No abstract

A large number of water- and climate-related applications, such as drought monitoring, are based on spaceborne-derived relationships between land surface temperature (LST) and the normalized difference vegetation index (NDVI). The majority of these applications rely on the existence of a negative slope between the two variables, as identified in site- and time-specific studies. The current paper investigates the generality of the LST–NDVI relationship over a wide range of moisture and climatic/radiation regimes encountered over the North American continent (up to 60°N) during the summer growing season (April–September). Information on LST and NDVI was obtained from long-term (21 years) datasets acquired with the Advanced Very High Resolution Radiometer (AVHRR). It was found that when water is the limiting factor for vegetation growth (the typical situation for low latitudes of the study area and during the midseason), the LST–NDVI correlation is negative. However, when energy is the limiting factor for vegetation growth (in higher latitudes and elevations, especially at the beginning of the growing season), a positive correlation exists between LST and NDVI. Multiple regression analysis revealed that during the beginning and the end of the growing season, solar radiation is the predominant factor driving the correlation between LST and NDVI, whereas other biophysical variables play a lesser role. Air temperature is the primary factor in midsummer. It is concluded that there is a need to use empirical LST–NDVI relationships with caution and to restrict their application to drought monitoring to areas and periods where negative correlations are observed, namely, to conditions when water—not energy—is the primary factor limiting vegetation growth.

We present a dryland irrigation mapping methodology that relies on remotely sensed inputs from the MODerate Resolution Imaging Spectroradiometer (MODIS) instrument, globally extensive ancillary sources of gridded climate and agricultural data and on an advanced image classification algorithm. The methodology involves four steps. First, we use climate-based indices of surface moisture status and a map of cultivated areas to generate a potential irrigation index. Next, we identify remotely-sensed temporal and spectral signatures that are associated with presence of irrigation defined as full or partial artificial application of water to agricultural areas under dryland conditions excluding irrigated pastures, paddy rice fields, and other semiaquatic crops. Here, the temporal indices are based on the difference in annual evolution of greenness between irrigated and non-irrigated crops, while spectral indices are based on the reflectance in the green and are sensitive to vegetation chlorophyll content associated with moisture stress. Third, we combine the climate-based potential irrigation index, remotely sensed indices, and learning samples within a decision tree supervised classification tool to make a binary irrigated/non-irrigated map. Finally, we apply a treebased regression algorithm to derive the fraction of irrigated area within each pixel that has been identified as irrigated. Application of the proposed procedure over the continental US in the year 2001 produces an objective and comprehensive map that exhibits expected patterns: there is a strong east-west divide where the majority of irrigated areas is concentrated in the arid west along dry lowland valleys. Qualitative assessment of the map across different climatic and agricultural zones reveals a high quality product with sufficient detail when compared to existing large area irrigation databases. Accuracy assessment indicates that the map is highly accurate in the western US but less accurate in the east. Comparison of area estimates made with the new procedure to those reported at the state and county levels shows a strong correlation with a small bias and an estimated RMSE of 2500 km 2 , or little over 2% of the total irrigated area in the US. As a result, the future application of the new procedure at a global scale is promising but may require a better potential irrigation index, as well as the use of remotely sensed skin temperature measurements.

No abstract

Abstract. The availability of advanced very high resolution radiometer (AVHRR) time series of global shortwave data for the past two decades motivated many scientists to investigate interannual variability and trends in land surface conditions. For these studies the observed change in radiances due to two varying factors, namely, sensor responsivity and illumination conditions, must be known a priori because of the degradation of AVHRR shortwave channels and the orbit drift of afternoon spacecraft. The current work analyzes the behavior of global land AVHRR shortwave time series data for the last 12 years, processed using postlaunch calibration, and investigates their usefulness for the monitoring of global land surface processes. Its focus is on verifying the postlaunch calibrations for the AVHRR sensors on board NOAA 11 and 14. It is assumed that the NOAA 9 AVHRR calibration is correct so that the changing illumination effects can be parameterized based on its data. After accounting for the illumination effects, the residual trends in data, averaged over global deserts and rain forests, are attributed to calibration discrepancies. In particular, NOAA 11 calibration was found to yield only small residuals, whereas NOAA 14 calibration produced significant unrealistic global increase in both reflectances and vegetation indices. The artificial trends caused by the combination of calibration residuals and satellite-orbit drift should be removed to alleviate their misidentification as real trends in Earth's climate system and to make statistical studies of anomalies more reliable. This study draws attention to the above aspects of time series analysis with the available global AVHRR data and suggests ways to improve these data for interannual variability studies.