Degree Type

Dissertation

Date of Award

2019

Degree Name

Doctor of Philosophy

Department

Agronomy

Major

Agricultural Meteorology

First Advisor

Brian K. Hornbuckle

Abstract

NASA’s Soil Moisture Active Passive (SMAP) satellite utilizes passive observations of L–band (f = 1.41 GHz, λ = 21 cm) brightness temperature to estimate surface soil moisture at a spatial scale of 33 km approximately once per day in the U. S. Corn Belt. These observations have the potential to improve weather forecasting models, increase agricultural productivity, and provide decision support for flood and drought management. However, SMAP Level 2 Soil Moisture (L2SM) performs poorly in croplands validation sites such as the South Fork Iowa River (located in central Iowa); we hypothesize that this is due to the use of generic croplands parameterizations during SMAP L2SM retrieval. We analyzed the ancillary inputs to the τ − ω retrieval model to determine if they could cause the observed seasonal component to SMAP L2SM bias and unbiased RMSE. After implementing a modified surface temperature, in which the SMAP–reported value is divided by the bias correction factor K = 1.02 to be more realistic for the South Fork, we identified roughness and vegetation to be the most likely sources of error in soil moisture retrieval.

At L–band, changes in soil surface roughness and vegetation produce the same effect on emissivity, leading to an inability to disentangle roughness–vegetation effects within L2SM retrievals. We utilize our conceptual knowledge of roughness–vegetation patterns, combined with South Fork in situ observations of soil moisture and temperature, to produce the first temporally–dynamic retrievals of HR (model roughness parameter) at satellite–scale. These are consistent with both the wide range of literature values and sampling of physical roughness conducted during the SMAPVEX16–IA campaign. However, when this roughness–vegetation concept is applied to retrieving L2SM the previously observed errors are not mitigated as initially hypothesized. We suggest that the next step towards improving SMAP L2SM in the U. S. Corn Belt includes adopting a first–order radiative transfer model to capture scattering that is currently considered to be negligible.

DOI

https://doi.org/10.31274/etd-20191121-0

Copyright Owner

Victoria A. Walker

Language

en

File Format

application/pdf

File Size

124 pages

Share

COinS