Applying Improved Estimates of MODIS Productivity to Characterize Grassland Vegetation Dynamics

Abstract

Prescribed fire was used in two semiarid grasslands to reduce shrub cover, promote grass production, and reduce erosional loss that represents a potential non-point-source of sediment to degrade water quality. This study measured transported soil sediment, dynamics in soil surface microtopography, cover of the woody shrub, grass, and bare ground cover classes, and soil fertility measured by nitrogen-mineralization potentials for the respective cover classes over a 9-year period during which 2 fires occurred. In general, the effects of two drought periods were equal to or greater than the effects of fire on the measured parameters. Following the second fire at the grama grass-dominated site, the number of live junipers was significantly lowered (P , 0.001). Fire consumed grass and shrub cover, which created the potential for greater transport of sediment and loss in soil surface elevation. Soil fertility under shrubs was significantly higher than soil fertility under grasses in both grassland sites before and after the first fire; however, that highly significant pattern was not present after the second fire for about 2 years at both sites. This suggests that the reestablishment of a natural fire frequency would likely dampen the well-documented pattern of greater soil fertility under shrub islands relative to grass islands. Repeat use of prescribed fire may keep shrub cover in check and promote fundamental changes in soil processes within semiarid grasslands, but its value must be weighed against the increased potential for erosion following fire. ResumenEl fuego prescrito se uso en dos pastizales semiá ridos para reducir la cobertura de arbustos, promover la producció n de zacates y reducir las perdidas de suelo por erosió n, la cual representa una fuente potencial no puntual de sedimentos para degradar la calidad del agua. Este estudio midió el sedimento de suelo transportado, las diná micas en la microtopografía de la superficie del suelo, las clases de cobertura de arbustos leñ osos, zacates y suelo desnudo y la fertilidad del suelo, medida por los potenciales de mineralizació n del nitró geno para las respectivas clases de cobertura. Las mediciones se realizaron por un periodo de nueve añ os, durante el cual ocurrieron dos fuegos. En general, los efectos de dos periodos de sequía fueron iguales o mayores que los efectos del fuego en los parámetros medidos. Después del segundo fuego, en el sitio dominado por ''Grama grass'', el nú mero de ''Junipers'' vivos fue reducido significativamente (P , 0.001). El fuego consumió la cobertura de zacates y arbustos, creando un mayor potencial de transporte de sedimentos y perdidas en la elevació n de la superficie del suelo. La fertilidad bajo los arbustos fue significativamente mayor que la fertilidad del suelo bajo los zacates, esto se detectó en ambos sitios de pastizal, antes y después del primer fuego. Sin embargo, ese patró n tan significativo no se presentó después del segundo fuego en ninguno de los dos sitios por aproximadamente dos añ os. Esto sugiere ...

The effects of phenology on indirect measures of aboveground biomass in annual grasses

“…The temporal lag we quantified between the points of maximum greenness and maximum aboveground biomass (figures 1 and 2) is an important consideration for remote sensing-driven estimates of carbon dynamics in grassland ecosystems. As in the annual grassland communities we studied, Reeves et al (2006) found that the highest degree of association between NDVI and biomass occurred near peak (maximum) greenness. While this period may also be the period of maximum photosynthesis, as commonly assumed (Reeves et al 2006, S. Los, personal communication, 11 May 2007, Inoue et al 2008, photosynthesis continues for some time afterwards and contributes, along with some reallocation from belowground components of plants, to increasing canopy biomass, as evident in this study.…”

“…As in the annual grassland communities we studied, Reeves et al (2006) found that the highest degree of association between NDVI and biomass occurred near peak (maximum) greenness. While this period may also be the period of maximum photosynthesis, as commonly assumed (Reeves et al 2006, S. Los, personal communication, 11 May 2007, Inoue et al 2008, photosynthesis continues for some time afterwards and contributes, along with some reallocation from belowground components of plants, to increasing canopy biomass, as evident in this study. Thus, if one wishes to estimate total biomass production or maximum standing biomass in grasslands, there will necessarily be elevated uncertainty in NDVI-based estimates for the period after maximum greenness.…”

“…Potter et al 1999), when it is assumed that photosynthetic production (i.e. carbon fixation) is at a maximum (Reeves et al 2006, Inoue et al 2008, S. Los, personal communication, 11 May 2007. However, Scurlock et al (2002) demonstrated that, in grasslands, productivity estimates that accounted for senescent biomass could be as much as two to five times higher than those based on green biomass only.…”

Assessing the multi-resolution information content of remotely sensed variables and elevation for evapotranspiration in a tall-grass prairie environment

“…Quantitative estimates of biomass are now available as an operational product from the Moderate Resolution Imaging Spectroradiometer (MODIS). These estimates offer a reliable measure of the vegetation dynamics in space and time, with the finest resolutions being 250 m, and daily coverage (Reeves et al, 2006). There are, however, issues associated with assuming that remotely sensed vegetation indices are sufficient for quantifying the dynamics of carbon cycling when compared to field measurements (Scurlock et al, 2002).…”

Classification management for grassland using MODIS data: a case study in the Gannan region, China

“…In the LOOCV method, all data items, except 1, are iteratively used to build the predictive equation. A single one is held out during each iteration and is then used as validation data (Reeves et al 2006). The standard procedure for LOOCV involves the use of a single observation from the original sample as the validation data and the remaining observations as the training data.…”