Modeling sediment and phosphorus losses in an agricultural watershed to meet TMDLs

Brent J. Dalzell, Prasanna H. Gowda, David J. Mulla

Research output: Contribution to journalArticlepeer-review

27 Scopus citations


This paper studies the effectiveness of alternative farm management strategies at improving water quality to meet Total Maximum Daily Loads (TMDLs) in agricultural watersheds. A spatial process model was calibrated using monthly flow, sediment, and phosphorus (P) losses (1994 to 1996) from Sand Creek watershed in south-central Minnesota. Statistical evaluation of predicted and observed data gave r2 coefficients of 0.75, 0.69, and 0.49 for flow (average 4.1 m3/s), sediment load (average 0.44 ton/ha), and phosphorus load (average 0.97 kg/ha), respectively. The calibrated model was used to evaluate the effects of conservation tillage, conversion of crop land to pasture, and changes in phosphorus fertilizer application rate on pollutant loads. TMDLs were developed for sediment and P losses based on existing water quality standards and guidelines. Observed annual sediment and P losses exceeded these TMDLs by 59 percent and 83 percent, respectively. A combination of increased conservation tillage, reduced application rates of phosphorus fertilizer, and conversion of crop land to pasture could reduce sediment and phosphorus loads by 23 percent and 20 percent of existing loads, respectively. These reductions are much less than needed to meet TMDLs, suggesting that control of sediment using buffer strips and control of point sources of phosphorus are needed for the remaining reductions.

Original languageEnglish (US)
Pages (from-to)533-543
Number of pages11
JournalJournal of the American Water Resources Association
Issue number2
StatePublished - Apr 2004


  • Agriculture tillage
  • Best management practices (BMPs)
  • Nonpoint source pollution
  • TMDL
  • Water quality


Dive into the research topics of 'Modeling sediment and phosphorus losses in an agricultural watershed to meet TMDLs'. Together they form a unique fingerprint.

Cite this