An improved cloud detection algorithm for monitoring agricultural growing conditions with NOAA AVHRR data in Texas

Near-real time access and large regional coverage are strong advantages of Advanced Very High Resolution Radiometer (AVHRR) data. Cloud contamination in almost each AVHRR scene influences solar reflectance data. Most environmental research requires cloud-free AVHRR data. An automated cloud detection algorithm was developed for the state of Texas. The time-dependent cloud detection algorithm effectively identified about 89% of cloud-contaminated pixels. Maximum normalized difference vegetation index (NDVI) composite was further applied to cloud-free AVHRR data to remove additional cloud contamination.; NDVI derived from AVHRR data has been successfully applied to global vegetation research. An observation of solar reflectance data for healthy vegetation produces a positive NDVI value. Negative NDVI values indicate the presence of clouds, snow, water, or bright non-vegetated surfaces. Weekly NDVI composites built from the cloud-free AVHRR data (named conditional NDVI composites) provided smooth temporal profiles during crop growing seasons, whereas traditional NDVI composites without the pre-process of cloud detection showed irregular patterns. This study suggested that cloud removal was important for composite products to be used for vegetation studies. Field data indicated that the sorghum needed 2 to 4 more weeks to reach heading in non-irrigated sites than in irrigated sites. The conditional NDVI composites showed that it took three more weeks for sorghum to attain maximum NDVI in non-irrigated sites than in irrigated sites. The conditional NDVI composites accurately depict in-field crop conditions. This study suggested that applying inexpensive NDVI data for crop monitoring could reduce fieldwork while achieving the same goals.; More National Oceanic and Atmospheric Administration (NOAA) satellites will be launched in the near future, which will provide more AVHRR data from different satellites for research. Since cloud contamination is an inevitable issue for AVHRR data users, combining AVHRR data from different satellites can help overcome cloud problems during compositing of daily AVHRR scenes using maximum NDVI selections. Results indicated that strong correlation was observed between NOAA-14 and NOAA-15 NDVI values derived from top-of-atmosphere (TOA) reflectance data. Potential use of merged NOAA-14 and -15 data for producing composites requires more study related to atmosphere correction and bi-directional reflectance distribution function (BRDF) for improving the accuracy of surface reflectance data.