Validation of a Pathfinder Satellite Drought Index
PI: Dr. Lloyd P. Queen, School of Forestry, University of Montana
INTRODUCTION
As part of NASA's Mission to Planet Earth, science teams are developing algorithms to generate standard sets of terrestrial science products. These algorithms will convert raw data streams from sensors such as MODIS into value-added image maps that are to be applied to resource investigations. Working with the MODLAND Science Team (who provided 20 days of summer salary), during the summer of 1997 I developed a prototype MODIS algorithm for a surface resistance product that is designed to be run on AVHRR weekly composites. Surface resistance is a biophysical term that considers the partitioning of sensible and latent heat fluxes over vegetated surfaces. Variations in resistance show strong correlations to moisture stress as a function of surface temperature. The existing algorithm uses vegetation indices and surface temperature as derived from multi-channel AVHRR data to derive image maps that are labeled into an ordinal scale of moisture stress or "drought condition."
Initial work during the June-September period of 1997 has allowed the Satellite Drought Index (SDI) algorithm to be completed and the processing stream (data flow) established, but the product has not yet been validated. The original development and application of this logic (Nemani and Running 1989) was limited to a small geographic area and the research was not extended over longer time intervals. The pathfinder algorithm that has been developed needs to be validated for three reasons. First, the logic has not been tested at large time and space scales. Second, because these information products have never before been systematically generated, they must be cross-validated to other image, map, and model products to examine their information content and consistency. And third, the logic behind the pathfinder algorithm will yield not only an historical record for EOS-based satellite drought maps, but this understanding will assist in development of the MODIS post-launch resistance product algorithm. The goal of the proposed research is to develop and apply a validation strategy for this model.
OBJECTIVES AND OUTCOMES
The proposed work consists of three objectives, all directed to comparing performance of the satellite drought index to existing methods of drought monitoring or characterization. The first objective will be to make historic runs of the code to develop a database of "long-term normal" conditions. The current NASA pathfinder data set for the AVHRR will allow the SDI to be propagated back in time to May 1989 for the eleven western states (the scale of published literature is on the order of 10,000 hectares) . A benefit of back-propagation is that I will be able to develop a derivative index based on deviations from the found normal. This will also make the index insensitive to biome stratification, a problem the algorithm currently suffers from. The second objective will compare spatial and thematic variation in the SDI to existing drought indices such as Palmer's drought severity index, Haines' drought index, the Keetch-Byram drought index (currently a part of the National Fire Danger Rating System or NFDRS) and the Standardized Precipitation Index. These indices share some of the logic of the SDI, but fail to address the physical aspect of vegetation stress. This second objective directly addresses the NFDRS as a potential end- user of the SDI. The NFDRS currently uses AVHRR vegetation index data to characterize surface condition. If the SDI proves to have a stronger biophysical rationale and it's information content can be verified, it will be considered for adoption and use by the NFDRS. The final objective involves deriving mathematical models of canopy surface resistance and the inverse stomatal conductance. While the current algorithm is efficient and error free, parallel runs of an ecosystem process model (BIOME-BGC) will be made so that I can derive the mathematical relationships between the SDI and canopy surface resistance and stomatal conductance. These differential equation models will then be embedded in the pathfinder algorithm code to produce dimensioned biophysical variables rather than the current ordinal-scale output.
The results of the proposed research are two-fold. First, the work will provide a proven method for monitoring significant terrestrial biophysical variables and fire/drought danger with high temporal and moderate spatial resolution. Second, the proposed research will yield not only publishable results, but also a significant comparative advantage in future work with EOS/MTPE sensors and data products. This will translate to increased cccompetitiveness for future applications research support.
REFERENCES
Nemani, R. and S. Running. 1989. Estimation of Regional Surface Resistance to Evapotranspiration from NDVI and Thermal-IR AVHRR Data. Journal of Applied Meteorology. 28(4): 276-283.
Montana Space Grant Home