Summary and future steps

1. MODTrendr-based estimates of disturbed forest area are in good agreement with Landsat-based results at four out of examined five test sites. Locations with high and low proportions of forest disturbance were identified correctly. A next step would be, adding test sites with intermediate proportion of forest disturbance (10-20% of total forest area over 12 years) to capture better the full range of disturbance regimes in Northern Eurasia. There remain substantial geographic gaps that could also be addressed by new sites in the northeast and the southwest of the study area. Finalized change detection at Yoshkar-Ola site is expected before the end of the year and can help with one of these geographic gaps.

2. The minimum disturbance magnitude threshold value of 0.02 dNBR was substantially lower than that established for the Pacific Northwest (PNW) of the USA (0.08 dNBR) or the other values tested for that region (Sulla-Menashe et al., 2014).

3. Pure forest pixels were used in calibration and validation in an effort to separate the effect of forest disturbance from that of interannual variation in non-forest land cover types. Among the five test sites, MODTrendr performed worst at AMUR site where forest cover was highly fragmented and pure pixels included only 38% of total forest area based on Landsat. The high level of forest cover fragmentation likely increased omission error because the noise from non-forest cover types obscured forest disturbance signal. At VASY site with very low level of disturbance this did not cause problems, but at AMUR site with large disturbed area the omission error undermined the overall MODTrendr performance.

4. Continental patterns of forest disturbance on MODTrendr forest disturbance map are broadly consistent with other sources (Potapov et al., 2008; Hansen et al., 2013) and general expectations of the prevalence of forest disturbance in different parts of Northern Eurasia. The MODTrendr algorithm is much simpler than other disturbance detection algorithms that have been applied to remote sensing data in this region.

5. Increasing the length of the MODIS time series can be expected to improve MODTrendr results; extending this analysis to include observations for 2014 is an important next step.

References

• Hansen, M.C., Potapov, P.V., Moore, R., Hancher, M., Turubanova, S.A., Tyukavina, A., Thau, D., Stehman, S.V., Goetz, S.J., Loveland, T.R., Kommareddy, A., Egorov, A., Chini, L., Justice, C.O., & Townshend, J.R.G. (2013). High-resolution global maps of 21st-century forest cover change. Science, 342, 850-853.
• Potapov, P., Hansen, M.C., Stehman, S.V., Loveland, T.R., & Pittman, K. (2008). Combining MODIS and Landsat imagery to estimate and map boreal forest cover loss. Remote Sensing of Environment, 112, 3708-3719.
• Sulla-Menashe, D., Kennedy, R.E., Yang, Z., Braaten, J., Krankina, O.N., & Friedl, M.A. 2014. Detecting forest disturbance in the Pacific Northwest from MODIS time series using temporal segmentation. Remote Sensing of Environment (in press). Available online 7 November 2013, ISSN 0034-4257, URL

Credit:

• D. Pflugmacher, D. Sulla-Menashe, O. Krankina. “ASSESSMENT OF THE MODTRENDR ALGORITHM FOR MAPPING FOREST DISTURBANCES IN NORTHERN EURASIA” NELDA-II REPORT (UNPUBLISHED)