Assessing Floodplain Forests: Using Flow Modeling and Remote Sensing to Determine the Best Places for Conservation

Mature and diverse floodplain forests are among the world's most diminished ecosystems and conservationists need a rapid method to identify the best remaining examples of these systems. Because large rivers and their dynamics bind the floodplain together, the method most go beyond simple inventory of remnant patches to evaluate flood processes and identify constraints in the surrounding watersheds. We develop such a method for a three million hectare watershed in New England using a combination of data types to evaluate key attributes of floodplain systems. Riparian and floodplain communities were modeled using a GIS analysis of river valley topography and riverine processes, and floodplain forest occurrences were identified in a classification and regression (CART) analysis. Current flooding was verified using overlays of remotely sensed imagery of spring and fall water levels. We evaluated the intactness of the floodplain occurrences using ratios of upstream dam storage to annual runoff, the length of the connected stream network, and the naturalness of surrounding land cover. Field-assigned ranks of forest quality were correlated with the Occurrence size, percent verified flooding, and percent natural cover. Predicted quality ranks reinforced the importance of these factors. Results indicate that the twenty top-ranking streams collectively contain 75 high quality areas Suitable for floodplain forest restoration and conservation. Independent verification of these areas strongly corroborated our results.