A precipitation-weighted landscape structure model to predict potential pollution contributions at watershed scales

Context

The impact of landscape structure—often described by landscape composition and configuration—on ecological processes is well-known. Appropriately quantifying landscape structure that critically affect nutrient processes within watersheds remains challenging.

Objective

A precipitation-weighted landscape structure model (LSM) was developed to predict nutrient concentrations at a large number of watersheds.

Methods

The LSM was developed based on the landscape location features including topography and precipitation within the watershed. The inequality function of Lorenz curve was used to quantify the spatial structure of different landscape types. The LSM was fitted and validated using the measurements of total nitrogen (TN) and total phosphorus (TP) in 132 watersheds. Regression models predicted the spatial patterns of TN and TP concentrations in 1578 watersheds of the Haihe River Basin, China.

Results

(1) Predictive models can explain 64 and 52% of the variation in total TN and TP, respectively. (2) Agricultural and residential lands served as nutrient sources. The contributions of agricultural land were 18 and 21% while those of residential land were 46 and 38% to TN and TP concentrations, respectively. (3) Grassland and forest land were nutrient sinks. Grassland had major contributions of 22 and 30% to TN and TP concentrations, respectively. The contributions of forest land were 7 and 11% to TN and TP concentrations, respectively.

Conclusions

The LSM focuses on the nutrient processes and is feasible to implement across a large number of watersheds. This study provides useful implications for quantifying landscape structure and predicting potential pollution under different landscape scenarios.