Limits...
An Overview of the “ Triangle Method ” for Estimating Surface Evapotranspiration and Soil Moisture from Satellite Imagery

View Article: PubMed Central

ABSTRACT

An overview of the ‘triangle’ method for estimating soil surface wetness and evapotranspiration fraction from satellite imagery is presented here. The method is insensitive to initial atmospheric and surface conditions, net radiation and atmospheric correction, yet can yield accuracies comparable to other methods. We describe the method first from the standpoint of the how the triangle is observed as obtained from aircraft and satellite image data and then show how the triangle can be created from a land surface model. By superimposing the model triangle over the observed one, pixel values from the image are determined for all points within the triangle. We further show how the stretched (or ‘universal’) triangle can be used to interpret pixel configurations within the triangle, showing how the temporal trajectories of points uniquely describe patterns of land use change. Finally, we conclude the paper with a brief assessment of the method's limitations.

No MeSH data available.


Related in: MedlinePlus

Scatter plot of NDVI versus Tir for an AVHRR image over Central Pennsylvania, 14 June, 1994. Tmax and Tmin, as defined in the text are shown, along with the limits for bare soil NDVI (NDVIo) and that for dense vegetation NDVIs. The horizontal dotted line suggests a possibly better value of NDVIo, than that originally chosen in the article by Owen et al. (1998).
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3814871&req=5

f7-sensors-07-01612: Scatter plot of NDVI versus Tir for an AVHRR image over Central Pennsylvania, 14 June, 1994. Tmax and Tmin, as defined in the text are shown, along with the limits for bare soil NDVI (NDVIo) and that for dense vegetation NDVIs. The horizontal dotted line suggests a possibly better value of NDVIo, than that originally chosen in the article by Owen et al. (1998).

Mentions: In practice, however, it is not necessary to specify many input parameters for the SVAT model with great accuracy. Nor is there need for the image to contain a complete spectrum of surface radiant temperatures and vegetation cover, as long as some patches of bare, dry ground and of dense vegetation can be resolved with sufficient number of pixels with to engender confidence in a representative value for each type of surface. Thus, it is often possible to locate a city center having enough pixels to assign a value of Tmax, while a stand of trees might adequately represent dense vegetation and a value of Tmin. These temperature extremes sometimes can be determined with some confidence even for AVHRR imagery even where the number of pixels in the image is not very large, as shown in Figure 7. Here, the warm edge is distinct, though the bare soil extreme is somewhat uncertain, having been set, probably incorrectly in this example (Owen et al., 1998), at zero; a value of 0.3 (dotted line) now seems a better choice. In a study by Carlson et al. (1995), in which high resolution (aircraft; NS001) imagery was successively degraded from that of 5 m, it was shown that the warm edge remains distinct to a resolution of 80 and possibly to at least 320 m.


An Overview of the “ Triangle Method ” for Estimating Surface Evapotranspiration and Soil Moisture from Satellite Imagery
Scatter plot of NDVI versus Tir for an AVHRR image over Central Pennsylvania, 14 June, 1994. Tmax and Tmin, as defined in the text are shown, along with the limits for bare soil NDVI (NDVIo) and that for dense vegetation NDVIs. The horizontal dotted line suggests a possibly better value of NDVIo, than that originally chosen in the article by Owen et al. (1998).
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3814871&req=5

f7-sensors-07-01612: Scatter plot of NDVI versus Tir for an AVHRR image over Central Pennsylvania, 14 June, 1994. Tmax and Tmin, as defined in the text are shown, along with the limits for bare soil NDVI (NDVIo) and that for dense vegetation NDVIs. The horizontal dotted line suggests a possibly better value of NDVIo, than that originally chosen in the article by Owen et al. (1998).
Mentions: In practice, however, it is not necessary to specify many input parameters for the SVAT model with great accuracy. Nor is there need for the image to contain a complete spectrum of surface radiant temperatures and vegetation cover, as long as some patches of bare, dry ground and of dense vegetation can be resolved with sufficient number of pixels with to engender confidence in a representative value for each type of surface. Thus, it is often possible to locate a city center having enough pixels to assign a value of Tmax, while a stand of trees might adequately represent dense vegetation and a value of Tmin. These temperature extremes sometimes can be determined with some confidence even for AVHRR imagery even where the number of pixels in the image is not very large, as shown in Figure 7. Here, the warm edge is distinct, though the bare soil extreme is somewhat uncertain, having been set, probably incorrectly in this example (Owen et al., 1998), at zero; a value of 0.3 (dotted line) now seems a better choice. In a study by Carlson et al. (1995), in which high resolution (aircraft; NS001) imagery was successively degraded from that of 5 m, it was shown that the warm edge remains distinct to a resolution of 80 and possibly to at least 320 m.

View Article: PubMed Central

ABSTRACT

An overview of the ‘triangle’ method for estimating soil surface wetness and evapotranspiration fraction from satellite imagery is presented here. The method is insensitive to initial atmospheric and surface conditions, net radiation and atmospheric correction, yet can yield accuracies comparable to other methods. We describe the method first from the standpoint of the how the triangle is observed as obtained from aircraft and satellite image data and then show how the triangle can be created from a land surface model. By superimposing the model triangle over the observed one, pixel values from the image are determined for all points within the triangle. We further show how the stretched (or ‘universal’) triangle can be used to interpret pixel configurations within the triangle, showing how the temporal trajectories of points uniquely describe patterns of land use change. Finally, we conclude the paper with a brief assessment of the method's limitations.

No MeSH data available.


Related in: MedlinePlus