Quantifying risk dynamics on Rawa Pening floodplain using optical images gathered by satellite and unmanned aerial vehicle

The floodplain of lake Rawa Pening, experience spatio-temporal dynamics due to regime shifts of wet and dry season as well as a more persisten land use changes in the upland area. The high yield of rice agriculture in the floodplain has also been bothered by additional entity rooted on the socio-economic value of the plain. Our research focused on floodplain in the vicinity of the Torong River, Banyubiru District that recently incurred river normalization project. Compare to the rest eight  catchments delivering effluents into the lake, we assume normalization will change sediment budget, in way the dynamics can be captured well by detailing imagery acquired from Unmanned Aerial Vehicle (UAV) photography.

Land use change is observed using high temporal resolution of optical satellite imagery and the verification using UAV images. Sentinel-2 optical imagery is used for the macrozonation. Because of the high temporal resolution, we eliminate images with cloud interference exceeding the specified threshold while assuring data continuity. At time when Sentinel-2 is planned to pass over, we also acquire UAV photos of different heights aimed to detail reality mapping of the area. To get land productivity, we use statistical information and semi-structured interviews of randomly selected samples for each land use class.

Our initial results using longer period Google Earth images showed both extreme and gradual changes of land use, partly due to irregular temporal captures. Sentinel-2 is available in shorter historical period providing denser images every 5 days. At the same capture time, UAV capture images to opens potentials for further color manipulations matching the productivity. For the moment, our investigation on land productivity still relied on manual delineation of straight skeleton visible in both approaches. High productivity of ricefield in the floodplain area also still relied on semi-structured interviews and statistical reports by village adminstrations. With the constraints, risk of land use change observed using current satellite images and UAV accords on the manual delineation process. As a result, we found Sentinel-2 images is sufficient to predict risk changes particularly for fish culture and tourism, while spatial ricefield productivity using satellite and UAV images still require complex experimentation on color spectrum and operational acquisition height of the UAV.