Hongliang Fang

Validating MODIS land surface reflectance and albedo products: methods and preliminary results

Abstract This paper presents the general methods and some preliminary results of validating Moderate-Resolution Imaging Spectroradiometer (MODIS) land surface reflectance and albedo products using ground measurements and Enhanced Thematic Mapper Plus (ETM+) imagery. Since ground “point” measurements are not suitable for direct comparisons with MODIS pixels of about 1 km over heterogeneous landscapes, upscaling based on high-resolution remotely sensed imagery is critical. In this study, ground measurements at Beltsville, MD were used to calibrate land surface reflectance and albedo products derived from ETM+ imagery at 30 m, which were then aggregated to the MODIS resolution for determining the accuracy of the following land surface products: (1) bidirectional reflectance from atmospheric correction, (2) bidirectional reflectance distribution function (BRDF), (3) broadband albedos, and (4) nadir BRDF-adjusted reflectance. The initial validation results from ground measurements and two ETM+ images acquired on October 2 and November 3, 2000 showed that these products are reasonably accurate, with typically less than 5% absolute error. Final conclusions on their accuracy depend on more validation results.

Atmospheric correction of Landsat ETM+ land surface imagery. I. Methods

To extract quantitative information from the Enhanced Thematic Mapper-Plus (ETM+) imagery accurately, atmospheric correction is a necessary step. After reviewing historical development of atmospheric correction of Landsat Thematic Mapper (TM) imagery, the authors present a new algorithm that can effectively estimate the spatial distribution of atmospheric aerosols and retrieve surface reflectance from ETM+ imagery under general atmospheric and surface conditions. This algorithm is therefore suitable for operational applications. A new formula that accounts for adjacency effects is also presented. Several examples are given to demonstrate that this new algorithm works very well under a variety of atmospheric and surface conditions.

Retrieving leaf area index using a genetic algorithm with a canopy radiative transfer model

Leaf area index (LAI) is an important structural property of vegetation canopy and is also one of the basic quantities driving the algorithms used in regional and global biogeochemical, ecological and meteorological applications. LAI can be estimated from remotely sensed data through the vegetation indices (VI) and the inversion of a canopy radiative transfer (RT) model. In recent years, applications of the genetic algorithms (GA) to a variety of optimization problems in remote sensing have been successfully demonstrated. In this study, we estimated LAI by integrating a canopy RT model and the GA optimization technique. This method was used to retrieve LAI from field measured reflectance as well as from atmospherically corrected Landsat ETM+ data. Four different ETM+ band combinations were tested to evaluate their effectiveness. The impacts of using the number of the genes were also examined. The results were very promising compared with field measured LAI data, and the best results were obtained with three genes in which the R 2 is 0.776 and the root-mean-square error (RMSE) 1.064. D 2003 Elsevier Science Inc. All rights reserved.

Estimation of incident photosynthetically active radiation from Moderate Resolution Imaging Spectrometer data

[ 1] Incident photosynthetically active radiation ( PAR) is a key variable needed by almost all terrestrial ecosystem models. Unfortunately, the current incident PAR products estimated from remotely sensed data at spatial and temporal resolutions are not sufficient for carbon cycle modeling and various applications. In this study, the authors develop a new method based on the look-up table approach for estimating instantaneous incident PAR from the polar-orbiting Moderate Resolution Imaging Spectrometer (MODIS) data. Since the top-of-atmosphere (TOA) radiance depends on both surface reflectance and atmospheric properties that largely determine the incident PAR, our first step is to estimate surface reflectance. The approach assumes known aerosol properties for the observations with minimum blue reflectance from a temporal window of each pixel. Their inverted surface reflectance is then interpolated to determine the surface reflectance of other observations. The second step is to calculate PAR by matching the computed TOA reflectance from the look-up table with the TOA values of the satellite observations. Both the direct and diffuse PAR components, as well as the total shortwave radiation, are determined in exactly the same fashion. The calculation of a daily average PAR value from one or two instantaneous PAR values is also explored. Ground measurements from seven FLUXNET sites are used for validating the algorithm. The results indicate that this approach can produce reasonable PAR product at 1 km resolution and is suitable for global applications, although more quantitative validation activities are still needed.

Narrowband to broadband conversions of land surface albedo: II. Validation

Abstract In the first paper of this series, we developed narrowband to broadband albedo conversion formulae for a series of sensors. These formulae were determined based on extensive radiative transfer simulations under different surface and atmospheric conditions. However, it is important to validate the simulation results using independent measurement data. In this paper, the validation results for three broadband albedos (total-shortwave, -visible and -near-IR albedos) using ground measurement of several cover types on five different days at Beltsville, MD are presented. Results show that the conversion formulae in the previous paper are very accurate and the average residual standard errors of the resulting broadband albedos for most sensors are around 0.02, which meets the required accuracy for land surface modeling.

Retrieving leaf area index with a neural network method: simulation and validation

Leaf area index (LAI) is a crucial biophysical parameter that is indispensable for many biophysical and climatic models. A neural network algorithm in conjunction with extensive canopy and atmospheric radiative transfer simulations is presented in this paper to estimate LAI from Landsat-7 Enhanced Thematic Mapper Plus data. Two schemes were explored; the first was based on surface reflectance, and the second on top-of-atmosphere (TOA) radiance. The implication of the second scheme is that atmospheric corrections are not needed for estimating the surface LAI. A soil reflectance index (SRI) was proposed to account for variable soil background reflectances. Ground-measured LAI data acquired at Beltsville, Maryland were used to validate both schemes. The results indicate that both methods can be used to estimate LAI accurately. The experiments also showed that the use of SRI is very critical.

Atmospheric correction of Landsat ETM+ land surface imagery. II. Validation and applications

For pt.I see ibid., vol.39, no.11, p.2490-8 (2001). This is the second paper of the series on atmospheric correction of Enhanced Thematic Mapper-Plus (ETM+) land surface imagery. In the first paper, a new algorithm that corrects heterogeneous aerosol scattering and surface adjacency effects was presented. In this study, our objectives are to (1) evaluate the accuracy of this new atmospheric correction algorithm using ground radiometric measurements, (2) apply this algorithm to correct Moderate-Resolution Imaging Spectroradiometer (MODIS) and SeaWiFS imagery, and (3) demonstrate how much atmospheric correction of ETM+ imagery can improve land cover classification, change detection, and broadband albedo calculations. Validation results indicate that this new algorithm can retrieve surface reflectance from ETM+ imagery accurately. All experimental cases demonstrate that this algorithm can be used for correcting both MODIS and SeaWiFS imagery. Although more tests and validation exercises are needed, it has been proven promising to correct different multispectral imagery operationally. We have also demonstrated that atmospheric correction does matter.

Integration of MODIS LAI and vegetation index products with the CSM-CERES-Maize model for corn yield estimation

Advanced information on crop yield is important for crop management and food policy making. A data assimilation approach was developed to integrate remotely sensed data with a crop growth model for crop yield estimation. The objective was to model the crop yield when the input data for the crop growth model are inadequate, and to make the yield forecast in the middle of the growing season. The Cropping System Model (CSM)–Crop Environment Resource Synthesis (CERES)–Maize and the Markov Chain canopy Reflectance Model (MCRM) were coupled in the data assimilation process. The Moderate Resolution Imaging Spectroradiometer (MODIS) Leaf Area Index (LAI) and vegetation index products were assimilated into the coupled model to estimate corn yield in Indiana, USA. Five different assimilation schemes were tested to study the effect of using different control variables: independent usage of LAI, normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI), and synergic usage of LAI and EVI or NDVI. Parameters of the CSM–CERES–Maize model were initiated with the remotely sensed data to estimate corn yield for each county of Indiana. Our results showed that the estimated corn yield agreed very well with the US Department of Agriculture (USDA) National Agricultural Statistics Service (NASS) data. Among different scenarios, the best results were obtained when both MODIS vegetation index and LAI products were assimilated and the relative deviations from the NASS data were less than 3.5%. Including only LAI in the model performed moderately well with a relative difference of 8.6%. The results from using only EVI or NDVI were unacceptable, as the deviations were as high as 21% and −13% for the EVI and NDVI schemes, respectively. Our study showed that corn yield at harvest could be successfully predicted using only a partial year of remotely sensed data.

Corn-yield estimation through assimilation of remotely sensed data into the CSM-CERES-Maize model

One of the applications of crop simulation models is to estimate crop yield during the current growing season. Several studies have tried to integrate crop simulation models with remotely sensed data through data‐assimilation methods. This approach has the advantage of allowing reinitialization of model parameters with remotely sensed observations to improve model performance. In this study, the Cropping System Model‐CERES‐Maize was integrated with the Moderate Resolution Imaging Spectroradiometer (MODIS) leaf area index (LAI) products for estimating corn yield in the state of Indiana, USA. This procedure, inversion of crop simulation model, facilitates several different user input modes and outputs a series of agronomic and biophysical parameters, including crop yield. The estimated corn yield in 2000 compared reasonably well with the US Department of Agriculture National Agricultural Statistics Service statistics for most counties. Using the seasonal LAI in the optimization procedure produced the best results compared with only the green‐up LAIs or the highest LAI values. Planting, emergence and maturation dates, and N fertilizer application rates were also estimated at a regional level. Further studies will include investigating model uncertainties and using other MODIS products, such as the enhanced vegetation index.

Calculating environmental moisture for per-field discrimination of rice crops

The accuracies of rice classifications determined from density slices of broadband moisture indices were compared to results from a standard supervised technique using six reflective Enhanced Thematic Mapper plus (ETM+) bands. Index-based methods resulted in higher accuracies early in the growing season when background moisture differences were at a maximum. Analysis of depth of ETM+ band 5 resulted in the highest accuracy over the growing season (97.74%). This was more accurate than the highest supervised classification accuracy (95.81%), demonstrating the usefulness of spectral feature selection of moisture for classifying rice.