Project Overview |

Images |

Products |

Publications |

Home |

• What lands in the contrasting Kansas regions are most susceptible to landuse change, and how does this change relate to various impacting factors such as availability of water, physical landscape condition (soil and surface geology), and changing climate patterns?
  
  
• How do annual and seasonal variations in precipitation and temperature patterns influence our ability to map and monitor landuse in the three regions, and can we integrate spectral, edaphic, climatic and other environmental factors into our mapping and monitoring procedures to produce dependable results and gain valuable input relative to global climate modeling?
  
  
• How can remotely sensed data be used to model climate variation that is the major factor influencing year-to-year landuse classification accuracy?
  
  
• How are regional landuse practices changing and what are the factors influencing change?
  
  
• Can climatic patterns be grouped into unique classes based on their influence on plant phenology and ability to significantly alter spectral classification of landuse?
  
  
• What is the relationships between remotely sensed measurements and various crop types, land conditions (e.g., coal, zinc or lead mined lands), productivity and water usage, and how do these relationships change from year-to-year, and from region to region;
  
  
• What are the spectral patterns associated with the phenological characteristics of Conservation Reserve lands throughout the four regions, and how can these lands be discriminated from natural rangelands, and how do we determine whether a tract of land has been converted back to a shortgrass prairie, a tallgrass prairie, or an agricultural landuse?
  
  
• How might data from various sensors (i.e., Landsat, SPOT, NAPP, radar and other remote sensing systems) be integrated into a model for addressing land use questions in the regions; and related to this;
  
  
• What remote sensor data are best suited for predicting rural land use at varying scales in the four regions?
  

1. Identify the grass land cover type from other vegetation types;
   
   
2. Distinguish the six grass land cover types (cool season CRP, cool season grazed, cool season hayed, warm season CRP, warm season grazed, warm season hayed) by multi-seasonal satellite images, Radar images, and combining with field data;
   
   
3. Figure out how landuse changes from year to year;
   
   
4. Create a GIS database include topographic condition, soil information, precipitation data, reflectance value, and so on for study area;
   
   
5. Figure out how other factors (precipitation, soil, etc.) affect classification;
   
   
6. Analyze the effects of CRP to preserve native prairie;
   
   
7. Develop a regression equation for predicting biomass and cover for grasslands in this area.

Field data collection

Field data were collected during the late spring and mid-summer of 1997. Sixty-nine field sites were randomly selected from the known grassland areas of Douglas County. The field sampling period was designed to coincide with the May and July satellite overpasses as closely as possible (approximately two weeks before and after the overflight dates). Ocular estimates of forb and grass cover (%) (n = 69) were collected in both May and July, and aboveground biomass (hereafter, "biomass") (g/m2) (n = 40) was collected between May and July. Cover and biomass were measured within five 0.25 m2 quadrats in each 90 m x 90 m study plot. Due to time constraints, biomass could only be collected at 40 of the 69 sites. Plant moisture content (g/m2) (n = 40) was calculated from the difference between wet and dry biomass. The geographic location of each sample site was determined using a Global Positioning System (GPS) that recorded positional accuracy within 15 m. Figure 3a shows the general geographic distribution of the warm and cool season grassland sample sites throughout the study area. Figure 3b shows a Landsat TM band 4 (near infrared) of the Douglas County study area. Areas that are brightest on the image have greater amounts of green photosynthetically active plant material (green biomass).

Spectral data collection

Landsat TM: Three nearly cloud-free Landsat 5 TM images (path 27, row 33) that covered the study area were acquired for the following dates, May 15, July 2, and September 4, 1997. After each image was examined for sensor errors (i.e., striping, banding, line dropout), the digital numbers were then converted to reflectance using the method described by Markham and Barker (1986). The reflectance values were then adjusted for atmospheric scatter using the Improved Dark Object Subtraction technique described by Chavez (1988). The images were corrected for geometric error by first transforming the July image to a Universal Transverse Mercator (UTM) projection. The geometric transformation equation was computed using 38 ground control points that produced a final RMS error of < 0.35 pixels (<10.5 m). The May and September images were then georegistered to the ground-rectified July image after the method described by Jensen (1996, pp. 125). The spectral values for each pixel were interpolated using a nearest neighbor resampling approach and the data were output to a 30 x 30 m pixel size. During the geometric correction process, the images were clipped to the Douglas County political boundaries, and the thermal bands were eliminated from each image resulting in an 18-band dataset (6 bands for each date). The six Landsat TM bands and their associated wavelengths that were used in this study include band 1 (blue-green, 450-520 nm), band 2 (green, 520-600 nm), band 3 (red, 630-690 nm), band 4 (near infrared, 760-900 nm), band 5 (middle infrared, 1550-1750 nm), and band 7 (middle infrared, 2080-2350 nm).

VIs were computed to determine whether their use improved our ability to characterize biophysical conditions of the various grassland types in the study area. For this purpose, the following VIs were computed: NDVI, the Tasseled Cap indices (BI, GVI, WI) (Crist, 1986), the green ratio (GR), and middle infrared ratio (MR). These indices were calculated using the following formulas:

NDVI = (TM4-TM3) / (TM4+TM3) [1]
GR = TM4 / TM2 [2]
MR = TM4 / TM5 [3]
BI = 0.2909(TM1)+0.2493(TM2)+0.4806(TM3)+0.5568(TM4)+ 0.4438(TM5)+0.1706(TM7) [4]
GVI = -0.2728(TM1)-0.2174(TM2)-0.5508(TM3)+0.7221(TM4)+0.0733(TM5)-0.1648(TM7) [5]
WI = 0.1446(TM1)+0.1761(TM2)+0.3322(TM3)+0.3396(TM4)-0.6210(TM5)- 0.4186(TM7) [6]

ERS-2 Synthetic Aperture Radar data: ERS-2 SAR images covering the study area were acquired for the following four dates, which are May 21, June 25, September 3, and October 8, 1997. These dates were selected to match the TM image overflight dates as closely as possible. The ERS-2 SAR sensor collects data in the C-band (5.3 Ghz, 5.565 cm) region of the spectrum. These data were collected in the VV-polarization mode using a 230 angle of incidence relative to nadir. Since the average topographic slope in Douglas County is less than 5% (USDA, 1977), no backscattering corrections for variation in slope were made.

Speckle within the SAR images was minimized using the median of values within a 3 x 3 pixel window. Fifteen ground control points were collected throughout the study area and used to develop the geometric transformation equations that resulted in a RMS error of < 0.5 pixels (< 6.25 m). The radar brightness for each pixel was estimated using a nearest neighbor resampling approach and the pixels were scaled to 12.5 m x 12.5 m. The output pixel sizes for the radar and TM data were kept similar to their original sensor-collected resolutions to minimize changing the data values during the resampling process.

Remote sensing image classification

The resulting 18-band dataset for each year were classified using an Iterative Self-Organizing Data Analysis technique (ISODATA_ERDAS 1994) to generate spectral statistics. These statistics were then submitted to a maximum likelihood classifier that assigned each pixel in the image to one of 100 spectral classes. These 100 classes were regrouped into three classes, cropland, forest or grassland. The spectral classes whose pixel did not fit well into the groups (confusion classes) were reclassified into 20-25 classes and the more refined spectral classes were again assigned to cropland, grassland, or woodland. This iterative process, sometimes called "cluster busting" was repeated until an acceptable level of classification accuracy was obtained.

Back to top.

Project Investigators and Staff:

	Dr. Kevin P. Price, KARS and Department of Geography Project co-principal investigator Associate Director, KARS Program Office: (785) 864-7723 email: k-price@ku.edu
	Dr. James M. Stiles, Department of Electrical Engineering and Computer Science and Radar Systems and Remote Sensing Laboratory Project co-principal investigator Office: (785) 864-8803 email: jstiles@rsl.ku.edu
	Dr. Xulin Guo, Kansas Applied Remote Sensing (KARS) Program Staff Office: (785) 864-3515 email: xguo@ku.edu

Project Overview |

Stage II Activities |

Images |

Products |

Publications |

Home |