11:00 〜 13:00
[HGG01-P03] 中央ケニア半乾燥地域の土地被覆変化と土壌浸食
キーワード：リモートセンシング、土壌浸食（土壌侵食）、ケニア、NDVI、NDSI、MNDWI
The objectives of our presentation are to discuss 1) the correlation between NDVI/NDSI and climatic elements, 2) the NDVI/NDSI trends over the past decades (1984 to 2021), and their relation to soil erosion, and 3) the availability of MNDWI, as a soil moisture indicator. The investigated area, the Il Polei sub-location (N 0°21'56", E 37°04'32") is situated in the semi-arid pastoral area, Laikipia North sub-county, central Kenya, which has an altitude of 1,750 to 1,850 m. Recent situation on natural resource use in this area was continuously reported in this session (Otsuki et al., 2018, 2020, 2021, Konno and Otsuki, 2020, etc.). We classified land cover of the investigated area mainly by using Landsat 4 to 8 images, and PlanetScope images with a spatial resolution of 3 m were also used in the process of land cover clustering.
Short-term increase and decrease of NDVI/NDSI were largely controlled by climatic elements, especially precipitation. In this area, precipitation is a factor controlling plant production due to the relatively small amount of subsurface water. Changes in precipitation, however, do not immediately affect plant production, but are highly correlated with the amount of vegetation after about one month. In addition, NDVI and NDSI show an inverse correlation trend with each climatic element, and this correlation indicates a direct link between shrinking vegetation area and expanding bare ground.
According to the long-term trends of NDVI/NDSI, pixels with predominant erosion tendencies over the past 40 years are intensively distributed in the area. The eroded cites chiefly correspond to gully, ephemeral river, and bare ground pixels excluding the gully and river ones. The pixels which show a significant increase in NDSI also tend to show a significant decrease in NDVI, and the trend of NDSI is more remarkable than that of NDVI. From NDVI and NDSI, the intensity of erosion tendency (absolute value of regression coefficient) is obviously A greater in the ephemeral river group among the three types of pixel groups. The gully and bare ground pixels suggest cites where down-cutting erosion, and planner erosion by sheetwash and rill flow, occur, respectively. The present ephemeral river is not considered to have undergone significant erosion, and its erosion tendency may also indicate recent anthropogenic impacts such as sand-harvesting (Otsuki et al., 2016, 2021, etc.).
The weakening of soil strength due to increased soil moisture is an important factor particularly in gully erosion; however, no correlation exists in this area, between the MNDWI and the antecedent 24-hour and 6-hour cumulative precipitation. Despite the small seasonal variation of MNDWI, there is an increase or decrease in MNDWI even in the absence of rainfall events. From these results, it is difficult to understand the quantitative changes in soil moisture caused by precipitation from MNDWI. In contrast, MNDWI shows a positive correlation with cloud cover at the time of imaging. Although clouds and cloud shadows are masked by QA (Quality Assessment) as a pre-processing step, the masking is inferred to be insufficient due to the increase in cloud cover, resulting in an increase in MNDWI, and further study is also necessary on these points.
Short-term increase and decrease of NDVI/NDSI were largely controlled by climatic elements, especially precipitation. In this area, precipitation is a factor controlling plant production due to the relatively small amount of subsurface water. Changes in precipitation, however, do not immediately affect plant production, but are highly correlated with the amount of vegetation after about one month. In addition, NDVI and NDSI show an inverse correlation trend with each climatic element, and this correlation indicates a direct link between shrinking vegetation area and expanding bare ground.
According to the long-term trends of NDVI/NDSI, pixels with predominant erosion tendencies over the past 40 years are intensively distributed in the area. The eroded cites chiefly correspond to gully, ephemeral river, and bare ground pixels excluding the gully and river ones. The pixels which show a significant increase in NDSI also tend to show a significant decrease in NDVI, and the trend of NDSI is more remarkable than that of NDVI. From NDVI and NDSI, the intensity of erosion tendency (absolute value of regression coefficient) is obviously A greater in the ephemeral river group among the three types of pixel groups. The gully and bare ground pixels suggest cites where down-cutting erosion, and planner erosion by sheetwash and rill flow, occur, respectively. The present ephemeral river is not considered to have undergone significant erosion, and its erosion tendency may also indicate recent anthropogenic impacts such as sand-harvesting (Otsuki et al., 2016, 2021, etc.).
The weakening of soil strength due to increased soil moisture is an important factor particularly in gully erosion; however, no correlation exists in this area, between the MNDWI and the antecedent 24-hour and 6-hour cumulative precipitation. Despite the small seasonal variation of MNDWI, there is an increase or decrease in MNDWI even in the absence of rainfall events. From these results, it is difficult to understand the quantitative changes in soil moisture caused by precipitation from MNDWI. In contrast, MNDWI shows a positive correlation with cloud cover at the time of imaging. Although clouds and cloud shadows are masked by QA (Quality Assessment) as a pre-processing step, the masking is inferred to be insufficient due to the increase in cloud cover, resulting in an increase in MNDWI, and further study is also necessary on these points.