Validation of a sediment connectivity binary model improved with a probabilistic approach in the effect of prairie strips.

Prairie strips (PS) have demonstrated remarkable effectiveness in sediment control and partially interrupting sediment connectivity (SC). Research projects such as STRIPS have devoted two decades to investigate this strategy and to promote associated and additional ecosystem services at agricultural landscapes.
The experimental area was within the Neal Smith National Wildlife Refuge (Iowa, USA). Since 2007, an assessment involving 12 watersheds (3 controls, 9 treatments) was conducted to evaluate the advantages of incorporating PS into rowcrop (Corn-Soybean rotation). Treatments consisted in altering the PS number and area. Helmers et al. (2012) exhibited sediment trapping efficiencies (STE) averaging above 90%.
To the extent of our knowledge, current models attempting to calculate STE are complex. Mahoney et al. (2018) parameterized the probability of SC with a binary model, combining various individual probabilities. One of these probabilities is buffer disconnectivity, where buffers interrupt runoff and disconnect the entire upstream area. If we consider a PS as a buffer, sediment will not pass when the PS is present, and vice versa. Another approach is the one from Muñoz et al. (2023), who analysed STE in vegetation strips using a probabilistic approach, finding a wide range of variation in STE, from -109 to 100%. 
This communication presents the integration of this approach to the previous model from Mahoney et al. (2018), going from a binary model in the buffer disconnectivity probability to a model with a range of values between -∞ and 1. Assuming a direct correlation between sediment load and connected pixels, we ran the model across the experimental watersheds during a period of 7 years to validate the result of SC in the model. The results of the model by events and in each watershed were poor due to the variability between precipitation and sediment load. However, considering a weighted arithmetic mean with the rainfall for sediment load and connected pixels, good positive relationships emerged between average sediment load and average connected pixels when the model was applied individually to each watershed for the whole period. As a final part, we extended the model to the set of watersheds, where the correlation was absent.
Nevertheless, the combination of both approaches allows one to factor the probability of STE for specific management practices without significant added complexity, resulting in a strong fit for small watersheds with management with PS. 

Acknowledgement: Work was funded by Spanish Ministry of Science and Innovation (PID2019-105793RB-I00), project SCALE (EUHorizon2020 GA 862695), and a predoctoral fellowship (PRE2020-093846). We also acknowledge and appreciate the numerous funders and researchers of previous STRIPS Project investigations.

References
Helmers et al. (2012). Sediment removal by prairie filter strips in row‐cropped ephemeral watersheds. Journal of Environmental Quality, 41(5), 1531-1539. 
Mahoney et al. (2018). Watershed erosion modeling using the probability of sediment connectivity in a gently rolling system. Journal of Hydrology, 561, 862–883. 
Muñoz et al. (2023). Appraising trapping efficiency of vegetative barriers in agricultural landscapes. Strategy based on a probabilistic approach based on a review of available information. International Soil and Water Conservation Research.