The monitoring of paddy fields in tropical region, where cropping patterns vary according to landforms and water conditions, requires spatio-temporal analysis using multi-temporal/seasonal satellite images. However, the large amount of rainfall in the humid tropics in Asian region increases the difficulty of obtaining cloud free satellite images. This study used images obtained by ALOS PALSAR, which is an all-weather sensor, in combination with optical pan-sharpened image from an ALOS AVNIR-2 and PRISM, which provides excellent spectral and spatial resolution, to evaluate paddy field detection in Cianjur District, an outstanding rice-producing area in Java Island, Indonesia.The land cover of the study area could be classified with an overall accuracy of 72.3% and an overall Kappa coefficient of 0.53. Above all, paddy fields could be well classified with Kappa coefficients between 0.73 and 0.91. The large scale paddy fields could be classified with Kappa coefficients above 0.81, on the other hand, complicated-shaped small scale paddy fields in the Gute volcanic mountain slope could be classified with a Kappa coefficient of lower than 0.76.The rice cropping pattern was investigated by temporally totaling the occupancy of water-covered areas. A semiannual crop pattern was primarily adopted in the basins where large scale and uniform paddy fields had been developed, except during the dry season between June and September. In contrast, terrace paddy fields exhibited unique seasonal patterns according to water conditions. These results essentially agreed with the cropping calendar that had been created based on the interviews survey. Therefore, the use of high-resolution optical sensor images obtained by ALOS AVNIR-2 and PRISM, together with temporal images obtained by PALSAR, could classify land cover and paddy fields with high spatio-temporal resolution in tropical regions.
The increased incidence of locally intense rainfall events in recent years hasprompted concern about the increased environmental loading due to elevated levels of suspended solids (SS) and nutrient salts such as phosphate in runoff from sloping farmlands to local watershed drainage systems as well as rivers and lakes in the catchment basin. Experimental sloping lysimeters have been generally utilized to collect samples for SS and nutrient salt runoff; however, sampling performed with small lysimeters does not accurately reflect the environmental conditions. On the other hand, on-site measurements made in remote farm fields are often unreliable because the researcher can easily miss crucial samplings in a runoff event due to the difficulty of predicting sudden, intense rainfalls. Equipment at remote sampling sites is also difficult to maintain, and technical problems with equipment can lead to reduced data quality. For example, conventional techniques using a partial flume with a water-stage recorder are often unreliable due to interference of sediment deposition and/or floating refuse at the sensor. Here we describe a practical, automated system that simultaneously records color images of the field surface during rainfall events, and measures the amount of runoff and collects samples of runoff water for subsequent chemical analyses. We tested the system at two fields with Andosol soil and sloping topography over the 2009 growing season for sweet corn in Tsukuba, Ibaraki, and over the 2009 and 2010 growing seasons for cabbage in Tsumagoi, Gumma, both in eastern Japan. The equipment successfully captured a series of photographs at 1-min intervals of runoff at the bottom of each sloping plot during rainfall events, day and night. The photographs were used to accurately determine the beginning and peak times of runoff, and this data was used to analyze the rainfall-runoff dynamics. Combining photographs and the field measurements of runoff water quantity with SS and nutrient salts has enabled us to better understand the impact of periods of high runoff.