Modeling grain-size dependent bias in estimating forest area: a regional application

Zheng, Daolan; Heath, Linda S.; Ducey, Mark J.

doi:10.1007/s10980-008-9272-1

Cited by 6 publications

(5 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, in a recent study in East Texas, USA, forest classifications from moderate spatial resolution Landsat satellite imagery (30×30 m pixel) were found to range from about 81 to 91% accuracy in predicting observed forest areas across six counties, with a net overestimation of forested area due largely to misclassification of other treed spaces (e.g., urban areas and pastures with significant tree cover) (Sivanpillai et al 2005). Misclassification error in forest area estimates tend to increase as pixel size increases, holding area constant, or as area decreases, holding pixel size constant (Zheng, Heath, and Ducey 2008a). While regional and global scale observations of forest variables are of great interest they often require that larger pixel sizes be used and hence the opportunity for greater uncertainty in the forest variable of interest due to forest/nonforest misclassification error (Zheng, Heath, and Duce 2008b).…”

Section: Analysis Of Forest Inventory Datamentioning

confidence: 99%

A Hierarchical Model for Quantifying Forest Variables Over Large Heterogeneous Landscapes With Uncertain Forest Areas

Finley

Banerjee

MacFarlane

2011

Journal of the American Statistical Association

View full text Add to dashboard Cite

We are interested in predicting one or more continuous forest variables (e.g., biomass, volume, age) at a fine resolution (e.g., pixel level) across a specified domain. Given a definition of forest/nonforest, this prediction is typically a two-step process. The first step predicts which locations are forested. The second step predicts the value of the variable for only those forested locations. Rarely is the forest/nonforest status predicted without error. However, the uncertainty in this prediction is typically not propagated through to the subsequent prediction of the forest variable of interest. Failure to acknowledge this error can result in biased estimates of forest variable totals within a domain. In response to this problem, we offer a modeling framework that will allow propagation of this uncertainty. Here we envision two latent processes generating the data. The first is a continuous spatial process while the second is a binary spatial process. The continuous spatial process controls the spatial association structure of the forest variable of interest, while the binary process indicates presence of a possible nonzero value for the forest variable at a given location. The proposed models are applied to georeferenced National Forest Inventory (NFI) data and spatially coinciding remotely sensed predictor variables. Due to the large number of observed locations in this dataset we seek dimension reduction not just in the likelihood, but also for unobserved stochastic processes. We demonstrate how a low-rank predictive process can be adapted to our setting and reduce the dimensionality of the data and ease the computational burden.

show abstract

Section: Analysis Of Forest Inventory Datamentioning

confidence: 99%

A Hierarchical Model for Quantifying Forest Variables Over Large Heterogeneous Landscapes With Uncertain Forest Areas

Finley

Banerjee

MacFarlane

2011

Journal of the American Statistical Association

View full text Add to dashboard Cite

show abstract

“…Nelson et al (2002) and Liknes et al (2004) evaluated how classified NLCD and MODIS products could be used as stratification tools in the Lake States of the USA, but these studies did not discuss scaling effects. Zheng et al (2008) suggested that forest area estimation varies with pixel size, with differences in forest cover percentages based on maps of 30 m and 1 km resolutions ranging from 0% to 17% within a 95% confidence interval using county-level data for several US states. Ahl et al (2005) reported a difference of up to 7% on NPP estimates as land-cover data were aggregated from 15 m to 1 km resolution.…”

Section: Introductionmentioning

confidence: 99%

“…We compared the 1 km model developed in this study, using observations at the state level, with the model from another study that used observations at the county level within the Lake States region of the USA (Zheng et al 2008). Both models are statistically significant (figure 5, p,0.001) and provide the best fits for their corresponding scales of study.…”

mentioning

confidence: 99%

“…The theoretical curve is expected to be symmetrical because area gain in one type is equal to the loss in the other. A symmetrical curve is also more feasible for general application (Zheng et al 2008). …”

mentioning

confidence: 99%

“…Comparison of 1 km scaling-effect models developed using state-level observations in this study and the county-level observations from three Lake States of the USA(Zheng et al 2008).…”

mentioning

confidence: 99%

See 2 more Smart Citations

Quantifying scaling effects on satellite‐derived forest area estimates for the conterminous USA

Zheng

Heath

Ducey

et al. 2009

International Journal of Remote Sensing

Self Cite

View full text Add to dashboard Cite

We quantified the scaling effects on forest area estimates for the conterminous USA using regression analysis and the National Land Cover Dataset 30 m satellite-derived maps in 2001 and 1992. The original data were aggregated to: (1) broad cover types (forest vs. non-forest); and (2) coarser resolutions (1 km and 10 km). Standard errors of the model estimates were 2.3% and 4.9% at 1 km and 10 km resolutions, respectively. Our model improved the accuracies for 1 km by 0.6% (12 556 km 2 ) in 2001 and 1.9% (43 198 km 2 ) in 1992, compared to the forest estimates before the adjustments. Forest area observed from Moderate Resolution Imaging Spectroradiometer (MODIS) 2001 1 km land-cover map for the conterminous USA might differ by 80 811 km 2 from what would be observed if MODIS was available at 30 m. Of this difference, 58% (46 870 km 2 ) could be a relatively small net improvement, equivalent to 1444 Tg (or 1.5%) of total non-soil forest CO 2 stocks. With increasing attention to accurate monitoring and evaluation of forest area changes for different regions of the globe, our results could facilitate the removal of bias from large-scale estimates based on remote sensors with coarse resolutions.

show abstract

The Influence of Human Demography on Land Cover Change in the Great Lakes States, USA

Ducey

Johnson

Belair

et al. 2018

Environmental Management

View full text Add to dashboard Cite

Modeling grain-size dependent bias in estimating forest area: a regional application

Cited by 6 publications

References 57 publications

A Hierarchical Model for Quantifying Forest Variables Over Large Heterogeneous Landscapes With Uncertain Forest Areas

A Hierarchical Model for Quantifying Forest Variables Over Large Heterogeneous Landscapes With Uncertain Forest Areas

Quantifying scaling effects on satellite‐derived forest area estimates for the conterminous USA

The Influence of Human Demography on Land Cover Change in the Great Lakes States, USA

Contact Info

Product

Resources

About