The MODIS NPP (Net Primary Productivity) Product - Concept, Definition and Validation

摘要

One of the major components of the carbon cycle is the sequestration of carbon from CO_2 in the atmosphere at the surface of the Earth, land and sea. It is considerably more difficult to estimate the magnitude of this than to estimate the amount of carbon emitted by combustion processes and (possibly) by respiration. This talk will comprise two components. The first will be a general tutorial presentation of the MODIS NPP product and the second will be concerned with reporting on two of our own research projects. This talk is only concerned with land surfaces, not the sea, so we are only covering part of the sequestration component of the carbon cycle. We are very familiar with the use of the NDVI (Normalised Difference Vegetation Index) for studying land surfaces at continental or global scale; the ideas involved in the NDVI are very simple and the use of the NDVI is a major success story of satellite remote sensing. However, carbon sequestration is much more complicated because in this case we are trying to obtain quantitative values for biogeophysical parameters all over the land surface. The algorithm is not simple, as is evidenced by the fact that the relevant MODIS Algorithm Theoretical Basis Document (ATBD) for the Net Primary Productivity (NPP) product runs to 130 pages. I shall try to summarise this. MODIS produces regularly on a global basis a value of NPP on a pixel by pixel basis with pixels approximately 1 km square over the global land surface and, of course, it has been validated with a number of test sites. However, these test sites are mostly in mid-latitudes and our interest is in the validation of the product for tropical areas, since the tropics are of major importance in carbon sequestration. Two research projects will be described, both of which are being carried out in conjunction with the Remote Sensing Department of Universiti Teknologi Malaysia (UTM). Both involve making the best estimate one can from in situ measurements of the quantities of vegetative material produced and thence the amount of carbon which has been sequestered, subtracting the loss by respiration. Given the pixel size of 1 km2 it can be seen that this is not a trivial task. The first project is based on a test site within the Pasoh Forest Reserve in Negri Sembilan in the central part of peninsular Malaysia, preceded by some work on a Japanese test site. The Pasoh site consists of primary lowland mixed forest. This work is complete, but because the test site is so small, relative to the MODIS NPP pixel size, the results are limited to a very small number of pixels. Our second project is only in its very early stages, but is concerned with plantations where there is the possibility of much larger areas of homogeneous vegetation so that we can match up with a much larger number of MODIS pixels.