Sea level rise (SLR) will reshape coastal ecosystems and human communities across multiple scales. Coastal landscapes are linked socio-ecological systems, and whether natural ecosystems migrate upslope as a response to SLR is ultimately determined by the interplay between ecological responses to SLR and human decision-making. Understanding these processes is important to identify potential hotspots of change and promote land-use adaptation to climate change and SLR in the coming decades. Here we focus on how SLR influences the land cover dynamics between natural ecosystems (marshes, tidal flats) and working lands, which we define as agriculture and forestry land uses, on the U.S. mid-Atlantic coastal plain. High resolution remote sensing data is critical for detecting changes in coastal landcover over timeframes that are relevant for connecting human decision-making to SLR-induced land changes.

Our objectives are: 1. Identify where SLR-caused landcover change and salinization on working lands is occurring and whether this change can be detected by satellite remote sensing on a yearly time scale. 2. Estimate whether the proximity or severity of SLR-caused coastal landcover change influence landowner decisions. 3. Estimate whether coastal landowner decisions influence land cover of natural ecosystems.

We will target the transition zone between marshes and upland agriculture and forestry lands in selected focal areas from New Jersey to Georgia (a region with a high proportion of working lands near coastal ecosystems and shallow topography). We are interested in two main processes: 1. areas of land cover change in the transition zone, which are larger signals that can be detected via Sentinel-2 (in combination with historic change from Landsat), and 2. change in land condition due to SLR and salinization that may lead to a change in land use. We will assess both the magnitude and periodicity of the spectral response through Fourier curve fitting of the Sentinel-2 response combined with very high-resolution imagery (i.e., NAIP for historic analysis and MAXAR for more recent change), that will be combined in a machine learning model to identify degradation in the wetland/upland transition zone. After model identification of areas with a range of land cover changes, we will identify landowners for those pieces of land (using the ParcelPoint dataset of landowners), and conduct targeted surveys to identify how the degree of change influences behaviors and decisions, as well as what other factors modulate or constrain those decisions (e.g., tax rates, market value for products). We will integrate both components (remote sensing model and analyses of landowner behaviors and decisions) in a predictive agent-based model (ABM), that we will validate with withheld data. We will create scenarios of SLR and economic pressures on landowners and project ABM estimates of relationships between landcover and landowner decisions to predict the type and degree of landcover change.

SLR is a pressing societal issue in this region – the U.S. mid-Atlantic is experiencing greater rates of SLR (3-4 mm/yr) compared to the global average (1-2 mm/yr) because local land subsidence exacerbates climate change-induced SLR. Landowners have several potential options for how to respond: by retreating (e.g., selling agriculture or forestry lands to housing developers), resisting (e.g., installing infrastructure to forestall inundation on productive land), adapting to changes (e.g., switching land use to accommodate flooding and salinization from SLR), or doing nothing (i.e., allowing inundation and land cover change to occur). These decisions will have consequences for both the upland land use, as well as the landcover of the adjacent natural system (e.g., sea walls typically lead to degradation of adjacent natural ecosystems). By making these social-ecological connections, we will be able to make better predictions of how land cover will change as a function both of direct SLR impacts and human responses to them.