of EOS MODIS, VIIRS data will be used to generate multiple land products in near-real time. The current baseline VIIRS Land Surface temperature (LST) algorithms, 2- and 4-band variants of the traditional "split window" LST method, were developed by the VIIRS sensor vendor (private industry). In this article, we evaluate these algorithms using MODIS data. Split-window coefficients are derived using the MODTRAN radiative transfer model and a large set of atmospheric profiles derived from radiosondes and TIROS Operational Vertical Sounder (TOVS) data. Fourteen surface emissivity spectral data sets are used to derive the coefficients for the IGBP 17 surface types. Tests of LST uncertainty with 50 MODIS granules consistently indicate that the 2-band split window algorithm is significantly more precise than is the 4-band "dual-split window" algorithm both in daytime and nighttime, and the nighttime algorithm is more precise than the daytime algorithm. In fact, our results suggest that an LST product derived from the dual split window algorithm would exceed NPOESS LST precision specification (0.5 K). Results from the split window algorithm are closer to that specification. Based on these results, we recommend that the dual-split window algorithm be eliminated from consideration, and that the 2-band split window algorithm become the VIIRS baseline operational algorithm for all land cover types and conditions.
B33D-05 15:00h
NDVI Intercalibration of AVHRR, SPOT Vegetation and MODIS
We have intercalibrated the NDVI time series of the AVHRR instruments with moderate- and coarse-resolution NDVI data from SPOT Vegetation and the MODIS instrument on Terra. Benefits of this include using longer time series means to calculate anomalies, being able to use SPOT Vegetation and MODIS Terra NDVI data quantitatively with respect to a 25-year NDVI record, and having multiple NDVI data sources in case one or more satellites or instruments fail. Here we present the results of the NDVI intercalibration using NDVI from the AVHRR instruments at 8 km, SPOT Vegetation at 1 km and a 16 day and monthly Terra MODIS dataset at 1 and 5 km. A description of a new AVHRR NDVI 1981-2004 data set and its intercalibration with SPOT Vegetation are also presented. Fourteen of the CEOS Land Validation core sites were used in conjunction with eight drought hotspots from around the world to evaluate the ability of all three sensors to detect variations during their periods of record. Analysis and statistics of time series, correlation among datasets, measures of autocorrelation, and an analysis of anomaly images using AVHRR means will be presented.
B33D-06 15:15h
Assessement of Land Biophysical Activity Over Multiple Years From a Sensor Independent Product
In the context of scientific research concerning global change issues, remote sensing products have been demonstrated to be essential tools to monitor the characteristics of both land surfaces and their temporal evolution. The biophysical activities on land surfaces are documented from spectral measurements made in space. Advances in the understanding of radiation transfer and availability of higher performance instruments have lead to the development of a new generation of geophysical products able to provide reliable, accurate information on the state and evolution of terrestrial environments. Specifically, a series of optimized algorithms have been developed to estimate the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) for various instruments. Such an approach allows the synergistic use of FAPAR products derived from different sensors and the construction of global FAPAR time series independent from the life time of these specific sensors. This paper will present inter-comparison procedure and results from the exercise conducted with SeaWiFS and MERIS (ENVISAT) global product. A study of European land surfaces response against 2003 drought will be illustrated.
B33D-07 15:30h
Multi- Sensor Translation, Continuity, and Scaling of Vegetation Indices Using Hyperspectral Data
Long term data records require the effective integration of new sensor technologies and improved algorithms to better characterize global and climate change impacts on ecosystems, while preserving the fundamental attributes of the existing data record. In this study, we use fine resolution, hyperspectral data sets from the AVIRIS and Hyperion sensors, to investigate inter-sensor translation and continuity issues related to the long term measurement of inherent surface properties and their spatial and temporal variations. Hyperspectral data sets were convolved to AVHRR, MODIS, and VIIRS sensor bandpasses and inter-sensor translation functions of reflectances and vegetation indices were derived and analyzed for a variety of surface and sun-target-sensor geometries and across a range of ecosystems and scales. The "continuity" relationships developed were then tested with interannual extracts of real MODIS and AVHRR data and included the normalized difference vegetation index (NDVI), the soil-adjusted vegetation index (SAVI), and the enhanced vegetation index (EVI). We found translation and scaling issues to be important in extension of long term, multiple-sensor datasets involving different spectral and spatial resolutions. We found that with calibration, a consistent atmosphere correction scheme, and generalized compositing procedure, translation of multiple-sensor datasets can be achieved with some limitations.
B33D-08 15:45h
Are the terrestrial long-term satellite observations from NOAA AVHRR adequate to justify the Earth Science conclusions that have been based on them?
The NOAA Advanced Very High Resolution Radiometer suffers from many disadvantages compared with more recent instruments, yet vegetation and other land surface observations have been made by these instruments for the entire globe since 1981, and they will not be replaced until sometime after 2008 - resulting in at least a 28 year record. The problems of the AVHRR include the lack of post-launch sensor calibration for its visible and near infrared channels; the gradual changes in equatorial crossing time through each satellite's duty cycle so the solar zenith angle of the observations change; the ephermeris and orbital data that are poor by modern standards, which affects the precision possible in mapping the data; the transmission of only a subsample of the measured surface reflected radiances for each pixel; and the few, rather broad spectral bands included in the available channels, which allow only limited atmospheric correction. Notwithstanding these handicaps, significant Earth system science conclusions have already been based on the AVHRR record and these are likely to increase as the length of its record extends. The conclusions relate to global, biome and agricultural primary production; interannual fluctuations in absorption of photosynthetically active radiation and leaf area index; impacts of ENSO on primary production; phenology; and changes in land cover. Such conclusions will be reviewed and their status considered in the light of the signal-noise relationships within the existing AVHRR record and recommendations made on the limits of interpretation possible with existing data sets. The aim will be to establish goals for improvements in post-processing of the long-term AVHRR data set.