Nondestructive Estimation of Standing Crop and Fuel Moisture Content in Tallgrass Prairie☆

ABSTRACT Accurate estimation of standing crop and herbaceous fuel moisture content (FMC) are important for grazing management and wildfire preparedness. Destructive sampling techniques have been used to accurately estimate standing crop and FMC, but those techniques are laborious and time consuming. Existing nondestructive methods for estimating standing crop in tallgrass prairie have limitations, and few studies have examined nondestructive estimation techniques for FMC in this environment. Therefore, our objective was to develop robust models for nondestructive estimation of standing crop and FMC in tallgrass prairie. We calibrated and validated stepwise multiple linear regression (SMLR) and artificial neural network (ANN) models for standing crop and FMC using data collected in tallgrass prairies near Stillwater, Oklahoma. Day of year (DOY), canopy height (CH), Normalized Difference Vegetation Index (NDVI), and percent reflectance in five wavelength bands were candidate input variables for the models. The study spanned two growing seasons and nine patches located within three pastures under patch burn management, and the resulting data set with >3 000 observations was split randomly with 85% for model calibration and 15% withheld for validation. Standing crop ranged from 0 to 852 g m-2, and FMC ranged from 0% to 204%. With DOY, CH, and NDVI as predictors, the SMLR model for standing crop produced a root mean squared error (RMSE) of 119 g m-2 on the validation data, while the RMSE of the corresponding ANN model was 116g m-2. With the same predictors, the SMLR model for FMC produced an RMSE of 26.7% compared with 23.8% for the corresponding ANN model. Thus, the ANN models provided better prediction accuracy but at the cost of added computational complexity. Given the large variability in the underlying datasets, the models developed here may prove useful for nondestructive estimation of standing crop and FMC in other similar grassland environments.

[1]  Weixing Cao,et al.  Monitoring Leaf Nitrogen Status in Rice with Canopy Spectral Reflectance , 2004, Agronomy Journal.

[2]  D. Riaño,et al.  Combining NDVI and surface temperature for the estimation of live fuel moisture content in forest fire danger rating , 2004 .

[3]  E. Bork,et al.  Integrating LIDAR data and multispectral imagery for enhanced classification of rangeland vegetation: A meta analysis , 2007 .

[4]  R. Gillen,et al.  Tiller defoliation patterns under short duration grazing in tallgrass prairie. , 1990 .

[5]  J. Paruelo,et al.  ANPP ESTIMATES FROM NDVI FOR THE CENTRAL GRASSLAND REGION OF THE UNITED STATES , 1997 .

[6]  R. A. Evans,et al.  Plant Height Times Ground Cover versus Clipped Samples for Estimating Forage Production1 , 1958 .

[7]  Gottfried Mandlburger,et al.  Beyond 3-D: The New Spectrum of Lidar Applications for Earth and Ecological Sciences , 2016 .

[8]  Dirac Twidwell,et al.  Surface Fuel Sampling Strategies: Linking Fuel Measurements and Fire Effects , 2009 .

[9]  Jeffrey G. Arnold,et al.  Model Evaluation Guidelines for Systematic Quantification of Accuracy in Watershed Simulations , 2007 .

[10]  M. Gardner,et al.  Confidence intervals rather than P values: estimation rather than hypothesis testing. , 1986, British medical journal.

[11]  R. King,et al.  Protocol for monitoring standing crop in grasslands using visual obstruction. , 2000 .

[12]  D. Engle,et al.  Application of the fire–grazing interaction to restore a shifting mosaic on tallgrass prairie , 2004 .

[13]  Brad Warner,et al.  Understanding Neural Networks as Statistical Tools , 1996 .

[14]  F. M. Danson,et al.  Estimating live fuel moisture content from remotely sensed reflectance , 2004 .

[15]  D. Riaño,et al.  Estimation of live fuel moisture content from MODIS images for fire risk assessment , 2008 .

[16]  D. Pontier,et al.  Assessing the performance of NDVI as a proxy for plant biomass using non-linear models: a case study on the Kerguelen archipelago , 2009, Polar Biology.

[17]  D. Riaño,et al.  Estimation of fuel moisture content from multitemporal analysis of Landsat Thematic Mapper reflectance data: Applications in fire danger assessment , 2002 .

[18]  K. Eskridge,et al.  Efficiency of different quadrat sizes and shapes for sampling standing crop , 1994 .

[19]  Chi Zhang,et al.  The spatiotemporal patterns of vegetation coverage and biomass of the temperate deserts in Central Asia and their relationships with climate controls , 2016 .

[20]  J. A. Schell,et al.  Monitoring vegetation systems in the great plains with ERTS , 1973 .

[21]  Mark Beale,et al.  Neural Network Toolbox™ User's Guide , 2015 .

[22]  N. M. Idaikkadar,et al.  CHAPTER 10 – Census of Agriculture , 1979 .

[23]  R L Pearson,et al.  Hand-held spectral radiometer to estimate gramineous biomass. , 1976, Applied optics.

[24]  M. Gevrey,et al.  Review and comparison of methods to study the contribution of variables in artificial neural network models , 2003 .

[25]  Indrajeet Chaubey,et al.  Artificial Neural Network Approach for Mapping Contrasting Tillage Practices , 2010, Remote. Sens..

[26]  A. Huete.  A soil-adjusted vegetation index (SAVI) , 1988 .

[27]  P. Dixon,et al.  Spectral Reflectance as a Covariate for Estimating Pasture Productivity and Composition , 2005 .

[28]  R. Lyman Ott.,et al.  An introduction to statistical methods and data analysis , 1977 .

[29]  K. Olson,et al.  RADIOMETRY FOR PREDICTING TALLGRASS PRAIRIE BIOMASS USING REGRESSION AND NEURAL MODELS , 1998 .