Elsevier

Remote Sensing of Environment

Volume 170, 1 December 2015, Pages 143-152
Remote Sensing of Environment

Combining NLCD and MODIS to create a land cover-albedo database for the continental United States

https://doi.org/10.1016/j.rse.2015.09.012Get rights and content

Abstract

Land surface albedo is an essential climate variable that is tightly linked to land cover, such that specific land cover classes (e.g., deciduous broadleaf forest, cropland) have characteristic albedos. Despite the normative of land-cover class specific albedos, there is considerable variability in albedo within a land cover class. The National Land Cover Database (NLCD) and the Moderate Resolution Imaging Spectroradiometer (MODIS) albedo product were combined to produce a long-term (14 years) integrated land cover-albedo database for the continental United States that can be used to examine the temporal behavior of albedo as a function of land cover. The integration identifies areas of homogeneous land cover at the nominal spatial resolution of the MODIS (MCD43A) albedo product (500 m × 500 m) from the NLCD product (30 m × 30 m), and provides an albedo data record per 500 m × 500 m pixel for 14 of the 16 NLCD land cover classes. Individual homogeneous land cover pixels have up to 605 albedo observations, and 75% of the pixels have at least 319 MODIS albedo observations (≥ 50% of the maximum possible number of observations) for the study period (2000–2013). We demonstrated the utility of the database by conducting a multivariate analysis of variance of albedo for each NLCD land cover class, showing that locational (pixel-to-pixel) and inter-annual variability were significant factors in addition to expected seasonal (intra-annual) and geographic (latitudinal) effects.

Introduction

Land surface albedo, the ratio of upwelling to downwelling radiative flux, is the proportion of the sun's radiant energy that is reflected by the earth's surface. It affects the partitioning of the sun's energy into sensible and latent heat and ground conduction, and is considered an essential climate variable (Schaaf, Cihlar, Belward, Dutton, & Verstraete, 2008). Although radiative and non-radiative factors can affect near surface air temperature (Zhao & Jackson, 2014), increases in albedo tend to lead to a negative radiative forcing (increased surface reflection) and decreases in albedo tend to lead to the opposite effect (Barnes and Roy, 2010, Barnes et al., 2013).

Remote measurement is needed to understand spatial heterogeneity and temporal variability of albedo at broad spatial scales (Gao et al., 2005, Schaaf et al., 2008). The promise of remotely sensed albedo has encouraged research on the performance of such measurements (see overviews by Román et al., 2009, Wang et al., 2014). Román et al. (2009) found agreement between albedo measured at flux towers and satellite-derived albedo values resolved at ~ 500 m × 500 m from the Moderate Resolution Imaging Spectroradiometer (MODIS). Similarly, Wang et al. (2014) found that flux tower and MODIS albedo values agreed when the measurements were acquired over areas of spatially homogeneous land cover.

There have been several studies comparing albedo derived from climate models, which model or prescribe albedo based on land cover (Gao et al., 2014), to albedo derived from MODIS. Wang et al. (2004) found that July albedo derived from a climate model was significantly greater than July albedo derived from MODIS for the eastern United States, whereas the reverse occurred for the southeastern United States in February. Tian et al. (2004) found that albedo derived from a climate model was greater than its MODIS counterpart throughout the continental United States for July and February. Matsui et al. (2007) reported that March, July, and November albedos derived from a climate model tended to be greater than their MODIS counterpart throughout much of the United States, although July black-sky near-infrared climate-model albedo was less than its MODIS counterpart. Lawrence and Chase (2007) reported that winter albedos derived from a climate model were significantly higher in the southeastern United States and significantly lower in the north-central United States than their MODIS counterparts, and that differences in albedo resulted in significant differences in modeled precipitation.

There is a strong correlation between land cover and albedo, such that land cover class specific albedos tend to be prescribed or modeled in most climate models (Bonan, 2002, Davidson and Wang, 2004, Davidson and Wang, 2005, Oleson et al., 2013). A typical rank order of albedo in climate models is snow, bare ground (e.g., desert pavement), cropland and grassland, deciduous forest, coniferous forest (Barnes and Roy, 2010, Betts, 2000, Bonan, 2002, Coakley, 2003, Davidson and Wang, 2004, Davidson and Wang, 2005). Notwithstanding the general trend in land cover class specific albedo, there can also be considerable variation in albedo related to canopy architecture (e.g., open versus closed forest stands), intra- and inter-annual variation in the timing, persistence, and depth of snow, soil color and wetness, and vegetation composition (Bonan, 1997, Davidson and Wang, 2004, Jackson et al., 2008, Myhre et al., 2013, Ghimire et al., 2014, Loranty et al., 2014). Davidson and Wang (2004), for example, reported essentially equivalent snow-free and snow-covered albedos for broadleaf deciduous and evergreen forested stands. Similarly, Barnes and Roy (2008) reported regionally averaged snow-free albedo values for forest and mechanically disturbed (e.g., clear-cut) that were strongly overlapping.

The advent of spatially and temporally explicit satellite derived land cover and albedo datasets creates the opportunity to advance parameterization of land cover albedo. In this paper, we describe the creation of a land cover-albedo dataset derived from 14 years of snow-covered and snow-free MODIS albedo data (Schaaf et al. 2002) for 14 of 16 land cover classes in the National Land Cover Database (NLCD 2011) (Jin et al., 2013, Homer et al., 2015). The two data products are linked by using the 30 m × 30 m NLCD to identify land cover homogeneity at the nominal spatial resolution of MODIS (500 m × 500 m). The main reasons for utilizing these two datasets are: 1) land cover homogeneity at the MODIS scale can be defined precisely using the higher resolution NLCD land cover data; 2) NLCD 2011 land cover is a time series product that includes data for 2001, 2006, and 2011, making it possible to examine land cover-related albedo change, and; 3) the MODIS albedo data have been produced since 2000, making it possible to examine the temporal behavior of albedo over a relatively long time series. We demonstrate the utility of the database by conducting a multivariate analysis of variance of albedo for each of the NLCD land cover classes.

Section snippets

Data

The database was developed from three data products: albedo from the MODIS satellite, land cover from the National Land Cover Database (NLCD) developed from the Landsat satellite, and topographic information (elevation, slope, and aspect) from the National Elevation Dataset (NED). The albedo data used in this study are from the MODIS Collection 5 BRDF/Albedo 500 m product (MCD43A3) (Schaaf et al., 2002) downloaded from the U.S. Geological Survey (USGS) (http://e4ftl01.cr.usgs.gov/MOTA) for the

Results

The number of per pixel snow-covered and snow-free high quality albedo observations ranged between 1 and 605 for the continental United States (Fig. 2a). The maximum possible value, 638, was not realized for any pixel, and only ~ 3% of all pixels have ≥ 90% of the maximum possible number of observations for the study period. The number of observations declined northward and eastward from the desert southwest. During winter months, there were few snow-covered and snow-free high quality

Discussion

The NLCD-MODIS land cover-albedo database is freely available on the MRLC website (http://www.mrlc.gov). The data, organized as described in the methods, are grouped into 16 sections (Table 5). Fourteen sections are devoted to the MODIS albedo data, one section for each year. For each year, there are six layers of albedo for each of the 46 observation dates and an additional binary layer identifying a pixel as snow-covered. Another section includes the number of high quality observations. This

Acknowledgments

The United States Environmental Protection Agency, through its Office of Research and Development, partly funded and managed the research described here. The article has been reviewed by the USEPA's Office of Research and Development and approved for publication. Approval does not signify that the contents reflect the views of the USEPA. C. Barnes' participation was underwritten by contract G13PC00028 between ASRC Federal InuTeq LLC and USGS. The authors are grateful for comments on earlier

References (45)

  • Z. Wang et al.

    Evaluation of MODIS albedo product (MDC43A) over grassland, agriculture, and forest surface types during dormant and snow-covered periods

    Remote Sensing of Environment

    (2014)
  • J. Wickham et al.

    Accuracy assessment of NLCD 2006 land cover and impervious surface

    Remote Sensing of Environment

    (2013)
  • C.A. Barnes et al.

    Radiative forcing over the conterminous United States due to contemporary land cover land use albedo change

    Geophysical Research Letters

    (2008)
  • C.A. Barnes et al.

    Radiative forcing over the conterminous United States due to contemporary land cover land use change and sensitivity to snow and interannual albedo variability

    Journal of Geophysical Research

    (2010)
  • C.A. Barnes et al.

    Projected surface radiative forcing due to 2000–2050 land-cover land-use albedo change over the eastern United States

    Journal of Land Use Science

    (2013)
  • R.A. Betts

    Offset of the potential carbon sink form boreal forestation by decreases in surface albedo

    Nature

    (2000)
  • G.B. Bonan

    Effects of land use on the climate of the United States

    Climatic Change

    (1997)
  • G.B. Bonan

    Ecological climatology: Concepts and applications

    (2002)
  • A. Davidson et al.

    Spatiotemporal variations in land surface albedo across Canada from MODIS observations

    Canadian Journal of Remote Sensing

    (2005)
  • H. Fang et al.

    Developing a spatial continuous 1 km surface albedo data set over North America from Terra MODIS products

    Journal of Geophysical Research

    (2007)
  • J.A. Fry et al.

    Completion of the 2006 National Land Cover Database for the conterminous United States

    Photogrammetric Engineering and Remote Sensing

    (2011)
  • F. Gao et al.

    Multiscale climatological albedo lookup maps derived from moderate resolution imaging spectroradiometer BRDF/albedo products

    Journal of Applied Remote Sensing

    (2014)
  • Cited by (10)

    • Albedo impacts of current agricultural land use: Crop-specific albedo from MODIS data and inclusion in LCA of crop production

      2022, Science of the Total Environment
      Citation Excerpt :

      Data are available from 2000 to present with nine days lag. MODIS products were used to characterise the albedo of contrasting land cover classes globally (Gao et al., 2005) and regionally (Wickham et al., 2015), and to model the impact of land cover change on RF (Myhre et al., 2005) or local temperature (Duveiller et al., 2018). However, implementing surface classifications at the level of individual crops is challenging.

    • A Bayesian geostatistical approach to modeling global distributions of Lygodium microphyllum under projected climate warming

      2017, Ecological Modelling
      Citation Excerpt :

      WorldClim rasters are at a resolution of 2.5 arc-minutes. Rasters with 30 m resolution are also obtained for elevation (Wickham et al., 2015) and human population density (CIESIN, 2005). Presence-only records for L. microphyllum are pooled, and then cleaned to remove duplicates and all attached attributes except those for the observed occurrence locations (coordinates).

    • Effect of land cover change on snow free surface albedo across the continental United States

      2016, Global and Planetary Change
      Citation Excerpt :

      Using coarse-scale GOES satellite data, Davidson and Wang (2004) reported nearly identical albedo values for needleleaf forests and grasslands in Canada in the presence of snow. However, retrieval of high quality MODIS albedo data in the presence of snow is problematic (Moody et al., 2008; see Fig. 2B in Wickham et al., 2015). Several studies have evaluated albedo data derived from MODIS against field measurements (Román et al., 2009; Cescatti et al., 2012; Jin et al., 2013; Wang et al., 2014).

    View all citing articles on Scopus
    1

    Retired.

    View full text