Assessing seasonal river-wetland connectivity using remote sensing-based monitoring in tropical environments (datasets)

===== General Overview ====
These datasets correspond to the manuscript "Assessing seasonal river-wetland connectivity using remote sensing-based monitoring in tropical environments," by Dylan Irvine, Kaline de Mello and Porni Mollick, which is currently under review at Ecological Indicators. The work used relationships between river stage, rainfall and gap-filled MNDWI to determine (1) annual inundation, (2) seasonal inundation, and (3) connectivity between the Daly River (Australia) and its floodplain wetlands.

==== Datasets ====
Datasets and code presented here include:
- Google Earth Engine (GEE) scripts to obtain the MNDWI time series (GEE_codes.zip)
- An approach to produce catchment-averaged, daily climate (rainfall, temperature, etc.) datasets from the Queensland Government SILO database (ClimateData.zip)
- An additional dataset/ approach to obtain multiple realisations of the climate data for a region (ClimateDataMonteCarlo.zip)
-The flow and river stage data used (FlowData.zip)
- The resulting MNDWI time series for a collection of pixels on key transects that join wetlands/billabongs to the river (MNDWI_Pixel_Datasets.zip)
- A gap-filling process to address issues with cloud cover (KalmanFilterMethod.zip)
- An approach to identify river stages where wetting or drying occurs, and to identify these by water year (Sep-Aug) (ApplyMNDWI_Stage_Threshold.zip)

Folders are set up to be self-contained; however, the various input files were constructed using a combination of Python and Excel. i.e., the approach is demonstrated here, but the files do not necessarily present a pure workflow from raw data to the final analyses (due to the intermediate steps to prepare data files).

==== Manuscript abstract ====
Understanding the timing of river-floodplain wetland connection is critical for anticipating ecological risks, including aquatic fauna strandings. In the wet–dry tropics of northern Australia, these risks may intensify due to climate change and water extraction. We combined Sentinel-2-derived modified normalised difference water index (MNDWI), river stage, and rainfall data to monitor inundation dynamics and connectivity between the Daly River (Australia) and three permanent wetlands that act as refugia for aquatic species. We assess annual flood frequency (2018–2025), monthly inundated area, and their relationships with rainfall and river stage. Data gaps due to cloud cover were gap-filled using a random walk model with Kalman filtering and smoothing. Gap-filled MNDWI enabled the detection of spatiotemporal wetness patterns along transects connecting the wetland to the river. Results reveal large interannual variability in inundation, with 2018–2019 and 2019–2020 exhibiting low persistence and extent of flooding, while 2023–2024 showed widespread and prolonged inundation. Connectivity duration differed among transects(6—112 days). We identify stage thresholds (m) for disconnection as an indicator of river-wetland connectivity, with first disconnection dates varying between February—July, depending on the transect. We also derive three pixel-based hydrological indicators: first wetting day, last drying day, and seasonal duration of wet conditions (days yr⁻¹). The strength of relationships between inundation and predictors supports the use of these readily available datasets for forecasting disconnection timing. We provide a practical approach to inform aquatic biodiversity conservation planning measures that can be readily adapted to other floodplain systems.