Limits...
Spectral measures and mixed models as valuable tools for investigating controls on land surface phenology in high arctic Greenland.

Tamstorf MP, Illeris L, Hansen BU, Wisz M - BMC Ecol. (2007)

Bottom Line: We find several non-linear growth responses to the environmental variables.We conclude that the uses of GAMMs are valuable for investigating growth dynamics in the Arctic.This indicates that although greening might occur wide-spread in the Arctic there are variations on the local scale that might influence the regional trends on the longer term.

View Article: PubMed Central - HTML - PubMed

Affiliation: University of Aarhus, National Environmental Research Institute, Dep, for Arctic Environment, Frederiksborgvej 399, PO Box 358, DK-4000 Roskilde, Denmark. mpt@dmu.dk

ABSTRACT

Background: Changes in land surface phenology are of major importance to the understanding of the impact of recent and future climate changes in the Arctic. This paper presents an extensive study from Zackenberg Ecological Research Operations (ZERO) where snow melt, climate and growing season characteristics of six major high arctic vegetation types has been monitored during 1999 to 2005. We investigate the growth dynamics for dry, mesic and wet types using hand held measurements of far red normalised difference vegetation index (NDVI-FR) and generalized additive mixed models (GAMM).

Results: Snow melt and temperature are of major importance for the timing of the maximum growth as well as for the seasonal growth. More than 85% of the variance in timing of the maximum growth is explained by the models and similar for the seasonal growth of mesic and wet vegetation types. We find several non-linear growth responses to the environmental variables.

Conclusion: We conclude that the uses of GAMMs are valuable for investigating growth dynamics in the Arctic. Contrary to several other studies in the Arctic we found a significant decreasing trend of the seasonally integrated NDVI-FR (SINDVI) in some vegetation types. This indicates that although greening might occur wide-spread in the Arctic there are variations on the local scale that might influence the regional trends on the longer term.

Show MeSH

Related in: MedlinePlus

DOYmax model results with all vegetation types included. Each graph shows the response curve between each explanatory variable and DOYmax. Each response curve is the result of backfitting the GAMM model to calculate the additive contribution of each variable using non parametric smoothing methods. Thus, the y-axis can be interpretted as a transformation of DOYmax. Low values on the y-axis correlate with low DOYmax (early maximum), while high values correlate with higher DOYmax (later maximum). Dashed lines indicate twice standard errors. Each short bar on the x-axis indicates an observation. A: End of snow melt, B: summed air temperature during the green-up period, C: summed rain during the green-up period, D: summed air temperature during the previous growing season. Estimated degrees of freedom are shown by each Y-axis.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2164935&req=5

Figure 5: DOYmax model results with all vegetation types included. Each graph shows the response curve between each explanatory variable and DOYmax. Each response curve is the result of backfitting the GAMM model to calculate the additive contribution of each variable using non parametric smoothing methods. Thus, the y-axis can be interpretted as a transformation of DOYmax. Low values on the y-axis correlate with low DOYmax (early maximum), while high values correlate with higher DOYmax (later maximum). Dashed lines indicate twice standard errors. Each short bar on the x-axis indicates an observation. A: End of snow melt, B: summed air temperature during the green-up period, C: summed rain during the green-up period, D: summed air temperature during the previous growing season. Estimated degrees of freedom are shown by each Y-axis.

Mentions: DOYmax GAMM models: The timing of the maximum NDVI-FR is explained a lot better than SINDVI. The lowest adjusted R2 is 0.85 and the highest 0.98 indicating that the used explanatory variables are controlling most of the variance of DOYmax. The response curves for the combined vegetation types are shown in Figure 5. Changes towards later snow melt clearly indicates a later occurring maximum as does increasing air temperatures during the green-up period although the positive response disappear with very high green-up temperatures. Both rain in the green-up period and temperatures in the previous year show an optimum range for later timing although none of the responses are as pronounced as for ESM or AIRUP. The model explains 89% of the variance in the timing of the combined vegetation types.


Spectral measures and mixed models as valuable tools for investigating controls on land surface phenology in high arctic Greenland.

Tamstorf MP, Illeris L, Hansen BU, Wisz M - BMC Ecol. (2007)

DOYmax model results with all vegetation types included. Each graph shows the response curve between each explanatory variable and DOYmax. Each response curve is the result of backfitting the GAMM model to calculate the additive contribution of each variable using non parametric smoothing methods. Thus, the y-axis can be interpretted as a transformation of DOYmax. Low values on the y-axis correlate with low DOYmax (early maximum), while high values correlate with higher DOYmax (later maximum). Dashed lines indicate twice standard errors. Each short bar on the x-axis indicates an observation. A: End of snow melt, B: summed air temperature during the green-up period, C: summed rain during the green-up period, D: summed air temperature during the previous growing season. Estimated degrees of freedom are shown by each Y-axis.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2164935&req=5

Figure 5: DOYmax model results with all vegetation types included. Each graph shows the response curve between each explanatory variable and DOYmax. Each response curve is the result of backfitting the GAMM model to calculate the additive contribution of each variable using non parametric smoothing methods. Thus, the y-axis can be interpretted as a transformation of DOYmax. Low values on the y-axis correlate with low DOYmax (early maximum), while high values correlate with higher DOYmax (later maximum). Dashed lines indicate twice standard errors. Each short bar on the x-axis indicates an observation. A: End of snow melt, B: summed air temperature during the green-up period, C: summed rain during the green-up period, D: summed air temperature during the previous growing season. Estimated degrees of freedom are shown by each Y-axis.
Mentions: DOYmax GAMM models: The timing of the maximum NDVI-FR is explained a lot better than SINDVI. The lowest adjusted R2 is 0.85 and the highest 0.98 indicating that the used explanatory variables are controlling most of the variance of DOYmax. The response curves for the combined vegetation types are shown in Figure 5. Changes towards later snow melt clearly indicates a later occurring maximum as does increasing air temperatures during the green-up period although the positive response disappear with very high green-up temperatures. Both rain in the green-up period and temperatures in the previous year show an optimum range for later timing although none of the responses are as pronounced as for ESM or AIRUP. The model explains 89% of the variance in the timing of the combined vegetation types.

Bottom Line: We find several non-linear growth responses to the environmental variables.We conclude that the uses of GAMMs are valuable for investigating growth dynamics in the Arctic.This indicates that although greening might occur wide-spread in the Arctic there are variations on the local scale that might influence the regional trends on the longer term.

View Article: PubMed Central - HTML - PubMed

Affiliation: University of Aarhus, National Environmental Research Institute, Dep, for Arctic Environment, Frederiksborgvej 399, PO Box 358, DK-4000 Roskilde, Denmark. mpt@dmu.dk

ABSTRACT

Background: Changes in land surface phenology are of major importance to the understanding of the impact of recent and future climate changes in the Arctic. This paper presents an extensive study from Zackenberg Ecological Research Operations (ZERO) where snow melt, climate and growing season characteristics of six major high arctic vegetation types has been monitored during 1999 to 2005. We investigate the growth dynamics for dry, mesic and wet types using hand held measurements of far red normalised difference vegetation index (NDVI-FR) and generalized additive mixed models (GAMM).

Results: Snow melt and temperature are of major importance for the timing of the maximum growth as well as for the seasonal growth. More than 85% of the variance in timing of the maximum growth is explained by the models and similar for the seasonal growth of mesic and wet vegetation types. We find several non-linear growth responses to the environmental variables.

Conclusion: We conclude that the uses of GAMMs are valuable for investigating growth dynamics in the Arctic. Contrary to several other studies in the Arctic we found a significant decreasing trend of the seasonally integrated NDVI-FR (SINDVI) in some vegetation types. This indicates that although greening might occur wide-spread in the Arctic there are variations on the local scale that might influence the regional trends on the longer term.

Show MeSH
Related in: MedlinePlus